Brian Tang
commited on
Commit
·
49ebb9c
0
Parent(s):
Snapshot of current state 4a58ca57710c49f51896e4bc820e202fbf64904b
Browse files- .gitattributes +36 -0
- .gitignore +73 -0
- README.md +366 -0
- adapters/adapter_config.json +31 -0
- adapters/adapter_model.safetensors +3 -0
- added_tokens.json +24 -0
- chat_template.json +3 -0
- config.json +108 -0
- config_sentence_transformers.json +13 -0
- configuration_jina_embeddings_v4.py +23 -0
- custom_lora_module.py +193 -0
- custom_st.py +185 -0
- generation_config.json +6 -0
- merges.txt +0 -0
- model-00001-of-00002.safetensors +3 -0
- model-00002-of-00002.safetensors +3 -0
- model.safetensors.index.json +833 -0
- modeling_jina_embeddings_v4.py +609 -0
- modules.json +9 -0
- preprocessor_config.json +33 -0
- qwen2_5_vl.py +0 -0
- results.json +582 -0
- special_tokens_map.json +31 -0
- tokenizer.json +3 -0
- tokenizer_config.json +209 -0
- vocab.json +0 -0
.gitattributes
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
*.7z filter=lfs diff=lfs merge=lfs -text
|
| 2 |
+
*.arrow filter=lfs diff=lfs merge=lfs -text
|
| 3 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
| 4 |
+
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
| 5 |
+
*.ckpt filter=lfs diff=lfs merge=lfs -text
|
| 6 |
+
*.ftz filter=lfs diff=lfs merge=lfs -text
|
| 7 |
+
*.gz filter=lfs diff=lfs merge=lfs -text
|
| 8 |
+
*.h5 filter=lfs diff=lfs merge=lfs -text
|
| 9 |
+
*.joblib filter=lfs diff=lfs merge=lfs -text
|
| 10 |
+
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
| 11 |
+
*.mlmodel filter=lfs diff=lfs merge=lfs -text
|
| 12 |
+
*.model filter=lfs diff=lfs merge=lfs -text
|
| 13 |
+
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
| 14 |
+
*.npy filter=lfs diff=lfs merge=lfs -text
|
| 15 |
+
*.npz filter=lfs diff=lfs merge=lfs -text
|
| 16 |
+
*.onnx filter=lfs diff=lfs merge=lfs -text
|
| 17 |
+
*.ot filter=lfs diff=lfs merge=lfs -text
|
| 18 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
| 19 |
+
*.pb filter=lfs diff=lfs merge=lfs -text
|
| 20 |
+
*.pickle filter=lfs diff=lfs merge=lfs -text
|
| 21 |
+
*.pkl filter=lfs diff=lfs merge=lfs -text
|
| 22 |
+
*.pt filter=lfs diff=lfs merge=lfs -text
|
| 23 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
| 24 |
+
*.rar filter=lfs diff=lfs merge=lfs -text
|
| 25 |
+
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
| 26 |
+
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
| 27 |
+
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
| 28 |
+
*.tar filter=lfs diff=lfs merge=lfs -text
|
| 29 |
+
*.tflite filter=lfs diff=lfs merge=lfs -text
|
| 30 |
+
*.tgz filter=lfs diff=lfs merge=lfs -text
|
| 31 |
+
*.wasm filter=lfs diff=lfs merge=lfs -text
|
| 32 |
+
*.xz filter=lfs diff=lfs merge=lfs -text
|
| 33 |
+
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
+
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
+
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
*.json filter=lfs diff=lfs merge=lfs -text
|
.gitignore
ADDED
|
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Python
|
| 2 |
+
__pycache__/
|
| 3 |
+
*.py[cod]
|
| 4 |
+
*$py.class
|
| 5 |
+
*.so
|
| 6 |
+
.Python
|
| 7 |
+
build/
|
| 8 |
+
develop-eggs/
|
| 9 |
+
dist/
|
| 10 |
+
downloads/
|
| 11 |
+
eggs/
|
| 12 |
+
.eggs/
|
| 13 |
+
lib/
|
| 14 |
+
lib64/
|
| 15 |
+
parts/
|
| 16 |
+
sdist/
|
| 17 |
+
var/
|
| 18 |
+
wheels/
|
| 19 |
+
*.egg-info/
|
| 20 |
+
.installed.cfg
|
| 21 |
+
*.egg
|
| 22 |
+
|
| 23 |
+
# Virtual Environment
|
| 24 |
+
venv/
|
| 25 |
+
env/
|
| 26 |
+
ENV/
|
| 27 |
+
.env
|
| 28 |
+
.venv
|
| 29 |
+
env.bak/
|
| 30 |
+
venv.bak/
|
| 31 |
+
|
| 32 |
+
# IDE
|
| 33 |
+
.idea/
|
| 34 |
+
.vscode/
|
| 35 |
+
*.swp
|
| 36 |
+
*.swo
|
| 37 |
+
.project
|
| 38 |
+
.pydevproject
|
| 39 |
+
.settings/
|
| 40 |
+
|
| 41 |
+
# Jupyter Notebook
|
| 42 |
+
.ipynb_checkpoints
|
| 43 |
+
*.ipynb
|
| 44 |
+
|
| 45 |
+
# Distribution / packaging
|
| 46 |
+
.Python
|
| 47 |
+
*.manifest
|
| 48 |
+
*.spec
|
| 49 |
+
|
| 50 |
+
# Unit test / coverage reports
|
| 51 |
+
htmlcov/
|
| 52 |
+
.tox/
|
| 53 |
+
.coverage
|
| 54 |
+
.coverage.*
|
| 55 |
+
.cache
|
| 56 |
+
nosetests.xml
|
| 57 |
+
coverage.xml
|
| 58 |
+
*.cover
|
| 59 |
+
.hypothesis/
|
| 60 |
+
|
| 61 |
+
# Logs and databases
|
| 62 |
+
*.log
|
| 63 |
+
*.sqlite
|
| 64 |
+
*.db
|
| 65 |
+
|
| 66 |
+
# OS generated files
|
| 67 |
+
.DS_Store
|
| 68 |
+
.DS_Store?
|
| 69 |
+
._*
|
| 70 |
+
.Spotlight-V100
|
| 71 |
+
.Trashes
|
| 72 |
+
ehthumbs.db
|
| 73 |
+
Thumbs.db
|
README.md
ADDED
|
@@ -0,0 +1,366 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
+
tags:
|
| 4 |
+
- vidore
|
| 5 |
+
- colpali
|
| 6 |
+
- multimodal-embedding
|
| 7 |
+
- multilingual-embedding
|
| 8 |
+
- Text-to-Visual Document (T→VD) retrieval
|
| 9 |
+
- feature-extraction
|
| 10 |
+
- sentence-similarity
|
| 11 |
+
- mteb
|
| 12 |
+
- sentence-transformers
|
| 13 |
+
language:
|
| 14 |
+
- multilingual
|
| 15 |
+
inference: false
|
| 16 |
+
library_name: transformers
|
| 17 |
+
pipeline_tag: visual-document-retrieval
|
| 18 |
+
---
|
| 19 |
+
<br><br>
|
| 20 |
+
|
| 21 |
+
<p align="center">
|
| 22 |
+
<img src="https://huggingface.co/datasets/jinaai/documentation-images/resolve/main/logo.webp" alt="Jina AI: Your Search Foundation, Supercharged!" width="150px">
|
| 23 |
+
</p>
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
<p align="center">
|
| 27 |
+
<b>The embedding model trained by <a href="https://jina.ai/"><b>Jina AI</b></a>.</b>
|
| 28 |
+
</p>
|
| 29 |
+
|
| 30 |
+
# Jina Embeddings v4: Universal Embeddings for Multimodal Multilingual Retrieval
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
[GGUF](https://github.com/jina-ai/jina-embeddings-v4-gguf) | [Blog](https://jina.ai/news/jina-embeddings-v4-universal-embeddings-for-multimodal-multilingual-retrieval) | [Technical Report](https://arxiv.org/abs/2506.18902) | [API](https://jina.ai/embeddings)
|
| 34 |
+
|
| 35 |
+
|
| 36 |
+
## Intended Usage & Model Info
|
| 37 |
+
`jina-embeddings-v4` is a universal embedding model for multimodal and multilingual retrieval.
|
| 38 |
+
The model is specially designed for complex document retrieval, including visually rich documents with charts, tables, and illustrations.
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
Built on [Qwen/Qwen2.5-VL-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct), `jina-embeddings-v4` features:
|
| 42 |
+
|
| 43 |
+
- **Unified embeddings** for text, images, and visual documents, supporting both dense (single-vector) and late-interaction (multi-vector) retrieval.
|
| 44 |
+
- **Multilingual support** (30+ languages) and compatibility with a wide range of domains, including technical and visually complex documents.
|
| 45 |
+
- **Task-specific adapters** for retrieval, text matching, and code-related tasks, which can be selected at inference time.
|
| 46 |
+
- **Flexible embedding size**: dense embeddings are 2048 dimensions by default but can be truncated to as low as 128 with minimal performance loss.
|
| 47 |
+
|
| 48 |
+
|
| 49 |
+
Summary of features:
|
| 50 |
+
|
| 51 |
+
| Feature | Jina Embeddings V4 |
|
| 52 |
+
|------------|------------|
|
| 53 |
+
| Base Model | Qwen2.5-VL-3B-Instruct |
|
| 54 |
+
| Supported Tasks | `retrieval`, `text-matching`, `code` |
|
| 55 |
+
| Model DType | BFloat 16 |
|
| 56 |
+
| Max Sequence Length | 32768 |
|
| 57 |
+
| Single-Vector Dimension | 2048 |
|
| 58 |
+
| Multi-Vector Dimension | 128 |
|
| 59 |
+
| Matryoshka dimensions | 128, 256, 512, 1024, 2048 |
|
| 60 |
+
| Pooling Strategy | Mean pooling |
|
| 61 |
+
| Attention Mechanism | FlashAttention2 |
|
| 62 |
+
|
| 63 |
+
|
| 64 |
+
|
| 65 |
+
## Training & Evaluation
|
| 66 |
+
|
| 67 |
+
Please refer to our [technical report of jina-embeddings-v4](https://arxiv.org/abs/2506.18902) for training details and benchmarks.
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
## Usage
|
| 71 |
+
|
| 72 |
+
<details>
|
| 73 |
+
<summary>Requirements</a></summary>
|
| 74 |
+
|
| 75 |
+
The following Python packages are required:
|
| 76 |
+
|
| 77 |
+
- `transformers>=4.52.0`
|
| 78 |
+
- `torch>=2.6.0`
|
| 79 |
+
- `peft>=0.15.2`
|
| 80 |
+
- `torchvision`
|
| 81 |
+
- `pillow`
|
| 82 |
+
|
| 83 |
+
### Optional / Recommended
|
| 84 |
+
- **flash-attention**: Installing [flash-attention](https://github.com/Dao-AILab/flash-attention) is recommended for improved inference speed and efficiency, but not mandatory.
|
| 85 |
+
- **sentence-transformers**: If you want to use the model via the `sentence-transformers` interface, install this package as well.
|
| 86 |
+
|
| 87 |
+
</details>
|
| 88 |
+
|
| 89 |
+
|
| 90 |
+
<details>
|
| 91 |
+
<summary>via <a href="https://jina.ai/embeddings/">Jina AI Embeddings API</a></summary>
|
| 92 |
+
|
| 93 |
+
|
| 94 |
+
```bash
|
| 95 |
+
curl https://api.jina.ai/v1/embeddings \
|
| 96 |
+
-H "Content-Type: application/json" \
|
| 97 |
+
-H "Authorization: Bearer $JINA_AI_API_TOKEN" \
|
| 98 |
+
-d @- <<EOFEOF
|
| 99 |
+
{
|
| 100 |
+
"model": "jina-embeddings-v4",
|
| 101 |
+
"task": "text-matching",
|
| 102 |
+
"input": [
|
| 103 |
+
{
|
| 104 |
+
"text": "غروب جميل على الشاطئ"
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"text": "海滩上美丽的日落"
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"text": "A beautiful sunset over the beach"
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"text": "Un beau coucher de soleil sur la plage"
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"text": "Ein wunderschöner Sonnenuntergang am Strand"
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"text": "Ένα όμορφο ηλιοβασίλεμα πάνω από την παραλία"
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"text": "समुद्र तट पर एक खूबसूरत सूर्यास्त"
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"text": "Un bellissimo tramonto sulla spiaggia"
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"text": "浜辺に沈む美しい夕日"
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"text": "해변 위로 아름다운 일몰"
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"image": "https://i.ibb.co/nQNGqL0/beach1.jpg"
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"image": "https://i.ibb.co/r5w8hG8/beach2.jpg"
|
| 138 |
+
}
|
| 139 |
+
]
|
| 140 |
+
}
|
| 141 |
+
EOFEOF
|
| 142 |
+
```
|
| 143 |
+
|
| 144 |
+
</details>
|
| 145 |
+
|
| 146 |
+
<details>
|
| 147 |
+
<summary>via <a href="https://huggingface.co/docs/transformers/en/index">transformers</a></summary>
|
| 148 |
+
|
| 149 |
+
```python
|
| 150 |
+
# !pip install transformers>=4.52.0 torch>=2.6.0 peft>=0.15.2 torchvision pillow
|
| 151 |
+
# !pip install
|
| 152 |
+
from transformers import AutoModel
|
| 153 |
+
import torch
|
| 154 |
+
|
| 155 |
+
# Initialize the model
|
| 156 |
+
model = AutoModel.from_pretrained("jinaai/jina-embeddings-v4", trust_remote_code=True, torch_dtype=torch.float16)
|
| 157 |
+
|
| 158 |
+
model.to("cuda")
|
| 159 |
+
|
| 160 |
+
# ========================
|
| 161 |
+
# 1. Retrieval Task
|
| 162 |
+
# ========================
|
| 163 |
+
# Configure truncate_dim, max_length (for texts), max_pixels (for images), vector_type, batch_size in the encode function if needed
|
| 164 |
+
|
| 165 |
+
# Encode query
|
| 166 |
+
query_embeddings = model.encode_text(
|
| 167 |
+
texts=["Overview of climate change impacts on coastal cities"],
|
| 168 |
+
task="retrieval",
|
| 169 |
+
prompt_name="query",
|
| 170 |
+
)
|
| 171 |
+
|
| 172 |
+
# Encode passage (text)
|
| 173 |
+
passage_embeddings = model.encode_text(
|
| 174 |
+
texts=[
|
| 175 |
+
"Climate change has led to rising sea levels, increased frequency of extreme weather events..."
|
| 176 |
+
],
|
| 177 |
+
task="retrieval",
|
| 178 |
+
prompt_name="passage",
|
| 179 |
+
)
|
| 180 |
+
|
| 181 |
+
# Encode image/document
|
| 182 |
+
image_embeddings = model.encode_image(
|
| 183 |
+
images=["https://i.ibb.co/nQNGqL0/beach1.jpg"],
|
| 184 |
+
task="retrieval",
|
| 185 |
+
)
|
| 186 |
+
|
| 187 |
+
# ========================
|
| 188 |
+
# 2. Text Matching Task
|
| 189 |
+
# ========================
|
| 190 |
+
texts = [
|
| 191 |
+
"غروب جميل على الشاطئ", # Arabic
|
| 192 |
+
"海滩上美丽的日落", # Chinese
|
| 193 |
+
"Un beau coucher de soleil sur la plage", # French
|
| 194 |
+
"Ein wunderschöner Sonnenuntergang am Strand", # German
|
| 195 |
+
"Ένα όμορφο ηλιοβασίλεμα πάνω από την παραλία", # Greek
|
| 196 |
+
"समुद्र तट पर एक खूबसूरत सूर्यास्त", # Hindi
|
| 197 |
+
"Un bellissimo tramonto sulla spiaggia", # Italian
|
| 198 |
+
"浜辺に沈む美しい夕日", # Japanese
|
| 199 |
+
"해변 위로 아름다운 일몰", # Korean
|
| 200 |
+
]
|
| 201 |
+
|
| 202 |
+
text_embeddings = model.encode_text(texts=texts, task="text-matching")
|
| 203 |
+
|
| 204 |
+
# ========================
|
| 205 |
+
# 3. Code Understanding Task
|
| 206 |
+
# ========================
|
| 207 |
+
|
| 208 |
+
# Encode query
|
| 209 |
+
query_embedding = model.encode_text(
|
| 210 |
+
texts=["Find a function that prints a greeting message to the console"],
|
| 211 |
+
task="code",
|
| 212 |
+
prompt_name="query",
|
| 213 |
+
)
|
| 214 |
+
|
| 215 |
+
# Encode code
|
| 216 |
+
code_embeddings = model.encode_text(
|
| 217 |
+
texts=["def hello_world():\n print('Hello, World!')"],
|
| 218 |
+
task="code",
|
| 219 |
+
prompt_name="passage",
|
| 220 |
+
)
|
| 221 |
+
|
| 222 |
+
# ========================
|
| 223 |
+
# 4. Use multivectors
|
| 224 |
+
# ========================
|
| 225 |
+
|
| 226 |
+
multivector_embeddings = model.encode_text(
|
| 227 |
+
texts=texts,
|
| 228 |
+
task="retrieval",
|
| 229 |
+
prompt_name="query",
|
| 230 |
+
return_multivector=True,
|
| 231 |
+
)
|
| 232 |
+
|
| 233 |
+
images = ["https://i.ibb.co/nQNGqL0/beach1.jpg", "https://i.ibb.co/r5w8hG8/beach2.jpg"]
|
| 234 |
+
multivector_image_embeddings = model.encode_image(
|
| 235 |
+
images=images,
|
| 236 |
+
task="retrieval",
|
| 237 |
+
return_multivector=True,
|
| 238 |
+
)
|
| 239 |
+
```
|
| 240 |
+
</details>
|
| 241 |
+
|
| 242 |
+
<details>
|
| 243 |
+
<summary>via <a href="https://sbert.net/">sentence-transformers</a></summary>
|
| 244 |
+
|
| 245 |
+
```python
|
| 246 |
+
from sentence_transformers import SentenceTransformer
|
| 247 |
+
|
| 248 |
+
# Initialize the model
|
| 249 |
+
model = SentenceTransformer("jinaai/jina-embeddings-v4", trust_remote_code=True)
|
| 250 |
+
# ========================
|
| 251 |
+
# 1. Retrieval Task
|
| 252 |
+
# ========================
|
| 253 |
+
# Encode query
|
| 254 |
+
query_embeddings = model.encode(
|
| 255 |
+
sentences=["Overview of climate change impacts on coastal cities"],
|
| 256 |
+
task="retrieval",
|
| 257 |
+
prompt_name="query",
|
| 258 |
+
)
|
| 259 |
+
|
| 260 |
+
print(f"query_embeddings.shape = {query_embeddings.shape}")
|
| 261 |
+
|
| 262 |
+
# Encode passage (text)
|
| 263 |
+
passage_embeddings = model.encode(
|
| 264 |
+
sentences=[
|
| 265 |
+
"Climate change has led to rising sea levels, increased frequency of extreme weather events..."
|
| 266 |
+
],
|
| 267 |
+
task="retrieval",
|
| 268 |
+
prompt_name="passage",
|
| 269 |
+
)
|
| 270 |
+
|
| 271 |
+
print(f"passage_embeddings.shape = {passage_embeddings.shape}")
|
| 272 |
+
|
| 273 |
+
# Encode image/document
|
| 274 |
+
image_embeddings = model.encode(
|
| 275 |
+
sentences=["https://i.ibb.co/nQNGqL0/beach1.jpg"],
|
| 276 |
+
task="retrieval",
|
| 277 |
+
)
|
| 278 |
+
|
| 279 |
+
print(f"image_embeddings.shape = {image_embeddings.shape}")
|
| 280 |
+
|
| 281 |
+
# ========================
|
| 282 |
+
# 2. Text Matching Task
|
| 283 |
+
# ========================
|
| 284 |
+
texts = [
|
| 285 |
+
"غروب جميل على الشاطئ", # Arabic
|
| 286 |
+
"海滩上美丽的日落", # Chinese
|
| 287 |
+
"Un beau coucher de soleil sur la plage", # French
|
| 288 |
+
"Ein wunderschöner Sonnenuntergang am Strand", # German
|
| 289 |
+
"Ένα όμορφο ηλιοβασίλεμα πάνω από την παραλία", # Greek
|
| 290 |
+
"समुद्र तट पर एक खूबसूरत सूर्यास्त", # Hindi
|
| 291 |
+
"Un bellissimo tramonto sulla spiaggia", # Italian
|
| 292 |
+
"浜辺に沈む美しい夕日", # Japanese
|
| 293 |
+
"해변 위로 아름다운 일몰", # Korean
|
| 294 |
+
]
|
| 295 |
+
|
| 296 |
+
text_embeddings = model.encode(sentences=texts, task="text-matching")
|
| 297 |
+
|
| 298 |
+
# ========================
|
| 299 |
+
# 3. Code Understanding Task
|
| 300 |
+
# ========================
|
| 301 |
+
|
| 302 |
+
# Encode query
|
| 303 |
+
query_embeddings = model.encode(
|
| 304 |
+
sentences=["Find a function that prints a greeting message to the console"],
|
| 305 |
+
task="code",
|
| 306 |
+
prompt_name="query",
|
| 307 |
+
)
|
| 308 |
+
|
| 309 |
+
# Encode code
|
| 310 |
+
code_embeddings = model.encode(
|
| 311 |
+
sentences=["def hello_world():\n print('Hello, World!')"],
|
| 312 |
+
task="code",
|
| 313 |
+
prompt_name="passage",
|
| 314 |
+
)
|
| 315 |
+
|
| 316 |
+
# ========================
|
| 317 |
+
# 4. Use multivectors
|
| 318 |
+
# ========================
|
| 319 |
+
# If you want to use multi-vector embeddings, please use the Hugging Face model directly.
|
| 320 |
+
```
|
| 321 |
+
</details>
|
| 322 |
+
|
| 323 |
+
<details>
|
| 324 |
+
<summary>via <a href="https://github.com/vllm-project/vllm">vLLM</a></summary>
|
| 325 |
+
|
| 326 |
+
We provide separate model versions for each task (`retrieval`, `text-matching`, `code`) where specific adapter is merged into the base `Qwen2.5-VL` weights.
|
| 327 |
+
This modification enables native compatibility with vLLM.
|
| 328 |
+
|
| 329 |
+
Instructions and usage examples for each task are available in their respective directories:
|
| 330 |
+
- [jina-embeddings-v4-vllm-retrieval](https://huggingface.co/jinaai/jina-embeddings-v4-vllm-retrieval)
|
| 331 |
+
- [jina-embeddings-v4-vllm-text-matching](https://huggingface.co/jinaai/jina-embeddings-v4-vllm-text-matching)
|
| 332 |
+
- [jina-embeddings-v4-vllm-code](https://huggingface.co/jinaai/jina-embeddings-v4-vllm-code)
|
| 333 |
+
|
| 334 |
+
Please refer to the directory that matches your task for more details.
|
| 335 |
+
|
| 336 |
+
</details>
|
| 337 |
+
|
| 338 |
+
|
| 339 |
+
## Jina-VDR
|
| 340 |
+
Alongside `jina-embeddings-v4`, we’re releasing [Jina VDR](https://github.com/jina-ai/jina-vdr), a multilingual, multi-domain benchmark for visual document retrieval. The task collection can be viewed [here](https://huggingface.co/collections/jinaai/jinavdr-visual-document-retrieval-684831c022c53b21c313b449), and evaluation instructions can be found [here](https://github.com/jina-ai/jina-vdr).
|
| 341 |
+
|
| 342 |
+
|
| 343 |
+
## License
|
| 344 |
+
|
| 345 |
+
This model is licensed to download and run under [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/deed.en).
|
| 346 |
+
|
| 347 |
+
|
| 348 |
+
## Contact
|
| 349 |
+
|
| 350 |
+
Join our [Discord community](https://discord.jina.ai) and chat with other community members about ideas.
|
| 351 |
+
|
| 352 |
+
|
| 353 |
+
## Citation
|
| 354 |
+
|
| 355 |
+
If you find `jina-embeddings-v4` useful in your research, please cite the following paper:
|
| 356 |
+
```
|
| 357 |
+
@misc{günther2025jinaembeddingsv4universalembeddingsmultimodal,
|
| 358 |
+
title={jina-embeddings-v4: Universal Embeddings for Multimodal Multilingual Retrieval},
|
| 359 |
+
author={Michael Günther and Saba Sturua and Mohammad Kalim Akram and Isabelle Mohr and Andrei Ungureanu and Sedigheh Eslami and Scott Martens and Bo Wang and Nan Wang and Han Xiao},
|
| 360 |
+
year={2025},
|
| 361 |
+
eprint={2506.18902},
|
| 362 |
+
archivePrefix={arXiv},
|
| 363 |
+
primaryClass={cs.AI},
|
| 364 |
+
url={https://arxiv.org/abs/2506.18902},
|
| 365 |
+
}
|
| 366 |
+
```
|
adapters/adapter_config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "jinaai/jina-embeddings-v4",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"corda_config": null,
|
| 7 |
+
"eva_config": null,
|
| 8 |
+
"exclude_modules": ".*visual.*",
|
| 9 |
+
"fan_in_fan_out": false,
|
| 10 |
+
"inference_mode": true,
|
| 11 |
+
"init_lora_weights": "gaussian",
|
| 12 |
+
"layer_replication": null,
|
| 13 |
+
"layers_pattern": null,
|
| 14 |
+
"layers_to_transform": null,
|
| 15 |
+
"loftq_config": {},
|
| 16 |
+
"lora_alpha": 32,
|
| 17 |
+
"lora_bias": false,
|
| 18 |
+
"lora_dropout": 0.1,
|
| 19 |
+
"megatron_config": null,
|
| 20 |
+
"megatron_core": "megatron.core",
|
| 21 |
+
"modules_to_save": null,
|
| 22 |
+
"peft_type": "LORA",
|
| 23 |
+
"r": 32,
|
| 24 |
+
"rank_pattern": {},
|
| 25 |
+
"revision": null,
|
| 26 |
+
"target_modules": "(.*(model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(single_vector_projector|multi_vector_projector).*$)",
|
| 27 |
+
"task_type": "FEATURE_EXTRACTION",
|
| 28 |
+
"trainable_token_indices": null,
|
| 29 |
+
"use_dora": false,
|
| 30 |
+
"use_rslora": false
|
| 31 |
+
}
|
adapters/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b6b7ab4a79daa3b4f3b5274500cc99d3dc89aa8c3419e9d79f89e366685e12e5
|
| 3 |
+
size 359863776
|
added_tokens.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"</tool_call>": 151658,
|
| 3 |
+
"<tool_call>": 151657,
|
| 4 |
+
"<|box_end|>": 151649,
|
| 5 |
+
"<|box_start|>": 151648,
|
| 6 |
+
"<|endoftext|>": 151643,
|
| 7 |
+
"<|file_sep|>": 151664,
|
| 8 |
+
"<|fim_middle|>": 151660,
|
| 9 |
+
"<|fim_pad|>": 151662,
|
| 10 |
+
"<|fim_prefix|>": 151659,
|
| 11 |
+
"<|fim_suffix|>": 151661,
|
| 12 |
+
"<|im_end|>": 151645,
|
| 13 |
+
"<|im_start|>": 151644,
|
| 14 |
+
"<|image_pad|>": 151655,
|
| 15 |
+
"<|object_ref_end|>": 151647,
|
| 16 |
+
"<|object_ref_start|>": 151646,
|
| 17 |
+
"<|quad_end|>": 151651,
|
| 18 |
+
"<|quad_start|>": 151650,
|
| 19 |
+
"<|repo_name|>": 151663,
|
| 20 |
+
"<|video_pad|>": 151656,
|
| 21 |
+
"<|vision_end|>": 151653,
|
| 22 |
+
"<|vision_pad|>": 151654,
|
| 23 |
+
"<|vision_start|>": 151652
|
| 24 |
+
}
|
chat_template.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}"
|
| 3 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,108 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "jinaai/jina-embeddings-v4",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"JinaEmbeddingsV4Model"
|
| 5 |
+
],
|
| 6 |
+
"auto_map": {
|
| 7 |
+
"AutoConfig": "configuration_jina_embeddings_v4.JinaEmbeddingsV4Config",
|
| 8 |
+
"AutoModel": "modeling_jina_embeddings_v4.JinaEmbeddingsV4Model"
|
| 9 |
+
},
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"bos_token_id": 151643,
|
| 12 |
+
"eos_token_id": 151645,
|
| 13 |
+
"hidden_act": "silu",
|
| 14 |
+
"hidden_size": 2048,
|
| 15 |
+
"image_token_id": 151655,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 11008,
|
| 18 |
+
"max_position_embeddings": 128000,
|
| 19 |
+
"max_window_layers": 70,
|
| 20 |
+
"multi_vector_projector_dim": 128,
|
| 21 |
+
"num_attention_heads": 16,
|
| 22 |
+
"num_hidden_layers": 36,
|
| 23 |
+
"num_key_value_heads": 2,
|
| 24 |
+
"rms_norm_eps": 1e-06,
|
| 25 |
+
"rope_scaling": {
|
| 26 |
+
"mrope_section": [
|
| 27 |
+
16,
|
| 28 |
+
24,
|
| 29 |
+
24
|
| 30 |
+
],
|
| 31 |
+
"rope_type": "default",
|
| 32 |
+
"type": "default"
|
| 33 |
+
},
|
| 34 |
+
"rope_theta": 1000000.0,
|
| 35 |
+
"single_vector_pool_strategy": "mean",
|
| 36 |
+
"sliding_window": 32768,
|
| 37 |
+
"tie_word_embeddings": true,
|
| 38 |
+
"text_config": {
|
| 39 |
+
"attention_dropout": 0.0,
|
| 40 |
+
"bos_token_id": 151643,
|
| 41 |
+
"eos_token_id": 151645,
|
| 42 |
+
"hidden_act": "silu",
|
| 43 |
+
"hidden_size": 2048,
|
| 44 |
+
"image_token_id": null,
|
| 45 |
+
"initializer_range": 0.02,
|
| 46 |
+
"intermediate_size": 11008,
|
| 47 |
+
"max_position_embeddings": 128000,
|
| 48 |
+
"max_window_layers": 70,
|
| 49 |
+
"model_type": "qwen2_5_vl_text",
|
| 50 |
+
"num_attention_heads": 16,
|
| 51 |
+
"num_hidden_layers": 36,
|
| 52 |
+
"num_key_value_heads": 2,
|
| 53 |
+
"rms_norm_eps": 1e-06,
|
| 54 |
+
"rope_scaling": {
|
| 55 |
+
"mrope_section": [
|
| 56 |
+
16,
|
| 57 |
+
24,
|
| 58 |
+
24
|
| 59 |
+
],
|
| 60 |
+
"rope_type": "default",
|
| 61 |
+
"type": "default"
|
| 62 |
+
},
|
| 63 |
+
"rope_theta": 1000000.0,
|
| 64 |
+
"sliding_window": null,
|
| 65 |
+
"tie_word_embeddings": true,
|
| 66 |
+
"torch_dtype": "bfloat16",
|
| 67 |
+
"use_cache": true,
|
| 68 |
+
"use_sliding_window": false,
|
| 69 |
+
"vocab_size": 151936
|
| 70 |
+
},
|
| 71 |
+
"torch_dtype": "bfloat16",
|
| 72 |
+
"transformers_version": "4.52.0",
|
| 73 |
+
"use_cache": true,
|
| 74 |
+
"use_sliding_window": false,
|
| 75 |
+
"video_token_id": 151656,
|
| 76 |
+
"vision_config": {
|
| 77 |
+
"depth": 32,
|
| 78 |
+
"fullatt_block_indexes": [
|
| 79 |
+
7,
|
| 80 |
+
15,
|
| 81 |
+
23,
|
| 82 |
+
31
|
| 83 |
+
],
|
| 84 |
+
"hidden_act": "silu",
|
| 85 |
+
"hidden_size": 1280,
|
| 86 |
+
"in_channels": 3,
|
| 87 |
+
"in_chans": 3,
|
| 88 |
+
"initializer_range": 0.02,
|
| 89 |
+
"intermediate_size": 3420,
|
| 90 |
+
"model_type": "qwen2_5_vl",
|
| 91 |
+
"num_heads": 16,
|
| 92 |
+
"out_hidden_size": 2048,
|
| 93 |
+
"patch_size": 14,
|
| 94 |
+
"spatial_merge_size": 2,
|
| 95 |
+
"spatial_patch_size": 14,
|
| 96 |
+
"temporal_patch_size": 2,
|
| 97 |
+
"tokens_per_second": 2,
|
| 98 |
+
"torch_dtype": "bfloat16",
|
| 99 |
+
"window_size": 112
|
| 100 |
+
},
|
| 101 |
+
"task_names": ["retrieval", "text-matching", "code"],
|
| 102 |
+
"matryoshka_dims": [128, 256, 512, 1024, 2048],
|
| 103 |
+
"_attn_implementation": "flash_attention_2",
|
| 104 |
+
"truncate_dim": null,
|
| 105 |
+
"vision_end_token_id": 151653,
|
| 106 |
+
"vision_start_token_id": 151652,
|
| 107 |
+
"vision_token_id": 151654
|
| 108 |
+
}
|
config_sentence_transformers.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"__version__": {
|
| 3 |
+
"sentence_transformers": "4.1.0",
|
| 4 |
+
"transformers": "4.50.0",
|
| 5 |
+
"pytorch": "2.6.0"
|
| 6 |
+
},
|
| 7 |
+
"prompts":{
|
| 8 |
+
"query":"Query: ",
|
| 9 |
+
"passage":"Passage: "
|
| 10 |
+
},
|
| 11 |
+
"default_prompt_name": null,
|
| 12 |
+
"similarity_fn_name": "cosine"
|
| 13 |
+
}
|
configuration_jina_embeddings_v4.py
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from transformers.models.qwen2_5_vl import Qwen2_5_VLConfig
|
| 2 |
+
|
| 3 |
+
from typing import Optional
|
| 4 |
+
|
| 5 |
+
|
| 6 |
+
class JinaEmbeddingsV4Config(Qwen2_5_VLConfig):
|
| 7 |
+
"""
|
| 8 |
+
Configuration for the JinaEmbeddingsV4 model.
|
| 9 |
+
"""
|
| 10 |
+
|
| 11 |
+
def __init__(
|
| 12 |
+
self,
|
| 13 |
+
single_vector_pool_strategy: str = "mean",
|
| 14 |
+
multi_vector_projector_dim: int = 128,
|
| 15 |
+
pretrained_peft_model_name_or_path: Optional[str] = None,
|
| 16 |
+
verbosity: int = 1,
|
| 17 |
+
**kwargs,
|
| 18 |
+
):
|
| 19 |
+
super().__init__(**kwargs)
|
| 20 |
+
self.single_vector_pool_strategy = single_vector_pool_strategy
|
| 21 |
+
self.multi_vector_projector_dim = multi_vector_projector_dim
|
| 22 |
+
self.pretrained_peft_model_name_or_path = pretrained_peft_model_name_or_path
|
| 23 |
+
self.verbosity = verbosity
|
custom_lora_module.py
ADDED
|
@@ -0,0 +1,193 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import math
|
| 4 |
+
import warnings
|
| 5 |
+
from typing import Any, Optional, Union, List
|
| 6 |
+
|
| 7 |
+
import torch
|
| 8 |
+
import torch.nn as nn
|
| 9 |
+
|
| 10 |
+
from peft.tuners.lora import LoraLayer
|
| 11 |
+
|
| 12 |
+
class MultiAdapterLinear(nn.Module, LoraLayer):
|
| 13 |
+
"""
|
| 14 |
+
Custom LoRA module supporting multiple adapters for a linear layer.
|
| 15 |
+
|
| 16 |
+
This module extends the standard LoRA implementation to support multiple task-specific
|
| 17 |
+
adapters that can be dynamically selected during the forward pass. The task_label
|
| 18 |
+
parameter passed to the forward function determines which LoRA adapter(s) to use:
|
| 19 |
+
- If task_label is a string, all examples in the batch use the same adapter
|
| 20 |
+
- If task_label is a list of strings, each example can use a different adapter
|
| 21 |
+
|
| 22 |
+
This enables efficient multi-task inference where all task-specific LoRA adapters
|
| 23 |
+
are loaded in memory simultaneously and dynamically selected per example, eliminating
|
| 24 |
+
the need to switch adapter states between tasks and allowing optimal throughput
|
| 25 |
+
for mixed-task batches.
|
| 26 |
+
|
| 27 |
+
Derived from peft.tuners.lora.Linear.
|
| 28 |
+
"""
|
| 29 |
+
def __init__(
|
| 30 |
+
self,
|
| 31 |
+
base_layer,
|
| 32 |
+
adapter_name: str,
|
| 33 |
+
task_names: List[str],
|
| 34 |
+
r: int = 0,
|
| 35 |
+
lora_alpha: int = 1,
|
| 36 |
+
lora_dropout: float = 0.0,
|
| 37 |
+
fan_in_fan_out: bool = False, # Set this to True if the layer to replace stores weight like (fan_in, fan_out)
|
| 38 |
+
is_target_conv_1d_layer: bool = False,
|
| 39 |
+
init_lora_weights: Union[bool, str] = True,
|
| 40 |
+
use_rslora: bool = False,
|
| 41 |
+
use_dora: bool = False,
|
| 42 |
+
lora_bias: bool = False,
|
| 43 |
+
**kwargs,
|
| 44 |
+
) -> None:
|
| 45 |
+
super().__init__()
|
| 46 |
+
LoraLayer.__init__(self, base_layer, **kwargs)
|
| 47 |
+
|
| 48 |
+
self.fan_in_fan_out = fan_in_fan_out
|
| 49 |
+
self.task_names = task_names
|
| 50 |
+
self._active_adapter = adapter_name
|
| 51 |
+
self.update_layer(
|
| 52 |
+
adapter_name,
|
| 53 |
+
r,
|
| 54 |
+
lora_alpha=lora_alpha,
|
| 55 |
+
lora_dropout=lora_dropout,
|
| 56 |
+
init_lora_weights=init_lora_weights,
|
| 57 |
+
use_rslora=use_rslora,
|
| 58 |
+
use_dora=use_dora,
|
| 59 |
+
lora_bias=lora_bias,
|
| 60 |
+
)
|
| 61 |
+
self.is_target_conv_1d_layer = is_target_conv_1d_layer
|
| 62 |
+
|
| 63 |
+
|
| 64 |
+
def forward(self, x: torch.Tensor, task_label: Union[str, List[str]], *args: Any, **kwargs: Any) -> torch.Tensor:
|
| 65 |
+
self._check_forward_args(x, *args, **kwargs)
|
| 66 |
+
|
| 67 |
+
if self.disable_adapters:
|
| 68 |
+
if self.merged:
|
| 69 |
+
self.unmerge()
|
| 70 |
+
result = self.base_layer(x, *args, **kwargs)
|
| 71 |
+
elif self.merged:
|
| 72 |
+
result = self.base_layer(x, *args, **kwargs)
|
| 73 |
+
else:
|
| 74 |
+
result = self.base_layer(x, *args, **kwargs)
|
| 75 |
+
torch_result_dtype = result.dtype
|
| 76 |
+
|
| 77 |
+
lora_A_keys = self.lora_A.keys()
|
| 78 |
+
for active_adapter in self.active_adapters:
|
| 79 |
+
if active_adapter not in lora_A_keys:
|
| 80 |
+
continue
|
| 81 |
+
|
| 82 |
+
if isinstance(task_label, str):
|
| 83 |
+
lora_A = self.lora_A[active_adapter][task_label]
|
| 84 |
+
lora_B = self.lora_B[active_adapter][task_label]
|
| 85 |
+
dropout = self.lora_dropout[active_adapter]
|
| 86 |
+
scaling = self.scaling[active_adapter]
|
| 87 |
+
x = self._cast_input_dtype(x, lora_A.weight.dtype)
|
| 88 |
+
result = result + lora_B(lora_A(dropout(x))) * scaling
|
| 89 |
+
else:
|
| 90 |
+
unique_tasks = list(set(task_label))
|
| 91 |
+
lora_output = torch.zeros_like(result)
|
| 92 |
+
|
| 93 |
+
for task in unique_tasks:
|
| 94 |
+
task_indices = [i for i, t in enumerate(task_label) if t == task]
|
| 95 |
+
task_x = x[task_indices]
|
| 96 |
+
|
| 97 |
+
lora_A = self.lora_A[active_adapter][task]
|
| 98 |
+
lora_B = self.lora_B[active_adapter][task]
|
| 99 |
+
dropout = self.lora_dropout[active_adapter]
|
| 100 |
+
scaling = self.scaling[active_adapter]
|
| 101 |
+
|
| 102 |
+
task_x = self._cast_input_dtype(task_x, lora_A.weight.dtype)
|
| 103 |
+
task_lora_value = lora_B(lora_A(dropout(task_x))) * scaling
|
| 104 |
+
|
| 105 |
+
for i, idx in enumerate(task_indices):
|
| 106 |
+
lora_output[idx] = task_lora_value[i]
|
| 107 |
+
|
| 108 |
+
result = result + lora_output
|
| 109 |
+
|
| 110 |
+
result = result.to(torch_result_dtype)
|
| 111 |
+
|
| 112 |
+
return result
|
| 113 |
+
|
| 114 |
+
def __repr__(self) -> str:
|
| 115 |
+
rep = super().__repr__()
|
| 116 |
+
return "lora." + rep
|
| 117 |
+
|
| 118 |
+
|
| 119 |
+
def update_layer(
|
| 120 |
+
self,
|
| 121 |
+
adapter_name,
|
| 122 |
+
r,
|
| 123 |
+
lora_alpha,
|
| 124 |
+
lora_dropout,
|
| 125 |
+
init_lora_weights,
|
| 126 |
+
use_rslora,
|
| 127 |
+
use_dora: bool = False,
|
| 128 |
+
lora_bias: bool = False,
|
| 129 |
+
):
|
| 130 |
+
# This code works for linear layers, override for other layer types
|
| 131 |
+
if r <= 0:
|
| 132 |
+
raise ValueError(f"`r` should be a positive integer value but the value passed is {r}")
|
| 133 |
+
|
| 134 |
+
self.r[adapter_name] = r
|
| 135 |
+
self.lora_alpha[adapter_name] = lora_alpha
|
| 136 |
+
if lora_dropout > 0.0:
|
| 137 |
+
lora_dropout_layer = nn.Dropout(p=lora_dropout)
|
| 138 |
+
else:
|
| 139 |
+
lora_dropout_layer = nn.Identity()
|
| 140 |
+
|
| 141 |
+
self.lora_dropout.update(nn.ModuleDict({adapter_name: lora_dropout_layer}))
|
| 142 |
+
# Actual trainable parameters
|
| 143 |
+
self.lora_A[adapter_name] = nn.ModuleDict({
|
| 144 |
+
task_name: nn.Linear(self.in_features, r, bias=False)
|
| 145 |
+
for task_name in self.task_names
|
| 146 |
+
})
|
| 147 |
+
self.lora_B[adapter_name] = nn.ModuleDict({
|
| 148 |
+
task_name: nn.Linear(r, self.out_features, bias=lora_bias)
|
| 149 |
+
for task_name in self.task_names
|
| 150 |
+
})
|
| 151 |
+
self.lora_bias[adapter_name] = lora_bias
|
| 152 |
+
|
| 153 |
+
if use_rslora:
|
| 154 |
+
self.scaling[adapter_name] = lora_alpha / math.sqrt(r)
|
| 155 |
+
else:
|
| 156 |
+
self.scaling[adapter_name] = lora_alpha / r
|
| 157 |
+
|
| 158 |
+
self.reset_lora_parameters(adapter_name, init_lora_weights)
|
| 159 |
+
self._move_adapter_to_device_of_base_layer(adapter_name)
|
| 160 |
+
self.use_dora[adapter_name] = False
|
| 161 |
+
self.set_adapter(self.active_adapters)
|
| 162 |
+
|
| 163 |
+
def reset_lora_parameters(self, adapter_name, init_lora_weights):
|
| 164 |
+
if init_lora_weights is False:
|
| 165 |
+
return
|
| 166 |
+
if init_lora_weights is True:
|
| 167 |
+
# initialize A the same way as the default for nn.Linear and B to zero
|
| 168 |
+
# https://github.com/microsoft/LoRA/blob/a0a92e0f26c067cf94747bdbf1ce73793fa44d19/loralib/layers.py#L124
|
| 169 |
+
for task_name in self.task_names:
|
| 170 |
+
nn.init.kaiming_uniform_(self.lora_A[adapter_name][task_name].weight, a=math.sqrt(5))
|
| 171 |
+
elif init_lora_weights.lower() == "gaussian":
|
| 172 |
+
for task_name in self.task_names:
|
| 173 |
+
nn.init.normal_(self.lora_A[adapter_name][task_name].weight, std=1 / self.r[adapter_name])
|
| 174 |
+
else:
|
| 175 |
+
raise ValueError(f"Unknown initialization {init_lora_weights=}")
|
| 176 |
+
for task_name in self.task_names:
|
| 177 |
+
nn.init.zeros_(self.lora_B[adapter_name][task_name].weight)
|
| 178 |
+
if self.lora_bias[adapter_name]:
|
| 179 |
+
for task_name in self.task_names:
|
| 180 |
+
nn.init.zeros_(self.lora_B[adapter_name][task_name].bias)
|
| 181 |
+
|
| 182 |
+
|
| 183 |
+
def merge(self, safe_merge: bool = False, adapter_names: Optional[list[str]] = None) -> None:
|
| 184 |
+
"""
|
| 185 |
+
Merge the active adapter weights into the base weights
|
| 186 |
+
"""
|
| 187 |
+
raise NotImplementedError("Merge operation is not supported")
|
| 188 |
+
|
| 189 |
+
def unmerge(self) -> None:
|
| 190 |
+
"""
|
| 191 |
+
This method unmerges all merged adapter layers from the base weights.
|
| 192 |
+
"""
|
| 193 |
+
raise NotImplementedError("Unmerge operation is not supported")
|
custom_st.py
ADDED
|
@@ -0,0 +1,185 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import json
|
| 2 |
+
import os
|
| 3 |
+
from io import BytesIO
|
| 4 |
+
from pathlib import Path
|
| 5 |
+
from typing import Any, Dict, List, Literal, Optional, Union
|
| 6 |
+
|
| 7 |
+
import requests
|
| 8 |
+
import torch
|
| 9 |
+
from PIL import Image
|
| 10 |
+
from torch import nn
|
| 11 |
+
from transformers import AutoConfig, AutoModel, AutoProcessor
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
class Transformer(nn.Module):
|
| 15 |
+
|
| 16 |
+
save_in_root: bool = True
|
| 17 |
+
|
| 18 |
+
def __init__(
|
| 19 |
+
self,
|
| 20 |
+
model_name_or_path: str = "jinaai/jina-embeddings-v4",
|
| 21 |
+
max_seq_length: Optional[int] = None,
|
| 22 |
+
config_args: Optional[Dict[str, Any]] = None,
|
| 23 |
+
model_args: Optional[Dict[str, Any]] = None,
|
| 24 |
+
tokenizer_args: Optional[Dict[str, Any]] = None,
|
| 25 |
+
cache_dir: Optional[str] = None,
|
| 26 |
+
backend: Literal["torch", "onnx", "openvino"] = "torch",
|
| 27 |
+
**kwargs,
|
| 28 |
+
) -> None:
|
| 29 |
+
super(Transformer, self).__init__()
|
| 30 |
+
if backend != "torch":
|
| 31 |
+
raise ValueError(
|
| 32 |
+
f"Backend '{backend}' is not supported, please use 'torch' instead"
|
| 33 |
+
)
|
| 34 |
+
config_kwargs = config_args or {}
|
| 35 |
+
model_kwargs = model_args or {}
|
| 36 |
+
tokenizer_kwargs = tokenizer_args or {}
|
| 37 |
+
|
| 38 |
+
self.config = AutoConfig.from_pretrained(
|
| 39 |
+
model_name_or_path, cache_dir=cache_dir, **config_kwargs
|
| 40 |
+
)
|
| 41 |
+
self.default_task = model_args.pop("default_task", None)
|
| 42 |
+
if self.default_task and self.default_task not in self.config.task_names:
|
| 43 |
+
raise ValueError(
|
| 44 |
+
f"Invalid task: {self.default_task}. Must be one of {self.config.task_names}."
|
| 45 |
+
)
|
| 46 |
+
|
| 47 |
+
self.model = AutoModel.from_pretrained(
|
| 48 |
+
model_name_or_path, config=self.config, cache_dir=cache_dir, **model_kwargs
|
| 49 |
+
)
|
| 50 |
+
self.processor = AutoProcessor.from_pretrained(
|
| 51 |
+
model_name_or_path,
|
| 52 |
+
cache_dir=cache_dir,
|
| 53 |
+
use_fast=True,
|
| 54 |
+
**tokenizer_kwargs,
|
| 55 |
+
)
|
| 56 |
+
self.max_seq_length = max_seq_length or 8192
|
| 57 |
+
|
| 58 |
+
def tokenize(
|
| 59 |
+
self, texts: List[Union[str, Image.Image]], padding: Union[str, bool] = True
|
| 60 |
+
) -> Dict[str, torch.Tensor]:
|
| 61 |
+
encoding = {}
|
| 62 |
+
text_indices = []
|
| 63 |
+
image_indices = []
|
| 64 |
+
for i, text in enumerate(texts):
|
| 65 |
+
if isinstance(text, str):
|
| 66 |
+
# Remove Query: or Passage: prefixes when checking for URLs or file paths
|
| 67 |
+
clean_text = text
|
| 68 |
+
if text.startswith("Query: "):
|
| 69 |
+
clean_text = text[len("Query: ") :]
|
| 70 |
+
elif text.startswith("Passage: "):
|
| 71 |
+
clean_text = text[len("Passage: ") :]
|
| 72 |
+
|
| 73 |
+
if clean_text.startswith("http"):
|
| 74 |
+
response = requests.get(clean_text)
|
| 75 |
+
texts[i] = Image.open(BytesIO(response.content)).convert("RGB")
|
| 76 |
+
image_indices.append(i)
|
| 77 |
+
else:
|
| 78 |
+
try:
|
| 79 |
+
if Path(clean_text).is_file():
|
| 80 |
+
texts[i] = Image.open(clean_text).convert("RGB")
|
| 81 |
+
image_indices.append(i)
|
| 82 |
+
else:
|
| 83 |
+
text_indices.append(i)
|
| 84 |
+
except Exception as e:
|
| 85 |
+
text_indices.append(i)
|
| 86 |
+
elif isinstance(text, Image.Image):
|
| 87 |
+
image_indices.append(i)
|
| 88 |
+
else:
|
| 89 |
+
raise ValueError(f"Invalid input type: {type(text)}")
|
| 90 |
+
if text_indices:
|
| 91 |
+
_texts = [texts[i] for i in text_indices]
|
| 92 |
+
text_features = self.processor.process_texts(
|
| 93 |
+
_texts, max_length=self.max_seq_length
|
| 94 |
+
)
|
| 95 |
+
for key, value in text_features.items():
|
| 96 |
+
encoding[f"text_{key}"] = value
|
| 97 |
+
encoding["text_indices"] = text_indices
|
| 98 |
+
|
| 99 |
+
if image_indices:
|
| 100 |
+
_images = [texts[i] for i in image_indices]
|
| 101 |
+
img_features = self.processor.process_images(_images)
|
| 102 |
+
for key, value in img_features.items():
|
| 103 |
+
encoding[f"image_{key}"] = value
|
| 104 |
+
encoding["image_indices"] = image_indices
|
| 105 |
+
|
| 106 |
+
return encoding
|
| 107 |
+
|
| 108 |
+
def forward(
|
| 109 |
+
self,
|
| 110 |
+
features: Dict[str, torch.Tensor],
|
| 111 |
+
task: Optional[str] = None,
|
| 112 |
+
truncate_dim: Optional[int] = None,
|
| 113 |
+
) -> Dict[str, torch.Tensor]:
|
| 114 |
+
self.model.eval()
|
| 115 |
+
|
| 116 |
+
if task is None:
|
| 117 |
+
if self.default_task is None:
|
| 118 |
+
raise ValueError(
|
| 119 |
+
"Task must be specified before encoding data. You can set it either during "
|
| 120 |
+
"loading the model (e.g., model_kwargs={'default_task': 'retrieval'}) or "
|
| 121 |
+
"pass it as an argument to the encode method (e.g., model.encode(texts, task='retrieval'))."
|
| 122 |
+
)
|
| 123 |
+
task = self.default_task
|
| 124 |
+
else:
|
| 125 |
+
if task not in self.config.task_names:
|
| 126 |
+
raise ValueError(
|
| 127 |
+
f"Invalid task: {task}. Must be one of {self.config.task_names}."
|
| 128 |
+
)
|
| 129 |
+
|
| 130 |
+
device = self.model.device.type
|
| 131 |
+
all_embeddings = []
|
| 132 |
+
|
| 133 |
+
with torch.no_grad():
|
| 134 |
+
if any(k.startswith("text_") for k in features.keys()):
|
| 135 |
+
text_batch = {
|
| 136 |
+
k[len("text_") :]: v.to(device)
|
| 137 |
+
for k, v in features.items()
|
| 138 |
+
if k.startswith("text_") and k != "text_indices"
|
| 139 |
+
}
|
| 140 |
+
text_indices = features.get("text_indices", [])
|
| 141 |
+
with torch.autocast(device_type=device, dtype=torch.bfloat16):
|
| 142 |
+
text_embeddings = self.model(
|
| 143 |
+
**text_batch, task_label=task
|
| 144 |
+
).single_vec_emb
|
| 145 |
+
if truncate_dim:
|
| 146 |
+
text_embeddings = text_embeddings[:, :truncate_dim]
|
| 147 |
+
text_embeddings = torch.nn.functional.normalize(
|
| 148 |
+
text_embeddings, p=2, dim=-1
|
| 149 |
+
)
|
| 150 |
+
for i, embedding in enumerate(text_embeddings):
|
| 151 |
+
all_embeddings.append((text_indices[i], embedding))
|
| 152 |
+
|
| 153 |
+
if any(k.startswith("image_") for k in features.keys()):
|
| 154 |
+
image_batch = {
|
| 155 |
+
k[len("image_") :]: v.to(device)
|
| 156 |
+
for k, v in features.items()
|
| 157 |
+
if k.startswith("image_") and k != "image_indices"
|
| 158 |
+
}
|
| 159 |
+
image_indices = features.get("image_indices", [])
|
| 160 |
+
|
| 161 |
+
with torch.autocast(device_type=device, dtype=torch.bfloat16):
|
| 162 |
+
img_embeddings = self.model(
|
| 163 |
+
**image_batch, task_label=task
|
| 164 |
+
).single_vec_emb
|
| 165 |
+
if truncate_dim:
|
| 166 |
+
img_embeddings = img_embeddings[:, :truncate_dim]
|
| 167 |
+
img_embeddings = torch.nn.functional.normalize(
|
| 168 |
+
img_embeddings, p=2, dim=-1
|
| 169 |
+
)
|
| 170 |
+
|
| 171 |
+
for i, embedding in enumerate(img_embeddings):
|
| 172 |
+
all_embeddings.append((image_indices[i], embedding))
|
| 173 |
+
|
| 174 |
+
if not all_embeddings:
|
| 175 |
+
raise RuntimeError("No embeddings were generated")
|
| 176 |
+
|
| 177 |
+
all_embeddings.sort(key=lambda x: x[0]) # sort by original index
|
| 178 |
+
combined_embeddings = torch.stack([emb for _, emb in all_embeddings])
|
| 179 |
+
features["sentence_embedding"] = combined_embeddings
|
| 180 |
+
|
| 181 |
+
return features
|
| 182 |
+
|
| 183 |
+
@classmethod
|
| 184 |
+
def load(cls, input_path: str) -> "Transformer":
|
| 185 |
+
return cls(model_name_or_path=input_path)
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 151643,
|
| 4 |
+
"eos_token_id": 151645,
|
| 5 |
+
"transformers_version": "4.50.0.dev0"
|
| 6 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:abb244162956ec2f26d944b6c10cbb96afe211d2aff908b8b2f498ec27a9100b
|
| 3 |
+
size 4997750728
|
model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d5252a7ede6469220b0e7386af53fea9a45fa299a1d2af6fe68cb29897de3e3
|
| 3 |
+
size 2512111904
|
model.safetensors.index.json
ADDED
|
@@ -0,0 +1,833 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metadata": {
|
| 3 |
+
"total_size": 7513966848
|
| 4 |
+
},
|
| 5 |
+
"weight_map": {
|
| 6 |
+
"model.embed_tokens.weight": "model-00001-of-00002.safetensors",
|
| 7 |
+
"model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 8 |
+
"model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 9 |
+
"model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 10 |
+
"model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 11 |
+
"model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 12 |
+
"model.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 13 |
+
"model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 14 |
+
"model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 15 |
+
"model.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 16 |
+
"model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 17 |
+
"model.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 18 |
+
"model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 19 |
+
"model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 20 |
+
"model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 21 |
+
"model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 22 |
+
"model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 23 |
+
"model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 24 |
+
"model.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 25 |
+
"model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 26 |
+
"model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 27 |
+
"model.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 28 |
+
"model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 29 |
+
"model.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 30 |
+
"model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 31 |
+
"model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 32 |
+
"model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 33 |
+
"model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 34 |
+
"model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 35 |
+
"model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 36 |
+
"model.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 37 |
+
"model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 38 |
+
"model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 39 |
+
"model.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 40 |
+
"model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 41 |
+
"model.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 42 |
+
"model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 43 |
+
"model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 44 |
+
"model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 45 |
+
"model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 46 |
+
"model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 47 |
+
"model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 48 |
+
"model.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 49 |
+
"model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 50 |
+
"model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 51 |
+
"model.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 52 |
+
"model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 53 |
+
"model.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 54 |
+
"model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 55 |
+
"model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 56 |
+
"model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 57 |
+
"model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 58 |
+
"model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 59 |
+
"model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 60 |
+
"model.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 61 |
+
"model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 62 |
+
"model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 63 |
+
"model.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 64 |
+
"model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 65 |
+
"model.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 66 |
+
"model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 67 |
+
"model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 68 |
+
"model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 69 |
+
"model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 70 |
+
"model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 71 |
+
"model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 72 |
+
"model.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 73 |
+
"model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 74 |
+
"model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 75 |
+
"model.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 76 |
+
"model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 77 |
+
"model.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 78 |
+
"model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 79 |
+
"model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 80 |
+
"model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 81 |
+
"model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 82 |
+
"model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 83 |
+
"model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 84 |
+
"model.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 85 |
+
"model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 86 |
+
"model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 87 |
+
"model.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 88 |
+
"model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 89 |
+
"model.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 90 |
+
"model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 91 |
+
"model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 92 |
+
"model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 93 |
+
"model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 94 |
+
"model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 95 |
+
"model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 96 |
+
"model.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 97 |
+
"model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 98 |
+
"model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 99 |
+
"model.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 100 |
+
"model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 101 |
+
"model.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 102 |
+
"model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 103 |
+
"model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 104 |
+
"model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 105 |
+
"model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 106 |
+
"model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 107 |
+
"model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 108 |
+
"model.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 109 |
+
"model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 110 |
+
"model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 111 |
+
"model.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 112 |
+
"model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 113 |
+
"model.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 114 |
+
"model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 115 |
+
"model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 116 |
+
"model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 117 |
+
"model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 118 |
+
"model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 119 |
+
"model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 120 |
+
"model.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 121 |
+
"model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 122 |
+
"model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 123 |
+
"model.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 124 |
+
"model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 125 |
+
"model.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 126 |
+
"model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 127 |
+
"model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 128 |
+
"model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 129 |
+
"model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 130 |
+
"model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 131 |
+
"model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 132 |
+
"model.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 133 |
+
"model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 134 |
+
"model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 135 |
+
"model.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 136 |
+
"model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 137 |
+
"model.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 138 |
+
"model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 139 |
+
"model.layers.19.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 140 |
+
"model.layers.19.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 141 |
+
"model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 142 |
+
"model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 143 |
+
"model.layers.19.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 144 |
+
"model.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 145 |
+
"model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 146 |
+
"model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 147 |
+
"model.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 148 |
+
"model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 149 |
+
"model.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 150 |
+
"model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 151 |
+
"model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 152 |
+
"model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 153 |
+
"model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 154 |
+
"model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 155 |
+
"model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 156 |
+
"model.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 157 |
+
"model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 158 |
+
"model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 159 |
+
"model.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 160 |
+
"model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 161 |
+
"model.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 162 |
+
"model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 163 |
+
"model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 164 |
+
"model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 165 |
+
"model.layers.20.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 166 |
+
"model.layers.20.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 167 |
+
"model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 168 |
+
"model.layers.20.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 169 |
+
"model.layers.20.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 170 |
+
"model.layers.20.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 171 |
+
"model.layers.20.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 172 |
+
"model.layers.20.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 173 |
+
"model.layers.20.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 174 |
+
"model.layers.20.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 175 |
+
"model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 176 |
+
"model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 177 |
+
"model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 178 |
+
"model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 179 |
+
"model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 180 |
+
"model.layers.21.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 181 |
+
"model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 182 |
+
"model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 183 |
+
"model.layers.21.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 184 |
+
"model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 185 |
+
"model.layers.21.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 186 |
+
"model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 187 |
+
"model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 188 |
+
"model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 189 |
+
"model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 190 |
+
"model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 191 |
+
"model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 192 |
+
"model.layers.22.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 193 |
+
"model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 194 |
+
"model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 195 |
+
"model.layers.22.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 196 |
+
"model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 197 |
+
"model.layers.22.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 198 |
+
"model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 199 |
+
"model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 200 |
+
"model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 201 |
+
"model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 202 |
+
"model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 203 |
+
"model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 204 |
+
"model.layers.23.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 205 |
+
"model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 206 |
+
"model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 207 |
+
"model.layers.23.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 208 |
+
"model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 209 |
+
"model.layers.23.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 210 |
+
"model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 211 |
+
"model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 212 |
+
"model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 213 |
+
"model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 214 |
+
"model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 215 |
+
"model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 216 |
+
"model.layers.24.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 217 |
+
"model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 218 |
+
"model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 219 |
+
"model.layers.24.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 220 |
+
"model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 221 |
+
"model.layers.24.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 222 |
+
"model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 223 |
+
"model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 224 |
+
"model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 225 |
+
"model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 226 |
+
"model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 227 |
+
"model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 228 |
+
"model.layers.25.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 229 |
+
"model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 230 |
+
"model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 231 |
+
"model.layers.25.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 232 |
+
"model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 233 |
+
"model.layers.25.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 234 |
+
"model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 235 |
+
"model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 236 |
+
"model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 237 |
+
"model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 238 |
+
"model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 239 |
+
"model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 240 |
+
"model.layers.26.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 241 |
+
"model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 242 |
+
"model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 243 |
+
"model.layers.26.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 244 |
+
"model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 245 |
+
"model.layers.26.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 246 |
+
"model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 247 |
+
"model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 248 |
+
"model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 249 |
+
"model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 250 |
+
"model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 251 |
+
"model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 252 |
+
"model.layers.27.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 253 |
+
"model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 254 |
+
"model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 255 |
+
"model.layers.27.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 256 |
+
"model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 257 |
+
"model.layers.27.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 258 |
+
"model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 259 |
+
"model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 260 |
+
"model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 261 |
+
"model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 262 |
+
"model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 263 |
+
"model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 264 |
+
"model.layers.28.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 265 |
+
"model.layers.28.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 266 |
+
"model.layers.28.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 267 |
+
"model.layers.28.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 268 |
+
"model.layers.28.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 269 |
+
"model.layers.28.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 270 |
+
"model.layers.28.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 271 |
+
"model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 272 |
+
"model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 273 |
+
"model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 274 |
+
"model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 275 |
+
"model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 276 |
+
"model.layers.29.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 277 |
+
"model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 278 |
+
"model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 279 |
+
"model.layers.29.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 280 |
+
"model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 281 |
+
"model.layers.29.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 282 |
+
"model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 283 |
+
"model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 284 |
+
"model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 285 |
+
"model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 286 |
+
"model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 287 |
+
"model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 288 |
+
"model.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 289 |
+
"model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 290 |
+
"model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 291 |
+
"model.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 292 |
+
"model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 293 |
+
"model.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 294 |
+
"model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 295 |
+
"model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 296 |
+
"model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 297 |
+
"model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 298 |
+
"model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 299 |
+
"model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 300 |
+
"model.layers.30.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 301 |
+
"model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 302 |
+
"model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 303 |
+
"model.layers.30.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 304 |
+
"model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 305 |
+
"model.layers.30.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 306 |
+
"model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 307 |
+
"model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 308 |
+
"model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 309 |
+
"model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 310 |
+
"model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 311 |
+
"model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 312 |
+
"model.layers.31.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 313 |
+
"model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 314 |
+
"model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 315 |
+
"model.layers.31.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 316 |
+
"model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 317 |
+
"model.layers.31.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 318 |
+
"model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 319 |
+
"model.layers.32.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 320 |
+
"model.layers.32.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 321 |
+
"model.layers.32.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 322 |
+
"model.layers.32.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 323 |
+
"model.layers.32.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 324 |
+
"model.layers.32.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 325 |
+
"model.layers.32.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 326 |
+
"model.layers.32.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 327 |
+
"model.layers.32.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 328 |
+
"model.layers.32.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 329 |
+
"model.layers.32.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 330 |
+
"model.layers.32.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 331 |
+
"model.layers.33.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 332 |
+
"model.layers.33.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 333 |
+
"model.layers.33.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 334 |
+
"model.layers.33.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 335 |
+
"model.layers.33.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 336 |
+
"model.layers.33.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 337 |
+
"model.layers.33.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 338 |
+
"model.layers.33.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 339 |
+
"model.layers.33.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 340 |
+
"model.layers.33.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 341 |
+
"model.layers.33.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 342 |
+
"model.layers.33.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 343 |
+
"model.layers.34.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 344 |
+
"model.layers.34.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 345 |
+
"model.layers.34.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 346 |
+
"model.layers.34.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 347 |
+
"model.layers.34.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 348 |
+
"model.layers.34.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 349 |
+
"model.layers.34.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 350 |
+
"model.layers.34.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 351 |
+
"model.layers.34.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 352 |
+
"model.layers.34.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 353 |
+
"model.layers.34.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 354 |
+
"model.layers.34.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 355 |
+
"model.layers.35.input_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 356 |
+
"model.layers.35.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
|
| 357 |
+
"model.layers.35.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
|
| 358 |
+
"model.layers.35.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
|
| 359 |
+
"model.layers.35.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
|
| 360 |
+
"model.layers.35.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
|
| 361 |
+
"model.layers.35.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
|
| 362 |
+
"model.layers.35.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
|
| 363 |
+
"model.layers.35.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
|
| 364 |
+
"model.layers.35.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
|
| 365 |
+
"model.layers.35.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
|
| 366 |
+
"model.layers.35.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
|
| 367 |
+
"model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 368 |
+
"model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 369 |
+
"model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 370 |
+
"model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 371 |
+
"model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 372 |
+
"model.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 373 |
+
"model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 374 |
+
"model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 375 |
+
"model.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 376 |
+
"model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 377 |
+
"model.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 378 |
+
"model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 379 |
+
"model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 380 |
+
"model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 381 |
+
"model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 382 |
+
"model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 383 |
+
"model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 384 |
+
"model.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 385 |
+
"model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 386 |
+
"model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 387 |
+
"model.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 388 |
+
"model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 389 |
+
"model.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 390 |
+
"model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 391 |
+
"model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 392 |
+
"model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 393 |
+
"model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 394 |
+
"model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 395 |
+
"model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 396 |
+
"model.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 397 |
+
"model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 398 |
+
"model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 399 |
+
"model.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 400 |
+
"model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 401 |
+
"model.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 402 |
+
"model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 403 |
+
"model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 404 |
+
"model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 405 |
+
"model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 406 |
+
"model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 407 |
+
"model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 408 |
+
"model.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 409 |
+
"model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 410 |
+
"model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 411 |
+
"model.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 412 |
+
"model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 413 |
+
"model.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 414 |
+
"model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 415 |
+
"model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 416 |
+
"model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 417 |
+
"model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 418 |
+
"model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 419 |
+
"model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 420 |
+
"model.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 421 |
+
"model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 422 |
+
"model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 423 |
+
"model.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 424 |
+
"model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 425 |
+
"model.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 426 |
+
"model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 427 |
+
"model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 428 |
+
"model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 429 |
+
"model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 430 |
+
"model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 431 |
+
"model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
|
| 432 |
+
"model.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
|
| 433 |
+
"model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
|
| 434 |
+
"model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
|
| 435 |
+
"model.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
|
| 436 |
+
"model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
|
| 437 |
+
"model.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
|
| 438 |
+
"model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
|
| 439 |
+
"model.norm.weight": "model-00002-of-00002.safetensors",
|
| 440 |
+
"multi_vector_projector.bias": "model-00002-of-00002.safetensors",
|
| 441 |
+
"multi_vector_projector.weight": "model-00002-of-00002.safetensors",
|
| 442 |
+
"visual.blocks.0.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 443 |
+
"visual.blocks.0.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 444 |
+
"visual.blocks.0.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 445 |
+
"visual.blocks.0.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 446 |
+
"visual.blocks.0.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 447 |
+
"visual.blocks.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 448 |
+
"visual.blocks.0.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 449 |
+
"visual.blocks.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 450 |
+
"visual.blocks.0.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 451 |
+
"visual.blocks.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 452 |
+
"visual.blocks.0.norm1.weight": "model-00001-of-00002.safetensors",
|
| 453 |
+
"visual.blocks.0.norm2.weight": "model-00001-of-00002.safetensors",
|
| 454 |
+
"visual.blocks.1.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 455 |
+
"visual.blocks.1.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 456 |
+
"visual.blocks.1.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 457 |
+
"visual.blocks.1.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 458 |
+
"visual.blocks.1.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 459 |
+
"visual.blocks.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 460 |
+
"visual.blocks.1.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 461 |
+
"visual.blocks.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 462 |
+
"visual.blocks.1.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 463 |
+
"visual.blocks.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 464 |
+
"visual.blocks.1.norm1.weight": "model-00001-of-00002.safetensors",
|
| 465 |
+
"visual.blocks.1.norm2.weight": "model-00001-of-00002.safetensors",
|
| 466 |
+
"visual.blocks.10.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 467 |
+
"visual.blocks.10.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 468 |
+
"visual.blocks.10.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 469 |
+
"visual.blocks.10.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 470 |
+
"visual.blocks.10.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 471 |
+
"visual.blocks.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 472 |
+
"visual.blocks.10.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 473 |
+
"visual.blocks.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 474 |
+
"visual.blocks.10.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 475 |
+
"visual.blocks.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 476 |
+
"visual.blocks.10.norm1.weight": "model-00001-of-00002.safetensors",
|
| 477 |
+
"visual.blocks.10.norm2.weight": "model-00001-of-00002.safetensors",
|
| 478 |
+
"visual.blocks.11.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 479 |
+
"visual.blocks.11.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 480 |
+
"visual.blocks.11.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 481 |
+
"visual.blocks.11.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 482 |
+
"visual.blocks.11.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 483 |
+
"visual.blocks.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 484 |
+
"visual.blocks.11.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 485 |
+
"visual.blocks.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 486 |
+
"visual.blocks.11.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 487 |
+
"visual.blocks.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 488 |
+
"visual.blocks.11.norm1.weight": "model-00001-of-00002.safetensors",
|
| 489 |
+
"visual.blocks.11.norm2.weight": "model-00001-of-00002.safetensors",
|
| 490 |
+
"visual.blocks.12.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 491 |
+
"visual.blocks.12.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 492 |
+
"visual.blocks.12.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 493 |
+
"visual.blocks.12.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 494 |
+
"visual.blocks.12.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 495 |
+
"visual.blocks.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 496 |
+
"visual.blocks.12.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 497 |
+
"visual.blocks.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 498 |
+
"visual.blocks.12.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 499 |
+
"visual.blocks.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 500 |
+
"visual.blocks.12.norm1.weight": "model-00001-of-00002.safetensors",
|
| 501 |
+
"visual.blocks.12.norm2.weight": "model-00001-of-00002.safetensors",
|
| 502 |
+
"visual.blocks.13.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 503 |
+
"visual.blocks.13.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 504 |
+
"visual.blocks.13.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 505 |
+
"visual.blocks.13.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 506 |
+
"visual.blocks.13.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 507 |
+
"visual.blocks.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 508 |
+
"visual.blocks.13.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 509 |
+
"visual.blocks.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 510 |
+
"visual.blocks.13.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 511 |
+
"visual.blocks.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 512 |
+
"visual.blocks.13.norm1.weight": "model-00001-of-00002.safetensors",
|
| 513 |
+
"visual.blocks.13.norm2.weight": "model-00001-of-00002.safetensors",
|
| 514 |
+
"visual.blocks.14.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 515 |
+
"visual.blocks.14.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 516 |
+
"visual.blocks.14.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 517 |
+
"visual.blocks.14.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 518 |
+
"visual.blocks.14.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 519 |
+
"visual.blocks.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 520 |
+
"visual.blocks.14.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 521 |
+
"visual.blocks.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 522 |
+
"visual.blocks.14.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 523 |
+
"visual.blocks.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 524 |
+
"visual.blocks.14.norm1.weight": "model-00001-of-00002.safetensors",
|
| 525 |
+
"visual.blocks.14.norm2.weight": "model-00001-of-00002.safetensors",
|
| 526 |
+
"visual.blocks.15.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 527 |
+
"visual.blocks.15.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 528 |
+
"visual.blocks.15.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 529 |
+
"visual.blocks.15.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 530 |
+
"visual.blocks.15.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 531 |
+
"visual.blocks.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 532 |
+
"visual.blocks.15.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 533 |
+
"visual.blocks.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 534 |
+
"visual.blocks.15.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 535 |
+
"visual.blocks.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 536 |
+
"visual.blocks.15.norm1.weight": "model-00001-of-00002.safetensors",
|
| 537 |
+
"visual.blocks.15.norm2.weight": "model-00001-of-00002.safetensors",
|
| 538 |
+
"visual.blocks.16.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 539 |
+
"visual.blocks.16.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 540 |
+
"visual.blocks.16.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 541 |
+
"visual.blocks.16.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 542 |
+
"visual.blocks.16.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 543 |
+
"visual.blocks.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 544 |
+
"visual.blocks.16.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 545 |
+
"visual.blocks.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 546 |
+
"visual.blocks.16.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 547 |
+
"visual.blocks.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 548 |
+
"visual.blocks.16.norm1.weight": "model-00001-of-00002.safetensors",
|
| 549 |
+
"visual.blocks.16.norm2.weight": "model-00001-of-00002.safetensors",
|
| 550 |
+
"visual.blocks.17.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 551 |
+
"visual.blocks.17.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 552 |
+
"visual.blocks.17.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 553 |
+
"visual.blocks.17.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 554 |
+
"visual.blocks.17.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 555 |
+
"visual.blocks.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 556 |
+
"visual.blocks.17.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 557 |
+
"visual.blocks.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 558 |
+
"visual.blocks.17.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 559 |
+
"visual.blocks.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 560 |
+
"visual.blocks.17.norm1.weight": "model-00001-of-00002.safetensors",
|
| 561 |
+
"visual.blocks.17.norm2.weight": "model-00001-of-00002.safetensors",
|
| 562 |
+
"visual.blocks.18.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 563 |
+
"visual.blocks.18.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 564 |
+
"visual.blocks.18.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 565 |
+
"visual.blocks.18.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 566 |
+
"visual.blocks.18.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 567 |
+
"visual.blocks.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 568 |
+
"visual.blocks.18.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 569 |
+
"visual.blocks.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 570 |
+
"visual.blocks.18.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 571 |
+
"visual.blocks.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 572 |
+
"visual.blocks.18.norm1.weight": "model-00001-of-00002.safetensors",
|
| 573 |
+
"visual.blocks.18.norm2.weight": "model-00001-of-00002.safetensors",
|
| 574 |
+
"visual.blocks.19.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 575 |
+
"visual.blocks.19.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 576 |
+
"visual.blocks.19.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 577 |
+
"visual.blocks.19.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 578 |
+
"visual.blocks.19.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 579 |
+
"visual.blocks.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 580 |
+
"visual.blocks.19.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 581 |
+
"visual.blocks.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 582 |
+
"visual.blocks.19.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 583 |
+
"visual.blocks.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 584 |
+
"visual.blocks.19.norm1.weight": "model-00001-of-00002.safetensors",
|
| 585 |
+
"visual.blocks.19.norm2.weight": "model-00001-of-00002.safetensors",
|
| 586 |
+
"visual.blocks.2.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 587 |
+
"visual.blocks.2.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 588 |
+
"visual.blocks.2.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 589 |
+
"visual.blocks.2.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 590 |
+
"visual.blocks.2.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 591 |
+
"visual.blocks.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 592 |
+
"visual.blocks.2.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 593 |
+
"visual.blocks.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 594 |
+
"visual.blocks.2.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 595 |
+
"visual.blocks.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 596 |
+
"visual.blocks.2.norm1.weight": "model-00001-of-00002.safetensors",
|
| 597 |
+
"visual.blocks.2.norm2.weight": "model-00001-of-00002.safetensors",
|
| 598 |
+
"visual.blocks.20.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 599 |
+
"visual.blocks.20.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 600 |
+
"visual.blocks.20.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 601 |
+
"visual.blocks.20.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 602 |
+
"visual.blocks.20.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 603 |
+
"visual.blocks.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 604 |
+
"visual.blocks.20.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 605 |
+
"visual.blocks.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 606 |
+
"visual.blocks.20.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 607 |
+
"visual.blocks.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 608 |
+
"visual.blocks.20.norm1.weight": "model-00001-of-00002.safetensors",
|
| 609 |
+
"visual.blocks.20.norm2.weight": "model-00001-of-00002.safetensors",
|
| 610 |
+
"visual.blocks.21.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 611 |
+
"visual.blocks.21.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 612 |
+
"visual.blocks.21.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 613 |
+
"visual.blocks.21.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 614 |
+
"visual.blocks.21.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 615 |
+
"visual.blocks.21.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 616 |
+
"visual.blocks.21.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 617 |
+
"visual.blocks.21.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 618 |
+
"visual.blocks.21.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 619 |
+
"visual.blocks.21.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 620 |
+
"visual.blocks.21.norm1.weight": "model-00001-of-00002.safetensors",
|
| 621 |
+
"visual.blocks.21.norm2.weight": "model-00001-of-00002.safetensors",
|
| 622 |
+
"visual.blocks.22.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 623 |
+
"visual.blocks.22.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 624 |
+
"visual.blocks.22.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 625 |
+
"visual.blocks.22.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 626 |
+
"visual.blocks.22.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 627 |
+
"visual.blocks.22.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 628 |
+
"visual.blocks.22.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 629 |
+
"visual.blocks.22.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 630 |
+
"visual.blocks.22.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 631 |
+
"visual.blocks.22.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 632 |
+
"visual.blocks.22.norm1.weight": "model-00001-of-00002.safetensors",
|
| 633 |
+
"visual.blocks.22.norm2.weight": "model-00001-of-00002.safetensors",
|
| 634 |
+
"visual.blocks.23.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 635 |
+
"visual.blocks.23.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 636 |
+
"visual.blocks.23.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 637 |
+
"visual.blocks.23.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 638 |
+
"visual.blocks.23.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 639 |
+
"visual.blocks.23.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 640 |
+
"visual.blocks.23.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 641 |
+
"visual.blocks.23.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 642 |
+
"visual.blocks.23.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 643 |
+
"visual.blocks.23.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 644 |
+
"visual.blocks.23.norm1.weight": "model-00001-of-00002.safetensors",
|
| 645 |
+
"visual.blocks.23.norm2.weight": "model-00001-of-00002.safetensors",
|
| 646 |
+
"visual.blocks.24.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 647 |
+
"visual.blocks.24.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 648 |
+
"visual.blocks.24.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 649 |
+
"visual.blocks.24.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 650 |
+
"visual.blocks.24.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 651 |
+
"visual.blocks.24.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 652 |
+
"visual.blocks.24.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 653 |
+
"visual.blocks.24.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 654 |
+
"visual.blocks.24.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 655 |
+
"visual.blocks.24.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 656 |
+
"visual.blocks.24.norm1.weight": "model-00001-of-00002.safetensors",
|
| 657 |
+
"visual.blocks.24.norm2.weight": "model-00001-of-00002.safetensors",
|
| 658 |
+
"visual.blocks.25.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 659 |
+
"visual.blocks.25.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 660 |
+
"visual.blocks.25.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 661 |
+
"visual.blocks.25.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 662 |
+
"visual.blocks.25.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 663 |
+
"visual.blocks.25.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 664 |
+
"visual.blocks.25.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 665 |
+
"visual.blocks.25.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 666 |
+
"visual.blocks.25.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 667 |
+
"visual.blocks.25.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 668 |
+
"visual.blocks.25.norm1.weight": "model-00001-of-00002.safetensors",
|
| 669 |
+
"visual.blocks.25.norm2.weight": "model-00001-of-00002.safetensors",
|
| 670 |
+
"visual.blocks.26.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 671 |
+
"visual.blocks.26.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 672 |
+
"visual.blocks.26.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 673 |
+
"visual.blocks.26.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 674 |
+
"visual.blocks.26.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 675 |
+
"visual.blocks.26.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 676 |
+
"visual.blocks.26.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 677 |
+
"visual.blocks.26.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 678 |
+
"visual.blocks.26.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 679 |
+
"visual.blocks.26.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 680 |
+
"visual.blocks.26.norm1.weight": "model-00001-of-00002.safetensors",
|
| 681 |
+
"visual.blocks.26.norm2.weight": "model-00001-of-00002.safetensors",
|
| 682 |
+
"visual.blocks.27.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 683 |
+
"visual.blocks.27.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 684 |
+
"visual.blocks.27.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 685 |
+
"visual.blocks.27.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 686 |
+
"visual.blocks.27.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 687 |
+
"visual.blocks.27.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 688 |
+
"visual.blocks.27.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 689 |
+
"visual.blocks.27.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 690 |
+
"visual.blocks.27.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 691 |
+
"visual.blocks.27.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 692 |
+
"visual.blocks.27.norm1.weight": "model-00001-of-00002.safetensors",
|
| 693 |
+
"visual.blocks.27.norm2.weight": "model-00001-of-00002.safetensors",
|
| 694 |
+
"visual.blocks.28.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 695 |
+
"visual.blocks.28.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 696 |
+
"visual.blocks.28.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 697 |
+
"visual.blocks.28.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 698 |
+
"visual.blocks.28.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 699 |
+
"visual.blocks.28.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 700 |
+
"visual.blocks.28.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 701 |
+
"visual.blocks.28.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 702 |
+
"visual.blocks.28.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 703 |
+
"visual.blocks.28.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 704 |
+
"visual.blocks.28.norm1.weight": "model-00001-of-00002.safetensors",
|
| 705 |
+
"visual.blocks.28.norm2.weight": "model-00001-of-00002.safetensors",
|
| 706 |
+
"visual.blocks.29.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 707 |
+
"visual.blocks.29.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 708 |
+
"visual.blocks.29.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 709 |
+
"visual.blocks.29.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 710 |
+
"visual.blocks.29.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 711 |
+
"visual.blocks.29.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 712 |
+
"visual.blocks.29.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 713 |
+
"visual.blocks.29.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 714 |
+
"visual.blocks.29.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 715 |
+
"visual.blocks.29.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 716 |
+
"visual.blocks.29.norm1.weight": "model-00001-of-00002.safetensors",
|
| 717 |
+
"visual.blocks.29.norm2.weight": "model-00001-of-00002.safetensors",
|
| 718 |
+
"visual.blocks.3.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 719 |
+
"visual.blocks.3.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 720 |
+
"visual.blocks.3.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 721 |
+
"visual.blocks.3.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 722 |
+
"visual.blocks.3.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 723 |
+
"visual.blocks.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 724 |
+
"visual.blocks.3.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 725 |
+
"visual.blocks.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 726 |
+
"visual.blocks.3.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 727 |
+
"visual.blocks.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 728 |
+
"visual.blocks.3.norm1.weight": "model-00001-of-00002.safetensors",
|
| 729 |
+
"visual.blocks.3.norm2.weight": "model-00001-of-00002.safetensors",
|
| 730 |
+
"visual.blocks.30.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 731 |
+
"visual.blocks.30.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 732 |
+
"visual.blocks.30.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 733 |
+
"visual.blocks.30.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 734 |
+
"visual.blocks.30.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 735 |
+
"visual.blocks.30.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 736 |
+
"visual.blocks.30.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 737 |
+
"visual.blocks.30.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 738 |
+
"visual.blocks.30.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 739 |
+
"visual.blocks.30.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 740 |
+
"visual.blocks.30.norm1.weight": "model-00001-of-00002.safetensors",
|
| 741 |
+
"visual.blocks.30.norm2.weight": "model-00001-of-00002.safetensors",
|
| 742 |
+
"visual.blocks.31.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 743 |
+
"visual.blocks.31.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 744 |
+
"visual.blocks.31.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 745 |
+
"visual.blocks.31.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 746 |
+
"visual.blocks.31.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 747 |
+
"visual.blocks.31.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 748 |
+
"visual.blocks.31.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 749 |
+
"visual.blocks.31.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 750 |
+
"visual.blocks.31.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 751 |
+
"visual.blocks.31.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 752 |
+
"visual.blocks.31.norm1.weight": "model-00001-of-00002.safetensors",
|
| 753 |
+
"visual.blocks.31.norm2.weight": "model-00001-of-00002.safetensors",
|
| 754 |
+
"visual.blocks.4.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 755 |
+
"visual.blocks.4.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 756 |
+
"visual.blocks.4.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 757 |
+
"visual.blocks.4.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 758 |
+
"visual.blocks.4.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 759 |
+
"visual.blocks.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 760 |
+
"visual.blocks.4.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 761 |
+
"visual.blocks.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 762 |
+
"visual.blocks.4.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 763 |
+
"visual.blocks.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 764 |
+
"visual.blocks.4.norm1.weight": "model-00001-of-00002.safetensors",
|
| 765 |
+
"visual.blocks.4.norm2.weight": "model-00001-of-00002.safetensors",
|
| 766 |
+
"visual.blocks.5.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 767 |
+
"visual.blocks.5.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 768 |
+
"visual.blocks.5.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 769 |
+
"visual.blocks.5.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 770 |
+
"visual.blocks.5.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 771 |
+
"visual.blocks.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 772 |
+
"visual.blocks.5.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 773 |
+
"visual.blocks.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 774 |
+
"visual.blocks.5.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 775 |
+
"visual.blocks.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 776 |
+
"visual.blocks.5.norm1.weight": "model-00001-of-00002.safetensors",
|
| 777 |
+
"visual.blocks.5.norm2.weight": "model-00001-of-00002.safetensors",
|
| 778 |
+
"visual.blocks.6.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 779 |
+
"visual.blocks.6.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 780 |
+
"visual.blocks.6.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 781 |
+
"visual.blocks.6.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 782 |
+
"visual.blocks.6.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 783 |
+
"visual.blocks.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 784 |
+
"visual.blocks.6.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 785 |
+
"visual.blocks.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 786 |
+
"visual.blocks.6.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 787 |
+
"visual.blocks.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 788 |
+
"visual.blocks.6.norm1.weight": "model-00001-of-00002.safetensors",
|
| 789 |
+
"visual.blocks.6.norm2.weight": "model-00001-of-00002.safetensors",
|
| 790 |
+
"visual.blocks.7.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 791 |
+
"visual.blocks.7.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 792 |
+
"visual.blocks.7.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 793 |
+
"visual.blocks.7.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 794 |
+
"visual.blocks.7.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 795 |
+
"visual.blocks.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 796 |
+
"visual.blocks.7.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 797 |
+
"visual.blocks.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 798 |
+
"visual.blocks.7.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 799 |
+
"visual.blocks.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 800 |
+
"visual.blocks.7.norm1.weight": "model-00001-of-00002.safetensors",
|
| 801 |
+
"visual.blocks.7.norm2.weight": "model-00001-of-00002.safetensors",
|
| 802 |
+
"visual.blocks.8.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 803 |
+
"visual.blocks.8.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 804 |
+
"visual.blocks.8.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 805 |
+
"visual.blocks.8.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 806 |
+
"visual.blocks.8.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 807 |
+
"visual.blocks.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 808 |
+
"visual.blocks.8.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 809 |
+
"visual.blocks.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 810 |
+
"visual.blocks.8.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 811 |
+
"visual.blocks.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 812 |
+
"visual.blocks.8.norm1.weight": "model-00001-of-00002.safetensors",
|
| 813 |
+
"visual.blocks.8.norm2.weight": "model-00001-of-00002.safetensors",
|
| 814 |
+
"visual.blocks.9.attn.proj.bias": "model-00001-of-00002.safetensors",
|
| 815 |
+
"visual.blocks.9.attn.proj.weight": "model-00001-of-00002.safetensors",
|
| 816 |
+
"visual.blocks.9.attn.qkv.bias": "model-00001-of-00002.safetensors",
|
| 817 |
+
"visual.blocks.9.attn.qkv.weight": "model-00001-of-00002.safetensors",
|
| 818 |
+
"visual.blocks.9.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
|
| 819 |
+
"visual.blocks.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
| 820 |
+
"visual.blocks.9.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
|
| 821 |
+
"visual.blocks.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
|
| 822 |
+
"visual.blocks.9.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
|
| 823 |
+
"visual.blocks.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
|
| 824 |
+
"visual.blocks.9.norm1.weight": "model-00001-of-00002.safetensors",
|
| 825 |
+
"visual.blocks.9.norm2.weight": "model-00001-of-00002.safetensors",
|
| 826 |
+
"visual.merger.ln_q.weight": "model-00001-of-00002.safetensors",
|
| 827 |
+
"visual.merger.mlp.0.bias": "model-00001-of-00002.safetensors",
|
| 828 |
+
"visual.merger.mlp.0.weight": "model-00001-of-00002.safetensors",
|
| 829 |
+
"visual.merger.mlp.2.bias": "model-00001-of-00002.safetensors",
|
| 830 |
+
"visual.merger.mlp.2.weight": "model-00001-of-00002.safetensors",
|
| 831 |
+
"visual.patch_embed.proj.weight": "model-00001-of-00002.safetensors"
|
| 832 |
+
}
|
| 833 |
+
}
|
modeling_jina_embeddings_v4.py
ADDED
|
@@ -0,0 +1,609 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Jina Embeddings V4 Model implementation was inspired by the ColPali codebase:
|
| 2 |
+
# https://github.com/illuin-tech/colpali
|
| 3 |
+
|
| 4 |
+
import os
|
| 5 |
+
from dataclasses import dataclass
|
| 6 |
+
from enum import Enum
|
| 7 |
+
from functools import partial
|
| 8 |
+
from io import BytesIO
|
| 9 |
+
from typing import Any, Callable, ClassVar, Dict, List, Optional, Union, cast
|
| 10 |
+
|
| 11 |
+
import numpy as np
|
| 12 |
+
import requests
|
| 13 |
+
import torch
|
| 14 |
+
from huggingface_hub import snapshot_download
|
| 15 |
+
from peft import LoraConfig, PeftModel
|
| 16 |
+
from PIL import Image
|
| 17 |
+
from torch import nn
|
| 18 |
+
from torch.utils.data import DataLoader
|
| 19 |
+
from tqdm import tqdm
|
| 20 |
+
from transformers import BatchFeature
|
| 21 |
+
from transformers.utils import is_flash_attn_2_available
|
| 22 |
+
|
| 23 |
+
from .configuration_jina_embeddings_v4 import JinaEmbeddingsV4Config
|
| 24 |
+
from .custom_lora_module import MultiAdapterLinear
|
| 25 |
+
from .qwen2_5_vl import Qwen2_5_VLForConditionalGeneration, Qwen2_5_VLProcessor
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
class PromptType(str, Enum):
|
| 29 |
+
query = "query"
|
| 30 |
+
passage = "passage"
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
PREFIX_DICT = {"query": "Query", "passage": "Passage"}
|
| 34 |
+
|
| 35 |
+
|
| 36 |
+
class JinaEmbeddingsV4Processor(Qwen2_5_VLProcessor):
|
| 37 |
+
def __init__(self, *args, **kwargs) -> None:
|
| 38 |
+
Qwen2_5_VLProcessor.__init__(self, *args, **kwargs)
|
| 39 |
+
self.assistant_prefix_len = 58
|
| 40 |
+
self.text_max_length = 32768
|
| 41 |
+
|
| 42 |
+
def process_images(
|
| 43 |
+
self,
|
| 44 |
+
images: Union[List[Image.Image], List[List[Image.Image]]],
|
| 45 |
+
) -> BatchFeature:
|
| 46 |
+
|
| 47 |
+
if isinstance(images[0], list):
|
| 48 |
+
images = cast(List[List[Image.Image]], images)
|
| 49 |
+
text_doc = []
|
| 50 |
+
for i in range(len(images)):
|
| 51 |
+
conversation = [
|
| 52 |
+
{"role": "user", "content": [{"type": "image"}] * len(images[i])}
|
| 53 |
+
]
|
| 54 |
+
template = self.apply_chat_template(
|
| 55 |
+
conversation, add_generation_prompt=False
|
| 56 |
+
)
|
| 57 |
+
text_doc.append(template[self.assistant_prefix_len :])
|
| 58 |
+
|
| 59 |
+
else:
|
| 60 |
+
images = cast(List[Image.Image], images)
|
| 61 |
+
text_doc = [
|
| 62 |
+
"<|im_start|>user\n<|vision_start|><|image_pad|><|vision_end|>Describe the image.<|im_end|>\n"
|
| 63 |
+
] * len(images)
|
| 64 |
+
|
| 65 |
+
# The following code is a hack to make sure the scatter in DDP is done correctly when training on multiple GPUs
|
| 66 |
+
batch_doc = self(text=text_doc, images=images, padding="longest", return_tensors="pt") # type: ignore
|
| 67 |
+
# Separate pixel_values for each image
|
| 68 |
+
offsets = batch_doc["image_grid_thw"][:, 1] * batch_doc["image_grid_thw"][:, 2]
|
| 69 |
+
# Pad pixel_values to the same length to be able to make it into a tensor
|
| 70 |
+
pixel_values = torch.split(batch_doc["pixel_values"], offsets.tolist())
|
| 71 |
+
|
| 72 |
+
max_length = max([len(pv) for pv in pixel_values])
|
| 73 |
+
|
| 74 |
+
pixel_values = [
|
| 75 |
+
torch.cat(
|
| 76 |
+
[
|
| 77 |
+
pv,
|
| 78 |
+
torch.zeros(
|
| 79 |
+
(max_length - len(pv), pv.shape[1]),
|
| 80 |
+
dtype=pv.dtype,
|
| 81 |
+
device=pv.device,
|
| 82 |
+
),
|
| 83 |
+
]
|
| 84 |
+
)
|
| 85 |
+
for pv in pixel_values
|
| 86 |
+
]
|
| 87 |
+
|
| 88 |
+
batch_doc["pixel_values"] = torch.stack(pixel_values)
|
| 89 |
+
return batch_doc
|
| 90 |
+
|
| 91 |
+
def process_texts(
|
| 92 |
+
self,
|
| 93 |
+
texts: List[str],
|
| 94 |
+
max_length: Optional[int] = None,
|
| 95 |
+
prefix: Optional[str] = None,
|
| 96 |
+
padding: Optional[str] = None,
|
| 97 |
+
) -> BatchFeature:
|
| 98 |
+
|
| 99 |
+
max_length = (
|
| 100 |
+
self.text_max_length
|
| 101 |
+
if max_length is None
|
| 102 |
+
else min(max_length, self.text_max_length)
|
| 103 |
+
)
|
| 104 |
+
padded_texts: List[str] = []
|
| 105 |
+
|
| 106 |
+
for text in texts:
|
| 107 |
+
if prefix:
|
| 108 |
+
text = f"{prefix}: {text}"
|
| 109 |
+
padded_texts.append(text)
|
| 110 |
+
|
| 111 |
+
text_batch = self(
|
| 112 |
+
text=padded_texts,
|
| 113 |
+
return_tensors="pt",
|
| 114 |
+
padding=padding or "longest",
|
| 115 |
+
max_length=max_length,
|
| 116 |
+
truncation=True,
|
| 117 |
+
)
|
| 118 |
+
|
| 119 |
+
return text_batch
|
| 120 |
+
|
| 121 |
+
|
| 122 |
+
@dataclass
|
| 123 |
+
class JinaEmbeddingsV4ModelOutput:
|
| 124 |
+
"""
|
| 125 |
+
Base class for the Hybrid Model outputs.
|
| 126 |
+
Args:
|
| 127 |
+
vlm_last_hidden_states (torch.Tensor, optional): Last hidden states of the VLM.
|
| 128 |
+
single_vec_emb (torch.Tensor, optional): Single-vector embeddings.
|
| 129 |
+
multi_vec_emb (torch.Tensor, optional): Multi-vector embeddings.
|
| 130 |
+
"""
|
| 131 |
+
|
| 132 |
+
vlm_last_hidden_states: Optional[torch.Tensor] = None
|
| 133 |
+
single_vec_emb: Optional[torch.Tensor] = None
|
| 134 |
+
multi_vec_emb: Optional[torch.Tensor] = None
|
| 135 |
+
|
| 136 |
+
|
| 137 |
+
class JinaEmbeddingsV4Model(Qwen2_5_VLForConditionalGeneration):
|
| 138 |
+
config_class = JinaEmbeddingsV4Config
|
| 139 |
+
main_input_name: ClassVar[str] = "doc_input_ids"
|
| 140 |
+
|
| 141 |
+
def __init__(self, config: JinaEmbeddingsV4Config):
|
| 142 |
+
Qwen2_5_VLForConditionalGeneration.__init__(self, config)
|
| 143 |
+
self._init_projection_layer(config)
|
| 144 |
+
self.post_init()
|
| 145 |
+
self.processor = JinaEmbeddingsV4Processor.from_pretrained(
|
| 146 |
+
self.name_or_path, trust_remote_code=True, use_fast=True
|
| 147 |
+
)
|
| 148 |
+
self.multi_vector_projector_dim = config.multi_vector_projector_dim
|
| 149 |
+
self.verbosity = config.verbosity
|
| 150 |
+
self._task = None
|
| 151 |
+
|
| 152 |
+
@property
|
| 153 |
+
def task(self) -> Optional[str]:
|
| 154 |
+
"""Get the current task set for the model."""
|
| 155 |
+
return self._task
|
| 156 |
+
|
| 157 |
+
@task.setter
|
| 158 |
+
def task(self, task: str):
|
| 159 |
+
"""
|
| 160 |
+
Set the task for the model.
|
| 161 |
+
|
| 162 |
+
Args:
|
| 163 |
+
task (str): The task name. Must be one of ['retrieval', 'text-matching', 'code']
|
| 164 |
+
"""
|
| 165 |
+
if task not in self.config.task_names:
|
| 166 |
+
raise ValueError(
|
| 167 |
+
f"Invalid task: {task}. Must be one of {self.config.task_names}."
|
| 168 |
+
)
|
| 169 |
+
self._task = task
|
| 170 |
+
|
| 171 |
+
def get_last_hidden_states(
|
| 172 |
+
self,
|
| 173 |
+
task_label: Union[str, List[str]],
|
| 174 |
+
input_ids: torch.LongTensor,
|
| 175 |
+
attention_mask: torch.Tensor,
|
| 176 |
+
**kwargs,
|
| 177 |
+
) -> torch.Tensor:
|
| 178 |
+
if "pixel_values" in kwargs:
|
| 179 |
+
offsets = kwargs["image_grid_thw"][:, 1] * kwargs["image_grid_thw"][:, 2]
|
| 180 |
+
kwargs["pixel_values"] = torch.cat(
|
| 181 |
+
[pv[:o] for pv, o in zip(kwargs["pixel_values"], offsets)], dim=0
|
| 182 |
+
)
|
| 183 |
+
position_ids, rope_deltas = self.model.get_rope_index(
|
| 184 |
+
input_ids=input_ids,
|
| 185 |
+
image_grid_thw=kwargs.get("image_grid_thw", None),
|
| 186 |
+
attention_mask=attention_mask,
|
| 187 |
+
)
|
| 188 |
+
|
| 189 |
+
kwargs["output_hidden_states"] = True
|
| 190 |
+
outputs = super().forward(
|
| 191 |
+
task_label=task_label,
|
| 192 |
+
input_ids=input_ids,
|
| 193 |
+
attention_mask=attention_mask,
|
| 194 |
+
**kwargs,
|
| 195 |
+
position_ids=position_ids,
|
| 196 |
+
rope_deltas=rope_deltas,
|
| 197 |
+
use_cache=False,
|
| 198 |
+
)
|
| 199 |
+
|
| 200 |
+
hidden_states = outputs.hidden_states
|
| 201 |
+
if not hidden_states:
|
| 202 |
+
raise ValueError("Hidden states not found in model output")
|
| 203 |
+
|
| 204 |
+
return hidden_states[-1]
|
| 205 |
+
|
| 206 |
+
def _init_projection_layer(self, config) -> None:
|
| 207 |
+
"""
|
| 208 |
+
Initializes projection layers.
|
| 209 |
+
"""
|
| 210 |
+
self.config.multi_vector_projector_dim = config.multi_vector_projector_dim
|
| 211 |
+
|
| 212 |
+
self.multi_vector_projector = nn.Linear(
|
| 213 |
+
in_features=self.config.text_config.hidden_size,
|
| 214 |
+
out_features=self.config.multi_vector_projector_dim,
|
| 215 |
+
)
|
| 216 |
+
|
| 217 |
+
def get_single_vector_embeddings(
|
| 218 |
+
self,
|
| 219 |
+
hidden_states: torch.Tensor,
|
| 220 |
+
attention_mask: torch.Tensor,
|
| 221 |
+
input_ids: Optional[torch.LongTensor] = None,
|
| 222 |
+
) -> torch.Tensor:
|
| 223 |
+
"""
|
| 224 |
+
Get the single-vector embeddings from the hidden states.
|
| 225 |
+
"""
|
| 226 |
+
if self._input_has_image(input_ids[0]): # got document image
|
| 227 |
+
img_start_positions = torch.where(
|
| 228 |
+
input_ids == self.config.vision_start_token_id
|
| 229 |
+
)[1]
|
| 230 |
+
img_end_positions = torch.where(
|
| 231 |
+
input_ids == self.config.vision_end_token_id
|
| 232 |
+
)[1]
|
| 233 |
+
|
| 234 |
+
batch_size, seq_len = input_ids.shape
|
| 235 |
+
position_indices = torch.arange(seq_len, device=input_ids.device).expand(
|
| 236 |
+
batch_size, -1
|
| 237 |
+
)
|
| 238 |
+
image_mask = (position_indices >= img_start_positions.unsqueeze(1)) & (
|
| 239 |
+
position_indices <= img_end_positions.unsqueeze(1)
|
| 240 |
+
)
|
| 241 |
+
|
| 242 |
+
masked_hidden_states = hidden_states * image_mask.unsqueeze(-1)
|
| 243 |
+
pooled_output = masked_hidden_states.sum(dim=1) / image_mask.sum(
|
| 244 |
+
dim=1, keepdim=True
|
| 245 |
+
)
|
| 246 |
+
else: # got query text
|
| 247 |
+
pooled_output = torch.sum(
|
| 248 |
+
hidden_states * attention_mask.unsqueeze(-1), dim=1
|
| 249 |
+
) / torch.sum(attention_mask, dim=1, keepdim=True)
|
| 250 |
+
|
| 251 |
+
return torch.nn.functional.normalize(pooled_output, dim=-1)
|
| 252 |
+
|
| 253 |
+
def get_multi_vector_embeddings(
|
| 254 |
+
self,
|
| 255 |
+
task_label: Union[str, List[str]],
|
| 256 |
+
hidden_states: torch.Tensor,
|
| 257 |
+
attention_mask: torch.Tensor,
|
| 258 |
+
) -> torch.Tensor:
|
| 259 |
+
"""
|
| 260 |
+
Project the hidden states to multi-vector embeddings.
|
| 261 |
+
"""
|
| 262 |
+
multi_vec_emb = self.multi_vector_projector(
|
| 263 |
+
hidden_states, task_label=task_label
|
| 264 |
+
)
|
| 265 |
+
multi_vec_emb = torch.nn.functional.normalize(multi_vec_emb, dim=-1)
|
| 266 |
+
return multi_vec_emb * attention_mask.unsqueeze(-1)
|
| 267 |
+
|
| 268 |
+
def _input_has_image(self, input_ids):
|
| 269 |
+
return self.config.vision_start_token_id in input_ids
|
| 270 |
+
|
| 271 |
+
def forward(
|
| 272 |
+
self,
|
| 273 |
+
task_label: Union[str, List[str]],
|
| 274 |
+
input_ids: torch.LongTensor,
|
| 275 |
+
attention_mask: torch.Tensor,
|
| 276 |
+
output_vlm_last_hidden_states: bool = False,
|
| 277 |
+
**kwargs,
|
| 278 |
+
) -> JinaEmbeddingsV4ModelOutput:
|
| 279 |
+
"""
|
| 280 |
+
Forward pass through the model. Returns both single-vector and multi-vector embeddings.
|
| 281 |
+
Args:
|
| 282 |
+
input_ids (torch.Tensor): The input tokens tensor.
|
| 283 |
+
attention_mask (torch.Tensor): The attention mask tensor.
|
| 284 |
+
Returns:
|
| 285 |
+
JinaEmbeddingsV4ModelOutput:
|
| 286 |
+
vlm_last_hidden_states (torch.Tensor, optional): Last hidden states of the VLM.
|
| 287 |
+
single_vec_emb (torch.Tensor, optional): Single-vector embeddings.
|
| 288 |
+
multi_vec_emb (torch.Tensor, optional): Multi-vector embeddings.
|
| 289 |
+
"""
|
| 290 |
+
# Forward pass through the VLM
|
| 291 |
+
hidden_states = self.get_last_hidden_states(
|
| 292 |
+
input_ids=input_ids,
|
| 293 |
+
attention_mask=attention_mask,
|
| 294 |
+
task_label=task_label,
|
| 295 |
+
**kwargs,
|
| 296 |
+
) # (batch_size, seq_length, hidden_size)
|
| 297 |
+
# Compute the embeddings
|
| 298 |
+
single_vec_emb = self.get_single_vector_embeddings(
|
| 299 |
+
hidden_states=hidden_states,
|
| 300 |
+
attention_mask=attention_mask,
|
| 301 |
+
input_ids=input_ids,
|
| 302 |
+
)
|
| 303 |
+
multi_vec_emb = self.get_multi_vector_embeddings(
|
| 304 |
+
hidden_states=hidden_states,
|
| 305 |
+
attention_mask=attention_mask,
|
| 306 |
+
task_label=task_label,
|
| 307 |
+
)
|
| 308 |
+
|
| 309 |
+
return JinaEmbeddingsV4ModelOutput(
|
| 310 |
+
vlm_last_hidden_states=(
|
| 311 |
+
hidden_states if output_vlm_last_hidden_states else None
|
| 312 |
+
),
|
| 313 |
+
single_vec_emb=single_vec_emb,
|
| 314 |
+
multi_vec_emb=multi_vec_emb,
|
| 315 |
+
)
|
| 316 |
+
|
| 317 |
+
def _process_batches(
|
| 318 |
+
self,
|
| 319 |
+
data: List[Union[str, Image.Image]],
|
| 320 |
+
task_label: Union[str, List[str]],
|
| 321 |
+
processor_fn: Callable,
|
| 322 |
+
desc: str,
|
| 323 |
+
return_multivector: bool = False,
|
| 324 |
+
return_numpy: bool = False,
|
| 325 |
+
batch_size: int = 32,
|
| 326 |
+
truncate_dim: Optional[int] = None,
|
| 327 |
+
) -> Union[np.ndarray, List[torch.Tensor]]:
|
| 328 |
+
dataloader = DataLoader(
|
| 329 |
+
dataset=data,
|
| 330 |
+
batch_size=batch_size,
|
| 331 |
+
shuffle=False,
|
| 332 |
+
collate_fn=processor_fn,
|
| 333 |
+
)
|
| 334 |
+
if return_multivector and len(data) > 1:
|
| 335 |
+
assert (
|
| 336 |
+
not return_numpy
|
| 337 |
+
), "`return_numpy` is not supported when `return_multivector=True` and more than one data is encoded"
|
| 338 |
+
results = []
|
| 339 |
+
self.eval()
|
| 340 |
+
for batch in tqdm(dataloader, desc=desc, disable=self.verbosity == 0):
|
| 341 |
+
with torch.no_grad():
|
| 342 |
+
batch = {k: v.to(self.device) for k, v in batch.items()}
|
| 343 |
+
with torch.autocast(
|
| 344 |
+
device_type=torch.device(self.device).type, dtype=torch.bfloat16
|
| 345 |
+
):
|
| 346 |
+
embeddings = self(**batch, task_label=task_label)
|
| 347 |
+
if not return_multivector:
|
| 348 |
+
embeddings = embeddings.single_vec_emb
|
| 349 |
+
if truncate_dim is not None:
|
| 350 |
+
embeddings = embeddings[:, :truncate_dim]
|
| 351 |
+
embeddings = torch.nn.functional.normalize(
|
| 352 |
+
embeddings, p=2, dim=-1
|
| 353 |
+
)
|
| 354 |
+
else:
|
| 355 |
+
embeddings = embeddings.multi_vec_emb
|
| 356 |
+
|
| 357 |
+
if return_multivector and not return_numpy:
|
| 358 |
+
valid_tokens = batch["attention_mask"].bool()
|
| 359 |
+
embeddings = [
|
| 360 |
+
emb[mask] for emb, mask in zip(embeddings, valid_tokens)
|
| 361 |
+
]
|
| 362 |
+
results.append(embeddings)
|
| 363 |
+
else:
|
| 364 |
+
results.append(
|
| 365 |
+
embeddings.cpu()
|
| 366 |
+
if return_numpy
|
| 367 |
+
else list(torch.unbind(embeddings))
|
| 368 |
+
)
|
| 369 |
+
if return_numpy:
|
| 370 |
+
return np.concatenate([result.numpy() for result in results], axis=0)
|
| 371 |
+
return [item for sublist in results for item in sublist]
|
| 372 |
+
|
| 373 |
+
def _validate_encoding_params(
|
| 374 |
+
self,
|
| 375 |
+
truncate_dim: Optional[int] = None,
|
| 376 |
+
prompt_name: Optional[str] = None,
|
| 377 |
+
) -> Dict[str, Any]:
|
| 378 |
+
encode_kwargs = {}
|
| 379 |
+
if prompt_name is not None:
|
| 380 |
+
if prompt_name not in PREFIX_DICT:
|
| 381 |
+
raise ValueError(
|
| 382 |
+
f"Invalid prompt_name: {prompt_name}. Must be one of {list(PREFIX_DICT.keys())}."
|
| 383 |
+
)
|
| 384 |
+
else:
|
| 385 |
+
encode_kwargs["prefix"] = (
|
| 386 |
+
PREFIX_DICT[prompt_name]
|
| 387 |
+
if self.task != "text-matching"
|
| 388 |
+
else PREFIX_DICT["query"]
|
| 389 |
+
)
|
| 390 |
+
|
| 391 |
+
truncate_dim = truncate_dim or self.config.truncate_dim
|
| 392 |
+
if truncate_dim is not None and truncate_dim not in self.config.matryoshka_dims:
|
| 393 |
+
raise ValueError(
|
| 394 |
+
f"Invalid truncate_dim: {truncate_dim}. Must be one of {self.config.matryoshka_dims}."
|
| 395 |
+
)
|
| 396 |
+
else:
|
| 397 |
+
encode_kwargs["truncate_dim"] = truncate_dim
|
| 398 |
+
|
| 399 |
+
return encode_kwargs
|
| 400 |
+
|
| 401 |
+
def _validate_task(self, task: Optional[str] = None) -> str:
|
| 402 |
+
if task is None:
|
| 403 |
+
if self.task is None:
|
| 404 |
+
raise ValueError(
|
| 405 |
+
"Task must be specified before encoding data. You can set it either as a model property "
|
| 406 |
+
"(e.g., model.task = 'retrieval') or pass it as an argument to the encode method."
|
| 407 |
+
)
|
| 408 |
+
task = self.task
|
| 409 |
+
else:
|
| 410 |
+
if task not in self.config.task_names:
|
| 411 |
+
raise ValueError(
|
| 412 |
+
f"Invalid task: {task}. Must be one of {self.config.task_names}."
|
| 413 |
+
)
|
| 414 |
+
return task
|
| 415 |
+
|
| 416 |
+
def encode_text(
|
| 417 |
+
self,
|
| 418 |
+
texts: Union[str, List[str]],
|
| 419 |
+
task: Optional[str] = None,
|
| 420 |
+
max_length: int = 32768,
|
| 421 |
+
batch_size: int = 8,
|
| 422 |
+
return_multivector: bool = False,
|
| 423 |
+
return_numpy: bool = False,
|
| 424 |
+
truncate_dim: Optional[int] = None,
|
| 425 |
+
prompt_name: Optional[str] = None,
|
| 426 |
+
) -> Union[List[torch.Tensor], torch.Tensor]:
|
| 427 |
+
"""
|
| 428 |
+
Encodes a list of texts into embeddings.
|
| 429 |
+
|
| 430 |
+
Args:
|
| 431 |
+
texts: text or list of text strings to encode
|
| 432 |
+
max_length: Maximum token length for text processing
|
| 433 |
+
batch_size: Number of texts to process at once
|
| 434 |
+
return_multivector: Whether to return multi-vector embeddings instead of single-vector embeddings
|
| 435 |
+
return_numpy: Whether to return numpy arrays instead of torch tensors
|
| 436 |
+
truncate_dim: Dimension to truncate embeddings to (128, 256, 512, or 1024)
|
| 437 |
+
prompt_name: Type of text being encoded ('query' or 'passage')
|
| 438 |
+
|
| 439 |
+
Returns:
|
| 440 |
+
List of text embeddings as tensors or numpy arrays when encoding multiple texts, or single text embedding as tensor when encoding a single text
|
| 441 |
+
"""
|
| 442 |
+
prompt_name = prompt_name or "query"
|
| 443 |
+
encode_kwargs = self._validate_encoding_params(
|
| 444 |
+
truncate_dim=truncate_dim, prompt_name=prompt_name
|
| 445 |
+
)
|
| 446 |
+
|
| 447 |
+
task = self._validate_task(task)
|
| 448 |
+
|
| 449 |
+
processor_fn = partial(
|
| 450 |
+
self.processor.process_texts,
|
| 451 |
+
max_length=max_length,
|
| 452 |
+
prefix=encode_kwargs.pop("prefix"),
|
| 453 |
+
)
|
| 454 |
+
|
| 455 |
+
return_list = isinstance(texts, list)
|
| 456 |
+
|
| 457 |
+
# If return_multivector is True and encoding multiple texts, ignore return_numpy
|
| 458 |
+
if return_multivector and return_list and len(texts) > 1:
|
| 459 |
+
if return_numpy:
|
| 460 |
+
print(
|
| 461 |
+
"Warning: `return_numpy` is ignored when `return_multivector=True` and `len(texts) > 1`"
|
| 462 |
+
)
|
| 463 |
+
return_numpy = False
|
| 464 |
+
|
| 465 |
+
if isinstance(texts, str):
|
| 466 |
+
texts = [texts]
|
| 467 |
+
|
| 468 |
+
embeddings = self._process_batches(
|
| 469 |
+
data=texts,
|
| 470 |
+
processor_fn=processor_fn,
|
| 471 |
+
desc="Encoding texts...",
|
| 472 |
+
task_label=task,
|
| 473 |
+
return_multivector=return_multivector,
|
| 474 |
+
return_numpy=return_numpy,
|
| 475 |
+
batch_size=batch_size,
|
| 476 |
+
**encode_kwargs,
|
| 477 |
+
)
|
| 478 |
+
|
| 479 |
+
return embeddings if return_list else embeddings[0]
|
| 480 |
+
|
| 481 |
+
def _load_images_if_needed(
|
| 482 |
+
self, images: List[Union[str, Image.Image]]
|
| 483 |
+
) -> List[Image.Image]:
|
| 484 |
+
loaded_images = []
|
| 485 |
+
for image in images:
|
| 486 |
+
if isinstance(image, str):
|
| 487 |
+
if image.startswith("http"):
|
| 488 |
+
response = requests.get(image)
|
| 489 |
+
image = Image.open(BytesIO(response.content)).convert("RGB")
|
| 490 |
+
else:
|
| 491 |
+
image = Image.open(image).convert("RGB")
|
| 492 |
+
loaded_images.append(image)
|
| 493 |
+
return loaded_images
|
| 494 |
+
|
| 495 |
+
def encode_image(
|
| 496 |
+
self,
|
| 497 |
+
images: Union[str, Image.Image, List[Union[str, Image.Image]]],
|
| 498 |
+
task: Optional[str] = None,
|
| 499 |
+
batch_size: int = 8,
|
| 500 |
+
return_multivector: bool = False,
|
| 501 |
+
return_numpy: bool = False,
|
| 502 |
+
truncate_dim: Optional[int] = None,
|
| 503 |
+
max_pixels: Optional[int] = None,
|
| 504 |
+
) -> Union[List[torch.Tensor], torch.Tensor]:
|
| 505 |
+
"""
|
| 506 |
+
Encodes a list of images or a single image into embedding(s).
|
| 507 |
+
|
| 508 |
+
Args:
|
| 509 |
+
images: image(s) to encode, can be PIL Image(s), URL(s), or local file path(s)
|
| 510 |
+
batch_size: Number of images to process at once
|
| 511 |
+
return_multivector: Whether to return multi-vector embeddings instead of single-vector embeddings
|
| 512 |
+
return_numpy: Whether to return numpy arrays instead of torch tensors. If `return_multivector` is `True` and more than one image is encoded, this parameter is ignored.
|
| 513 |
+
truncate_dim: Dimension to truncate embeddings to (128, 256, 512, or 1024)
|
| 514 |
+
max_pixels: Maximum number of pixels to process per image
|
| 515 |
+
|
| 516 |
+
Returns:
|
| 517 |
+
List of image embeddings as tensors or numpy arrays when encoding multiple images, or single image embedding as tensor when encoding a single image
|
| 518 |
+
"""
|
| 519 |
+
if max_pixels:
|
| 520 |
+
default_max_pixels = self.processor.image_processor.max_pixels
|
| 521 |
+
self.processor.image_processor.max_pixels = (
|
| 522 |
+
max_pixels # change during encoding
|
| 523 |
+
)
|
| 524 |
+
encode_kwargs = self._validate_encoding_params(truncate_dim=truncate_dim)
|
| 525 |
+
task = self._validate_task(task)
|
| 526 |
+
|
| 527 |
+
return_list = isinstance(images, list)
|
| 528 |
+
|
| 529 |
+
# If return_multivector is True and encoding multiple images, ignore return_numpy
|
| 530 |
+
if return_multivector and return_list and len(images) > 1:
|
| 531 |
+
if return_numpy:
|
| 532 |
+
print(
|
| 533 |
+
"Warning: `return_numpy` is ignored when `return_multivector=True` and `len(images) > 1`"
|
| 534 |
+
)
|
| 535 |
+
return_numpy = False
|
| 536 |
+
|
| 537 |
+
# Convert single image to list
|
| 538 |
+
if isinstance(images, (str, Image.Image)):
|
| 539 |
+
images = [images]
|
| 540 |
+
|
| 541 |
+
images = self._load_images_if_needed(images)
|
| 542 |
+
embeddings = self._process_batches(
|
| 543 |
+
data=images,
|
| 544 |
+
processor_fn=self.processor.process_images,
|
| 545 |
+
desc="Encoding images...",
|
| 546 |
+
task_label=task,
|
| 547 |
+
batch_size=batch_size,
|
| 548 |
+
return_multivector=return_multivector,
|
| 549 |
+
return_numpy=return_numpy,
|
| 550 |
+
**encode_kwargs,
|
| 551 |
+
)
|
| 552 |
+
|
| 553 |
+
if max_pixels:
|
| 554 |
+
self.processor.image_processor.max_pixels = default_max_pixels
|
| 555 |
+
|
| 556 |
+
return embeddings if return_list else embeddings[0]
|
| 557 |
+
|
| 558 |
+
@classmethod
|
| 559 |
+
def from_pretrained(
|
| 560 |
+
cls,
|
| 561 |
+
pretrained_model_name_or_path,
|
| 562 |
+
*args,
|
| 563 |
+
**kwargs,
|
| 564 |
+
):
|
| 565 |
+
"""
|
| 566 |
+
Loads a pretrained model and configures it with the appropriate task adapter (`retrieval` by default).
|
| 567 |
+
"""
|
| 568 |
+
if "torch_dtype" not in kwargs:
|
| 569 |
+
kwargs["torch_dtype"] = "auto"
|
| 570 |
+
|
| 571 |
+
kwargs["key_mapping"] = super()._checkpoint_conversion_mapping
|
| 572 |
+
if not is_flash_attn_2_available():
|
| 573 |
+
kwargs["attn_implementation"] = "sdpa"
|
| 574 |
+
|
| 575 |
+
base_model = super().from_pretrained(
|
| 576 |
+
pretrained_model_name_or_path, *args, **kwargs
|
| 577 |
+
)
|
| 578 |
+
|
| 579 |
+
# Configure adapter directory
|
| 580 |
+
if os.path.isdir(base_model.name_or_path):
|
| 581 |
+
adapter_dir = os.path.join(base_model.name_or_path, "adapters")
|
| 582 |
+
else:
|
| 583 |
+
adapter_cache_path = snapshot_download(
|
| 584 |
+
repo_id=base_model.name_or_path, allow_patterns=["adapters/*"]
|
| 585 |
+
)
|
| 586 |
+
adapter_dir = os.path.join(adapter_cache_path, "adapters")
|
| 587 |
+
|
| 588 |
+
lora_config = LoraConfig.from_pretrained(adapter_dir)
|
| 589 |
+
lora_config._custom_modules = {
|
| 590 |
+
torch.nn.modules.linear.Linear: partial(
|
| 591 |
+
MultiAdapterLinear,
|
| 592 |
+
task_names=base_model.config.task_names,
|
| 593 |
+
)
|
| 594 |
+
}
|
| 595 |
+
peft_model = PeftModel.from_pretrained(
|
| 596 |
+
model=base_model,
|
| 597 |
+
model_id=adapter_dir,
|
| 598 |
+
config=lora_config,
|
| 599 |
+
)
|
| 600 |
+
|
| 601 |
+
def task_getter(self):
|
| 602 |
+
return self.model.task
|
| 603 |
+
|
| 604 |
+
def task_setter(self, value):
|
| 605 |
+
self.model.task = value
|
| 606 |
+
|
| 607 |
+
peft_model.__class__.task = property(task_getter, task_setter)
|
| 608 |
+
|
| 609 |
+
return peft_model
|
modules.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"idx": 0,
|
| 4 |
+
"name": "transformer",
|
| 5 |
+
"path": "",
|
| 6 |
+
"type": "custom_st.Transformer",
|
| 7 |
+
"kwargs": ["task", "truncate_dim"]
|
| 8 |
+
}
|
| 9 |
+
]
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": true,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.48145466,
|
| 8 |
+
0.4578275,
|
| 9 |
+
0.40821073
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "Qwen2VLImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.26862954,
|
| 14 |
+
0.26130258,
|
| 15 |
+
0.27577711
|
| 16 |
+
],
|
| 17 |
+
"max_pixels": 602112,
|
| 18 |
+
"merge_size": 2,
|
| 19 |
+
"min_pixels": 3136,
|
| 20 |
+
"patch_size": 14,
|
| 21 |
+
"processor_class": "JinaEmbeddingsV4Processor",
|
| 22 |
+
"resample": 3,
|
| 23 |
+
"rescale_factor": 0.00392156862745098,
|
| 24 |
+
"video_processor_type": "Qwen2VLVideoProcessor",
|
| 25 |
+
"size": {
|
| 26 |
+
"longest_edge": 602112,
|
| 27 |
+
"shortest_edge": 3136
|
| 28 |
+
},
|
| 29 |
+
"temporal_patch_size": 2,
|
| 30 |
+
"auto_map": {
|
| 31 |
+
"AutoProcessor": "modeling_jina_embeddings_v4.JinaEmbeddingsV4Processor"
|
| 32 |
+
}
|
| 33 |
+
}
|
qwen2_5_vl.py
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
results.json
ADDED
|
@@ -0,0 +1,582 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"arxivqa_test_subsampled": {
|
| 3 |
+
"ndcg_at_1": 0.844,
|
| 4 |
+
"ndcg_at_3": 0.88524,
|
| 5 |
+
"ndcg_at_5": 0.88954,
|
| 6 |
+
"ndcg_at_10": 0.89512,
|
| 7 |
+
"ndcg_at_20": 0.90085,
|
| 8 |
+
"ndcg_at_50": 0.90479,
|
| 9 |
+
"ndcg_at_100": 0.90578,
|
| 10 |
+
"map_at_1": 0.844,
|
| 11 |
+
"map_at_3": 0.87467,
|
| 12 |
+
"map_at_5": 0.87717,
|
| 13 |
+
"map_at_10": 0.87933,
|
| 14 |
+
"map_at_20": 0.88099,
|
| 15 |
+
"map_at_50": 0.88161,
|
| 16 |
+
"map_at_100": 0.8817,
|
| 17 |
+
"recall_at_1": 0.844,
|
| 18 |
+
"recall_at_3": 0.916,
|
| 19 |
+
"recall_at_5": 0.926,
|
| 20 |
+
"recall_at_10": 0.944,
|
| 21 |
+
"recall_at_20": 0.966,
|
| 22 |
+
"recall_at_50": 0.986,
|
| 23 |
+
"recall_at_100": 0.992,
|
| 24 |
+
"precision_at_1": 0.844,
|
| 25 |
+
"precision_at_3": 0.30533,
|
| 26 |
+
"precision_at_5": 0.1852,
|
| 27 |
+
"precision_at_10": 0.0944,
|
| 28 |
+
"precision_at_20": 0.0483,
|
| 29 |
+
"precision_at_50": 0.01972,
|
| 30 |
+
"precision_at_100": 0.00992,
|
| 31 |
+
"mrr_at_1": 0.844,
|
| 32 |
+
"mrr_at_3": 0.8746666666666665,
|
| 33 |
+
"mrr_at_5": 0.8771666666666665,
|
| 34 |
+
"mrr_at_10": 0.8793301587301586,
|
| 35 |
+
"mrr_at_20": 0.880986183261183,
|
| 36 |
+
"mrr_at_50": 0.8816066058267283,
|
| 37 |
+
"mrr_at_100": 0.8816959272950264,
|
| 38 |
+
"naucs_at_1_max": 0.7413901379085128,
|
| 39 |
+
"naucs_at_1_std": 0.3454872013866209,
|
| 40 |
+
"naucs_at_1_diff1": 0.9600906830113787,
|
| 41 |
+
"naucs_at_3_max": 0.7713307545240329,
|
| 42 |
+
"naucs_at_3_std": 0.4801698457160663,
|
| 43 |
+
"naucs_at_3_diff1": 0.9489240140500664,
|
| 44 |
+
"naucs_at_5_max": 0.7514699573523106,
|
| 45 |
+
"naucs_at_5_std": 0.4375552022610836,
|
| 46 |
+
"naucs_at_5_diff1": 0.9526206879148043,
|
| 47 |
+
"naucs_at_10_max": 0.8086901427237575,
|
| 48 |
+
"naucs_at_10_std": 0.5144891289849284,
|
| 49 |
+
"naucs_at_10_diff1": 0.9513972255568919,
|
| 50 |
+
"naucs_at_20_max": 0.907453177349375,
|
| 51 |
+
"naucs_at_20_std": 0.5683802932937894,
|
| 52 |
+
"naucs_at_20_diff1": 0.9692425990003846,
|
| 53 |
+
"naucs_at_50_max": 0.8709483793517359,
|
| 54 |
+
"naucs_at_50_std": 0.7055488862211612,
|
| 55 |
+
"naucs_at_50_diff1": 0.9626517273576126,
|
| 56 |
+
"naucs_at_100_max": 0.8068394024276366,
|
| 57 |
+
"naucs_at_100_std": 0.7076330532212914,
|
| 58 |
+
"naucs_at_100_diff1": 0.9673202614378978
|
| 59 |
+
},
|
| 60 |
+
"docvqa_test_subsampled": {
|
| 61 |
+
"ndcg_at_1": 0.52328,
|
| 62 |
+
"ndcg_at_3": 0.5841,
|
| 63 |
+
"ndcg_at_5": 0.59975,
|
| 64 |
+
"ndcg_at_10": 0.62669,
|
| 65 |
+
"ndcg_at_20": 0.64245,
|
| 66 |
+
"ndcg_at_50": 0.65661,
|
| 67 |
+
"ndcg_at_100": 0.66492,
|
| 68 |
+
"map_at_1": 0.52328,
|
| 69 |
+
"map_at_3": 0.56911,
|
| 70 |
+
"map_at_5": 0.57786,
|
| 71 |
+
"map_at_10": 0.58881,
|
| 72 |
+
"map_at_20": 0.59317,
|
| 73 |
+
"map_at_50": 0.59548,
|
| 74 |
+
"map_at_100": 0.59622,
|
| 75 |
+
"recall_at_1": 0.52328,
|
| 76 |
+
"recall_at_3": 0.62749,
|
| 77 |
+
"recall_at_5": 0.66519,
|
| 78 |
+
"recall_at_10": 0.74945,
|
| 79 |
+
"recall_at_20": 0.81153,
|
| 80 |
+
"recall_at_50": 0.88248,
|
| 81 |
+
"recall_at_100": 0.93348,
|
| 82 |
+
"precision_at_1": 0.52328,
|
| 83 |
+
"precision_at_3": 0.20916,
|
| 84 |
+
"precision_at_5": 0.13304,
|
| 85 |
+
"precision_at_10": 0.07494,
|
| 86 |
+
"precision_at_20": 0.04058,
|
| 87 |
+
"precision_at_50": 0.01765,
|
| 88 |
+
"precision_at_100": 0.00933,
|
| 89 |
+
"mrr_at_1": 0.5232815964523282,
|
| 90 |
+
"mrr_at_3": 0.5691056910569108,
|
| 91 |
+
"mrr_at_5": 0.5778640059127865,
|
| 92 |
+
"mrr_at_10": 0.5888132193010243,
|
| 93 |
+
"mrr_at_20": 0.5931663069177401,
|
| 94 |
+
"mrr_at_50": 0.5954783504735428,
|
| 95 |
+
"mrr_at_100": 0.5962169799244146,
|
| 96 |
+
"naucs_at_1_max": 0.46089368028029637,
|
| 97 |
+
"naucs_at_1_std": 0.19359243300005127,
|
| 98 |
+
"naucs_at_1_diff1": 0.8483527783001977,
|
| 99 |
+
"naucs_at_3_max": 0.4640279399849662,
|
| 100 |
+
"naucs_at_3_std": 0.1814509120980464,
|
| 101 |
+
"naucs_at_3_diff1": 0.7719022256243834,
|
| 102 |
+
"naucs_at_5_max": 0.45716016762761796,
|
| 103 |
+
"naucs_at_5_std": 0.16428980258139747,
|
| 104 |
+
"naucs_at_5_diff1": 0.750196647594659,
|
| 105 |
+
"naucs_at_10_max": 0.3956528364820721,
|
| 106 |
+
"naucs_at_10_std": 0.09973122080056422,
|
| 107 |
+
"naucs_at_10_diff1": 0.7237863238311393,
|
| 108 |
+
"naucs_at_20_max": 0.35927664451426317,
|
| 109 |
+
"naucs_at_20_std": 0.09080366240903168,
|
| 110 |
+
"naucs_at_20_diff1": 0.6946736504983693,
|
| 111 |
+
"naucs_at_50_max": 0.3626447370884348,
|
| 112 |
+
"naucs_at_50_std": 0.2775120087087966,
|
| 113 |
+
"naucs_at_50_diff1": 0.6534710933108262,
|
| 114 |
+
"naucs_at_100_max": 0.32155287639122004,
|
| 115 |
+
"naucs_at_100_std": 0.3495021025151782,
|
| 116 |
+
"naucs_at_100_diff1": 0.6165810885563539
|
| 117 |
+
},
|
| 118 |
+
"infovqa_test_subsampled": {
|
| 119 |
+
"ndcg_at_1": 0.90283,
|
| 120 |
+
"ndcg_at_3": 0.93062,
|
| 121 |
+
"ndcg_at_5": 0.93567,
|
| 122 |
+
"ndcg_at_10": 0.93969,
|
| 123 |
+
"ndcg_at_20": 0.94324,
|
| 124 |
+
"ndcg_at_50": 0.94401,
|
| 125 |
+
"ndcg_at_100": 0.945,
|
| 126 |
+
"map_at_1": 0.90283,
|
| 127 |
+
"map_at_3": 0.92409,
|
| 128 |
+
"map_at_5": 0.92692,
|
| 129 |
+
"map_at_10": 0.92863,
|
| 130 |
+
"map_at_20": 0.92959,
|
| 131 |
+
"map_at_50": 0.9297,
|
| 132 |
+
"map_at_100": 0.92979,
|
| 133 |
+
"recall_at_1": 0.90283,
|
| 134 |
+
"recall_at_3": 0.94939,
|
| 135 |
+
"recall_at_5": 0.96154,
|
| 136 |
+
"recall_at_10": 0.97368,
|
| 137 |
+
"recall_at_20": 0.98785,
|
| 138 |
+
"recall_at_50": 0.9919,
|
| 139 |
+
"recall_at_100": 0.99798,
|
| 140 |
+
"precision_at_1": 0.90283,
|
| 141 |
+
"precision_at_3": 0.31646,
|
| 142 |
+
"precision_at_5": 0.19231,
|
| 143 |
+
"precision_at_10": 0.09737,
|
| 144 |
+
"precision_at_20": 0.04939,
|
| 145 |
+
"precision_at_50": 0.01984,
|
| 146 |
+
"precision_at_100": 0.00998,
|
| 147 |
+
"mrr_at_1": 0.902834008097166,
|
| 148 |
+
"mrr_at_3": 0.9240890688259108,
|
| 149 |
+
"mrr_at_5": 0.9269230769230767,
|
| 150 |
+
"mrr_at_10": 0.9286316753422016,
|
| 151 |
+
"mrr_at_20": 0.9295898610333593,
|
| 152 |
+
"mrr_at_50": 0.929699602843506,
|
| 153 |
+
"mrr_at_100": 0.929788457049907,
|
| 154 |
+
"naucs_at_1_max": 0.6026903076230651,
|
| 155 |
+
"naucs_at_1_std": 0.261936050485784,
|
| 156 |
+
"naucs_at_1_diff1": 0.9396804875719484,
|
| 157 |
+
"naucs_at_3_max": 0.7565375225904929,
|
| 158 |
+
"naucs_at_3_std": 0.45980620999702715,
|
| 159 |
+
"naucs_at_3_diff1": 0.9534218386220948,
|
| 160 |
+
"naucs_at_5_max": 0.8235249494008307,
|
| 161 |
+
"naucs_at_5_std": 0.5316999544043512,
|
| 162 |
+
"naucs_at_5_diff1": 0.9524604670358964,
|
| 163 |
+
"naucs_at_10_max": 0.8684766575602219,
|
| 164 |
+
"naucs_at_10_std": 0.5944713216706646,
|
| 165 |
+
"naucs_at_10_diff1": 0.9405654098266761,
|
| 166 |
+
"naucs_at_20_max": 0.7830887900175995,
|
| 167 |
+
"naucs_at_20_std": 0.5643438299512757,
|
| 168 |
+
"naucs_at_20_diff1": 0.8929919636352566,
|
| 169 |
+
"naucs_at_50_max": 0.7072835485426375,
|
| 170 |
+
"naucs_at_50_std": 0.5764614839135555,
|
| 171 |
+
"naucs_at_50_diff1": 0.8394879454528887,
|
| 172 |
+
"naucs_at_100_max": 1.0,
|
| 173 |
+
"naucs_at_100_std": 1.0,
|
| 174 |
+
"naucs_at_100_diff1": 1.0
|
| 175 |
+
},
|
| 176 |
+
"tabfquad_test_subsampled": {
|
| 177 |
+
"ndcg_at_1": 0.9,
|
| 178 |
+
"ndcg_at_3": 0.94685,
|
| 179 |
+
"ndcg_at_5": 0.95131,
|
| 180 |
+
"ndcg_at_10": 0.95366,
|
| 181 |
+
"ndcg_at_20": 0.95455,
|
| 182 |
+
"ndcg_at_50": 0.9553,
|
| 183 |
+
"ndcg_at_100": 0.9553,
|
| 184 |
+
"map_at_1": 0.9,
|
| 185 |
+
"map_at_3": 0.9369,
|
| 186 |
+
"map_at_5": 0.9394,
|
| 187 |
+
"map_at_10": 0.9404,
|
| 188 |
+
"map_at_20": 0.94063,
|
| 189 |
+
"map_at_50": 0.94077,
|
| 190 |
+
"map_at_100": 0.94077,
|
| 191 |
+
"recall_at_1": 0.9,
|
| 192 |
+
"recall_at_3": 0.975,
|
| 193 |
+
"recall_at_5": 0.98571,
|
| 194 |
+
"recall_at_10": 0.99286,
|
| 195 |
+
"recall_at_20": 0.99643,
|
| 196 |
+
"recall_at_50": 1.0,
|
| 197 |
+
"recall_at_100": 1.0,
|
| 198 |
+
"precision_at_1": 0.9,
|
| 199 |
+
"precision_at_3": 0.325,
|
| 200 |
+
"precision_at_5": 0.19714,
|
| 201 |
+
"precision_at_10": 0.09929,
|
| 202 |
+
"precision_at_20": 0.04982,
|
| 203 |
+
"precision_at_50": 0.02,
|
| 204 |
+
"precision_at_100": 0.01,
|
| 205 |
+
"mrr_at_1": 0.9,
|
| 206 |
+
"mrr_at_3": 0.936904761904762,
|
| 207 |
+
"mrr_at_5": 0.9394047619047617,
|
| 208 |
+
"mrr_at_10": 0.9403968253968255,
|
| 209 |
+
"mrr_at_20": 0.9406349206349207,
|
| 210 |
+
"mrr_at_50": 0.9407722832722833,
|
| 211 |
+
"mrr_at_100": 0.9407722832722833,
|
| 212 |
+
"naucs_at_1_max": 0.39284046952114193,
|
| 213 |
+
"naucs_at_1_std": 0.06274176337201544,
|
| 214 |
+
"naucs_at_1_diff1": 0.9321395224756563,
|
| 215 |
+
"naucs_at_3_max": 0.98132586367881,
|
| 216 |
+
"naucs_at_3_std": 0.9042950513538718,
|
| 217 |
+
"naucs_at_3_diff1": 0.98132586367881,
|
| 218 |
+
"naucs_at_5_max": 0.967320261437913,
|
| 219 |
+
"naucs_at_5_std": 0.8978758169934754,
|
| 220 |
+
"naucs_at_5_diff1": 1.0,
|
| 221 |
+
"naucs_at_10_max": 1.0,
|
| 222 |
+
"naucs_at_10_std": 0.9346405228758269,
|
| 223 |
+
"naucs_at_10_diff1": 1.0,
|
| 224 |
+
"naucs_at_20_max": 1.0,
|
| 225 |
+
"naucs_at_20_std": 1.0,
|
| 226 |
+
"naucs_at_20_diff1": 1.0,
|
| 227 |
+
"naucs_at_50_max": 1.0,
|
| 228 |
+
"naucs_at_50_std": 1.0,
|
| 229 |
+
"naucs_at_50_diff1": 1.0,
|
| 230 |
+
"naucs_at_100_max": 1.0,
|
| 231 |
+
"naucs_at_100_std": 1.0,
|
| 232 |
+
"naucs_at_100_diff1": 1.0
|
| 233 |
+
},
|
| 234 |
+
"tatdqa_test": {
|
| 235 |
+
"ndcg_at_1": 0.68834,
|
| 236 |
+
"ndcg_at_3": 0.7834,
|
| 237 |
+
"ndcg_at_5": 0.80344,
|
| 238 |
+
"ndcg_at_10": 0.81851,
|
| 239 |
+
"ndcg_at_20": 0.82469,
|
| 240 |
+
"ndcg_at_50": 0.82852,
|
| 241 |
+
"ndcg_at_100": 0.82981,
|
| 242 |
+
"map_at_1": 0.68834,
|
| 243 |
+
"map_at_3": 0.76073,
|
| 244 |
+
"map_at_5": 0.772,
|
| 245 |
+
"map_at_10": 0.7783,
|
| 246 |
+
"map_at_20": 0.78002,
|
| 247 |
+
"map_at_50": 0.78067,
|
| 248 |
+
"map_at_100": 0.78079,
|
| 249 |
+
"recall_at_1": 0.68834,
|
| 250 |
+
"recall_at_3": 0.84872,
|
| 251 |
+
"recall_at_5": 0.89672,
|
| 252 |
+
"recall_at_10": 0.94289,
|
| 253 |
+
"recall_at_20": 0.96719,
|
| 254 |
+
"recall_at_50": 0.98603,
|
| 255 |
+
"recall_at_100": 0.99392,
|
| 256 |
+
"precision_at_1": 0.68834,
|
| 257 |
+
"precision_at_3": 0.28291,
|
| 258 |
+
"precision_at_5": 0.17934,
|
| 259 |
+
"precision_at_10": 0.09429,
|
| 260 |
+
"precision_at_20": 0.04836,
|
| 261 |
+
"precision_at_50": 0.01972,
|
| 262 |
+
"precision_at_100": 0.00994,
|
| 263 |
+
"mrr_at_1": 0.6865127582017011,
|
| 264 |
+
"mrr_at_3": 0.7598217901984609,
|
| 265 |
+
"mrr_at_5": 0.7710307816929933,
|
| 266 |
+
"mrr_at_10": 0.7773322532739296,
|
| 267 |
+
"mrr_at_20": 0.7790656715075932,
|
| 268 |
+
"mrr_at_50": 0.7797137179788176,
|
| 269 |
+
"mrr_at_100": 0.7798294471430899,
|
| 270 |
+
"naucs_at_1_max": 0.19289339347399329,
|
| 271 |
+
"naucs_at_1_std": -0.05373436574034402,
|
| 272 |
+
"naucs_at_1_diff1": 0.8118815353915732,
|
| 273 |
+
"naucs_at_3_max": 0.24444248974914928,
|
| 274 |
+
"naucs_at_3_std": 0.012951438245694854,
|
| 275 |
+
"naucs_at_3_diff1": 0.7252009696977523,
|
| 276 |
+
"naucs_at_5_max": 0.27477480629269946,
|
| 277 |
+
"naucs_at_5_std": 0.10687833140288663,
|
| 278 |
+
"naucs_at_5_diff1": 0.7019146338300569,
|
| 279 |
+
"naucs_at_10_max": 0.23474834180340118,
|
| 280 |
+
"naucs_at_10_std": 0.13375117651376378,
|
| 281 |
+
"naucs_at_10_diff1": 0.6766342016471449,
|
| 282 |
+
"naucs_at_20_max": 0.3762582961131715,
|
| 283 |
+
"naucs_at_20_std": 0.29216428469292166,
|
| 284 |
+
"naucs_at_20_diff1": 0.6564671335087516,
|
| 285 |
+
"naucs_at_50_max": 0.4691053847445,
|
| 286 |
+
"naucs_at_50_std": 0.4359718488363951,
|
| 287 |
+
"naucs_at_50_diff1": 0.7152604718494652,
|
| 288 |
+
"naucs_at_100_max": 0.5259975902909616,
|
| 289 |
+
"naucs_at_100_std": 0.651086653120611,
|
| 290 |
+
"naucs_at_100_diff1": 0.7663843453532901
|
| 291 |
+
},
|
| 292 |
+
"shiftproject_test": {
|
| 293 |
+
"ndcg_at_1": 0.85,
|
| 294 |
+
"ndcg_at_3": 0.91917,
|
| 295 |
+
"ndcg_at_5": 0.92347,
|
| 296 |
+
"ndcg_at_10": 0.92949,
|
| 297 |
+
"ndcg_at_20": 0.92949,
|
| 298 |
+
"ndcg_at_50": 0.92949,
|
| 299 |
+
"ndcg_at_100": 0.92949,
|
| 300 |
+
"map_at_1": 0.85,
|
| 301 |
+
"map_at_3": 0.90167,
|
| 302 |
+
"map_at_5": 0.90417,
|
| 303 |
+
"map_at_10": 0.90639,
|
| 304 |
+
"map_at_20": 0.90639,
|
| 305 |
+
"map_at_50": 0.90639,
|
| 306 |
+
"map_at_100": 0.90639,
|
| 307 |
+
"recall_at_1": 0.85,
|
| 308 |
+
"recall_at_3": 0.97,
|
| 309 |
+
"recall_at_5": 0.98,
|
| 310 |
+
"recall_at_10": 1.0,
|
| 311 |
+
"recall_at_20": 1.0,
|
| 312 |
+
"recall_at_50": 1.0,
|
| 313 |
+
"recall_at_100": 1.0,
|
| 314 |
+
"precision_at_1": 0.85,
|
| 315 |
+
"precision_at_3": 0.32333,
|
| 316 |
+
"precision_at_5": 0.196,
|
| 317 |
+
"precision_at_10": 0.1,
|
| 318 |
+
"precision_at_20": 0.05,
|
| 319 |
+
"precision_at_50": 0.02,
|
| 320 |
+
"precision_at_100": 0.01,
|
| 321 |
+
"mrr_at_1": 0.85,
|
| 322 |
+
"mrr_at_3": 0.9016666666666666,
|
| 323 |
+
"mrr_at_5": 0.9041666666666666,
|
| 324 |
+
"mrr_at_10": 0.9063888888888889,
|
| 325 |
+
"mrr_at_20": 0.9063888888888889,
|
| 326 |
+
"mrr_at_50": 0.9063888888888889,
|
| 327 |
+
"mrr_at_100": 0.9063888888888889,
|
| 328 |
+
"naucs_at_1_max": 0.029189716889034732,
|
| 329 |
+
"naucs_at_1_std": -0.37507321835340074,
|
| 330 |
+
"naucs_at_1_diff1": 0.7931012040351454,
|
| 331 |
+
"naucs_at_3_max": 0.5589791472144446,
|
| 332 |
+
"naucs_at_3_std": 0.09056956115779448,
|
| 333 |
+
"naucs_at_3_diff1": 0.9564270152505466,
|
| 334 |
+
"naucs_at_5_max": 0.3384687208216692,
|
| 335 |
+
"naucs_at_5_std": -0.2987861811391239,
|
| 336 |
+
"naucs_at_5_diff1": 1.0,
|
| 337 |
+
"naucs_at_10_max": 1.0,
|
| 338 |
+
"naucs_at_10_std": 1.0,
|
| 339 |
+
"naucs_at_10_diff1": 1.0,
|
| 340 |
+
"naucs_at_20_max": 1.0,
|
| 341 |
+
"naucs_at_20_std": 1.0,
|
| 342 |
+
"naucs_at_20_diff1": 1.0,
|
| 343 |
+
"naucs_at_50_max": null,
|
| 344 |
+
"naucs_at_50_std": null,
|
| 345 |
+
"naucs_at_50_diff1": null,
|
| 346 |
+
"naucs_at_100_max": null,
|
| 347 |
+
"naucs_at_100_std": null,
|
| 348 |
+
"naucs_at_100_diff1": null
|
| 349 |
+
},
|
| 350 |
+
"syntheticDocQA_artificial_intelligence_test": {
|
| 351 |
+
"ndcg_at_1": 0.98,
|
| 352 |
+
"ndcg_at_3": 0.99262,
|
| 353 |
+
"ndcg_at_5": 0.99262,
|
| 354 |
+
"ndcg_at_10": 0.99262,
|
| 355 |
+
"ndcg_at_20": 0.99262,
|
| 356 |
+
"ndcg_at_50": 0.99262,
|
| 357 |
+
"ndcg_at_100": 0.99262,
|
| 358 |
+
"map_at_1": 0.98,
|
| 359 |
+
"map_at_3": 0.99,
|
| 360 |
+
"map_at_5": 0.99,
|
| 361 |
+
"map_at_10": 0.99,
|
| 362 |
+
"map_at_20": 0.99,
|
| 363 |
+
"map_at_50": 0.99,
|
| 364 |
+
"map_at_100": 0.99,
|
| 365 |
+
"recall_at_1": 0.98,
|
| 366 |
+
"recall_at_3": 1.0,
|
| 367 |
+
"recall_at_5": 1.0,
|
| 368 |
+
"recall_at_10": 1.0,
|
| 369 |
+
"recall_at_20": 1.0,
|
| 370 |
+
"recall_at_50": 1.0,
|
| 371 |
+
"recall_at_100": 1.0,
|
| 372 |
+
"precision_at_1": 0.98,
|
| 373 |
+
"precision_at_3": 0.33333,
|
| 374 |
+
"precision_at_5": 0.2,
|
| 375 |
+
"precision_at_10": 0.1,
|
| 376 |
+
"precision_at_20": 0.05,
|
| 377 |
+
"precision_at_50": 0.02,
|
| 378 |
+
"precision_at_100": 0.01,
|
| 379 |
+
"mrr_at_1": 0.98,
|
| 380 |
+
"mrr_at_3": 0.99,
|
| 381 |
+
"mrr_at_5": 0.99,
|
| 382 |
+
"mrr_at_10": 0.99,
|
| 383 |
+
"mrr_at_20": 0.99,
|
| 384 |
+
"mrr_at_50": 0.99,
|
| 385 |
+
"mrr_at_100": 0.99,
|
| 386 |
+
"naucs_at_1_max": 0.540149393090569,
|
| 387 |
+
"naucs_at_1_std": 0.3384687208216605,
|
| 388 |
+
"naucs_at_1_diff1": 0.9346405228758133,
|
| 389 |
+
"naucs_at_3_max": 1.0,
|
| 390 |
+
"naucs_at_3_std": 1.0,
|
| 391 |
+
"naucs_at_3_diff1": 1.0,
|
| 392 |
+
"naucs_at_5_max": 1.0,
|
| 393 |
+
"naucs_at_5_std": 1.0,
|
| 394 |
+
"naucs_at_5_diff1": 1.0,
|
| 395 |
+
"naucs_at_10_max": 1.0,
|
| 396 |
+
"naucs_at_10_std": 1.0,
|
| 397 |
+
"naucs_at_10_diff1": 1.0,
|
| 398 |
+
"naucs_at_20_max": 1.0,
|
| 399 |
+
"naucs_at_20_std": 1.0,
|
| 400 |
+
"naucs_at_20_diff1": 1.0,
|
| 401 |
+
"naucs_at_50_max": null,
|
| 402 |
+
"naucs_at_50_std": null,
|
| 403 |
+
"naucs_at_50_diff1": null,
|
| 404 |
+
"naucs_at_100_max": null,
|
| 405 |
+
"naucs_at_100_std": null,
|
| 406 |
+
"naucs_at_100_diff1": null
|
| 407 |
+
},
|
| 408 |
+
"syntheticDocQA_energy_test": {
|
| 409 |
+
"ndcg_at_1": 0.95,
|
| 410 |
+
"ndcg_at_3": 0.96762,
|
| 411 |
+
"ndcg_at_5": 0.96762,
|
| 412 |
+
"ndcg_at_10": 0.97118,
|
| 413 |
+
"ndcg_at_20": 0.97118,
|
| 414 |
+
"ndcg_at_50": 0.973,
|
| 415 |
+
"ndcg_at_100": 0.973,
|
| 416 |
+
"map_at_1": 0.95,
|
| 417 |
+
"map_at_3": 0.96333,
|
| 418 |
+
"map_at_5": 0.96333,
|
| 419 |
+
"map_at_10": 0.965,
|
| 420 |
+
"map_at_20": 0.965,
|
| 421 |
+
"map_at_50": 0.96523,
|
| 422 |
+
"map_at_100": 0.96523,
|
| 423 |
+
"recall_at_1": 0.95,
|
| 424 |
+
"recall_at_3": 0.98,
|
| 425 |
+
"recall_at_5": 0.98,
|
| 426 |
+
"recall_at_10": 0.99,
|
| 427 |
+
"recall_at_20": 0.99,
|
| 428 |
+
"recall_at_50": 1.0,
|
| 429 |
+
"recall_at_100": 1.0,
|
| 430 |
+
"precision_at_1": 0.95,
|
| 431 |
+
"precision_at_3": 0.32667,
|
| 432 |
+
"precision_at_5": 0.196,
|
| 433 |
+
"precision_at_10": 0.099,
|
| 434 |
+
"precision_at_20": 0.0495,
|
| 435 |
+
"precision_at_50": 0.02,
|
| 436 |
+
"precision_at_100": 0.01,
|
| 437 |
+
"mrr_at_1": 0.95,
|
| 438 |
+
"mrr_at_3": 0.9633333333333333,
|
| 439 |
+
"mrr_at_5": 0.9633333333333333,
|
| 440 |
+
"mrr_at_10": 0.965,
|
| 441 |
+
"mrr_at_20": 0.965,
|
| 442 |
+
"mrr_at_50": 0.9652272727272727,
|
| 443 |
+
"mrr_at_100": 0.9652272727272727,
|
| 444 |
+
"naucs_at_1_max": 0.42726423902894384,
|
| 445 |
+
"naucs_at_1_std": -0.4889822595704953,
|
| 446 |
+
"naucs_at_1_diff1": 1.0,
|
| 447 |
+
"naucs_at_3_max": 0.6136788048552655,
|
| 448 |
+
"naucs_at_3_std": -0.6909430438842241,
|
| 449 |
+
"naucs_at_3_diff1": 1.0,
|
| 450 |
+
"naucs_at_5_max": 0.6136788048552745,
|
| 451 |
+
"naucs_at_5_std": -0.690943043884218,
|
| 452 |
+
"naucs_at_5_diff1": 1.0,
|
| 453 |
+
"naucs_at_10_max": 0.8692810457516413,
|
| 454 |
+
"naucs_at_10_std": 0.35807656395891135,
|
| 455 |
+
"naucs_at_10_diff1": 1.0,
|
| 456 |
+
"naucs_at_20_max": 0.8692810457516413,
|
| 457 |
+
"naucs_at_20_std": 0.35807656395891135,
|
| 458 |
+
"naucs_at_20_diff1": 1.0,
|
| 459 |
+
"naucs_at_50_max": null,
|
| 460 |
+
"naucs_at_50_std": null,
|
| 461 |
+
"naucs_at_50_diff1": null,
|
| 462 |
+
"naucs_at_100_max": null,
|
| 463 |
+
"naucs_at_100_std": null,
|
| 464 |
+
"naucs_at_100_diff1": null
|
| 465 |
+
},
|
| 466 |
+
"syntheticDocQA_government_reports_test": {
|
| 467 |
+
"ndcg_at_1": 0.93,
|
| 468 |
+
"ndcg_at_3": 0.96524,
|
| 469 |
+
"ndcg_at_5": 0.96954,
|
| 470 |
+
"ndcg_at_10": 0.96954,
|
| 471 |
+
"ndcg_at_20": 0.96954,
|
| 472 |
+
"ndcg_at_50": 0.96954,
|
| 473 |
+
"ndcg_at_100": 0.96954,
|
| 474 |
+
"map_at_1": 0.93,
|
| 475 |
+
"map_at_3": 0.95667,
|
| 476 |
+
"map_at_5": 0.95917,
|
| 477 |
+
"map_at_10": 0.95917,
|
| 478 |
+
"map_at_20": 0.95917,
|
| 479 |
+
"map_at_50": 0.95917,
|
| 480 |
+
"map_at_100": 0.95917,
|
| 481 |
+
"recall_at_1": 0.93,
|
| 482 |
+
"recall_at_3": 0.99,
|
| 483 |
+
"recall_at_5": 1.0,
|
| 484 |
+
"recall_at_10": 1.0,
|
| 485 |
+
"recall_at_20": 1.0,
|
| 486 |
+
"recall_at_50": 1.0,
|
| 487 |
+
"recall_at_100": 1.0,
|
| 488 |
+
"precision_at_1": 0.93,
|
| 489 |
+
"precision_at_3": 0.33,
|
| 490 |
+
"precision_at_5": 0.2,
|
| 491 |
+
"precision_at_10": 0.1,
|
| 492 |
+
"precision_at_20": 0.05,
|
| 493 |
+
"precision_at_50": 0.02,
|
| 494 |
+
"precision_at_100": 0.01,
|
| 495 |
+
"mrr_at_1": 0.93,
|
| 496 |
+
"mrr_at_3": 0.9566666666666667,
|
| 497 |
+
"mrr_at_5": 0.9591666666666667,
|
| 498 |
+
"mrr_at_10": 0.9591666666666667,
|
| 499 |
+
"mrr_at_20": 0.9591666666666667,
|
| 500 |
+
"mrr_at_50": 0.9591666666666667,
|
| 501 |
+
"mrr_at_100": 0.9591666666666667,
|
| 502 |
+
"naucs_at_1_max": 0.6809390422835813,
|
| 503 |
+
"naucs_at_1_std": 0.5458850206749362,
|
| 504 |
+
"naucs_at_1_diff1": 0.9229691876750709,
|
| 505 |
+
"naucs_at_3_max": 1.0,
|
| 506 |
+
"naucs_at_3_std": 1.0,
|
| 507 |
+
"naucs_at_3_diff1": 1.0,
|
| 508 |
+
"naucs_at_5_max": 1.0,
|
| 509 |
+
"naucs_at_5_std": 1.0,
|
| 510 |
+
"naucs_at_5_diff1": 1.0,
|
| 511 |
+
"naucs_at_10_max": 1.0,
|
| 512 |
+
"naucs_at_10_std": 1.0,
|
| 513 |
+
"naucs_at_10_diff1": 1.0,
|
| 514 |
+
"naucs_at_20_max": 1.0,
|
| 515 |
+
"naucs_at_20_std": 1.0,
|
| 516 |
+
"naucs_at_20_diff1": 1.0,
|
| 517 |
+
"naucs_at_50_max": null,
|
| 518 |
+
"naucs_at_50_std": null,
|
| 519 |
+
"naucs_at_50_diff1": null,
|
| 520 |
+
"naucs_at_100_max": null,
|
| 521 |
+
"naucs_at_100_std": null,
|
| 522 |
+
"naucs_at_100_diff1": null
|
| 523 |
+
},
|
| 524 |
+
"syntheticDocQA_healthcare_industry_test": {
|
| 525 |
+
"ndcg_at_1": 0.96,
|
| 526 |
+
"ndcg_at_3": 0.98393,
|
| 527 |
+
"ndcg_at_5": 0.98393,
|
| 528 |
+
"ndcg_at_10": 0.98393,
|
| 529 |
+
"ndcg_at_20": 0.98393,
|
| 530 |
+
"ndcg_at_50": 0.98393,
|
| 531 |
+
"ndcg_at_100": 0.98393,
|
| 532 |
+
"map_at_1": 0.96,
|
| 533 |
+
"map_at_3": 0.97833,
|
| 534 |
+
"map_at_5": 0.97833,
|
| 535 |
+
"map_at_10": 0.97833,
|
| 536 |
+
"map_at_20": 0.97833,
|
| 537 |
+
"map_at_50": 0.97833,
|
| 538 |
+
"map_at_100": 0.97833,
|
| 539 |
+
"recall_at_1": 0.96,
|
| 540 |
+
"recall_at_3": 1.0,
|
| 541 |
+
"recall_at_5": 1.0,
|
| 542 |
+
"recall_at_10": 1.0,
|
| 543 |
+
"recall_at_20": 1.0,
|
| 544 |
+
"recall_at_50": 1.0,
|
| 545 |
+
"recall_at_100": 1.0,
|
| 546 |
+
"precision_at_1": 0.96,
|
| 547 |
+
"precision_at_3": 0.33333,
|
| 548 |
+
"precision_at_5": 0.2,
|
| 549 |
+
"precision_at_10": 0.1,
|
| 550 |
+
"precision_at_20": 0.05,
|
| 551 |
+
"precision_at_50": 0.02,
|
| 552 |
+
"precision_at_100": 0.01,
|
| 553 |
+
"mrr_at_1": 0.96,
|
| 554 |
+
"mrr_at_3": 0.9783333333333333,
|
| 555 |
+
"mrr_at_5": 0.9783333333333333,
|
| 556 |
+
"mrr_at_10": 0.9783333333333333,
|
| 557 |
+
"mrr_at_20": 0.9783333333333333,
|
| 558 |
+
"mrr_at_50": 0.9783333333333333,
|
| 559 |
+
"mrr_at_100": 0.9783333333333333,
|
| 560 |
+
"naucs_at_1_max": 0.7047152194211012,
|
| 561 |
+
"naucs_at_1_std": 0.32037815126050734,
|
| 562 |
+
"naucs_at_1_diff1": 1.0,
|
| 563 |
+
"naucs_at_3_max": 1.0,
|
| 564 |
+
"naucs_at_3_std": 1.0,
|
| 565 |
+
"naucs_at_3_diff1": 1.0,
|
| 566 |
+
"naucs_at_5_max": 1.0,
|
| 567 |
+
"naucs_at_5_std": 1.0,
|
| 568 |
+
"naucs_at_5_diff1": 1.0,
|
| 569 |
+
"naucs_at_10_max": 1.0,
|
| 570 |
+
"naucs_at_10_std": 1.0,
|
| 571 |
+
"naucs_at_10_diff1": 1.0,
|
| 572 |
+
"naucs_at_20_max": 1.0,
|
| 573 |
+
"naucs_at_20_std": 1.0,
|
| 574 |
+
"naucs_at_20_diff1": 1.0,
|
| 575 |
+
"naucs_at_50_max": null,
|
| 576 |
+
"naucs_at_50_std": null,
|
| 577 |
+
"naucs_at_50_diff1": null,
|
| 578 |
+
"naucs_at_100_max": null,
|
| 579 |
+
"naucs_at_100_std": null,
|
| 580 |
+
"naucs_at_100_diff1": null
|
| 581 |
+
}
|
| 582 |
+
}
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>",
|
| 5 |
+
"<|object_ref_start|>",
|
| 6 |
+
"<|object_ref_end|>",
|
| 7 |
+
"<|box_start|>",
|
| 8 |
+
"<|box_end|>",
|
| 9 |
+
"<|quad_start|>",
|
| 10 |
+
"<|quad_end|>",
|
| 11 |
+
"<|vision_start|>",
|
| 12 |
+
"<|vision_end|>",
|
| 13 |
+
"<|vision_pad|>",
|
| 14 |
+
"<|image_pad|>",
|
| 15 |
+
"<|video_pad|>"
|
| 16 |
+
],
|
| 17 |
+
"eos_token": {
|
| 18 |
+
"content": "<|im_end|>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": false,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
},
|
| 24 |
+
"pad_token": {
|
| 25 |
+
"content": "<|endoftext|>",
|
| 26 |
+
"lstrip": false,
|
| 27 |
+
"normalized": false,
|
| 28 |
+
"rstrip": false,
|
| 29 |
+
"single_word": false
|
| 30 |
+
}
|
| 31 |
+
}
|
tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
|
| 3 |
+
size 11421896
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,209 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_prefix_space": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"151643": {
|
| 6 |
+
"content": "<|endoftext|>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"151644": {
|
| 14 |
+
"content": "<|im_start|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"151645": {
|
| 22 |
+
"content": "<|im_end|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
},
|
| 29 |
+
"151646": {
|
| 30 |
+
"content": "<|object_ref_start|>",
|
| 31 |
+
"lstrip": false,
|
| 32 |
+
"normalized": false,
|
| 33 |
+
"rstrip": false,
|
| 34 |
+
"single_word": false,
|
| 35 |
+
"special": true
|
| 36 |
+
},
|
| 37 |
+
"151647": {
|
| 38 |
+
"content": "<|object_ref_end|>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false,
|
| 43 |
+
"special": true
|
| 44 |
+
},
|
| 45 |
+
"151648": {
|
| 46 |
+
"content": "<|box_start|>",
|
| 47 |
+
"lstrip": false,
|
| 48 |
+
"normalized": false,
|
| 49 |
+
"rstrip": false,
|
| 50 |
+
"single_word": false,
|
| 51 |
+
"special": true
|
| 52 |
+
},
|
| 53 |
+
"151649": {
|
| 54 |
+
"content": "<|box_end|>",
|
| 55 |
+
"lstrip": false,
|
| 56 |
+
"normalized": false,
|
| 57 |
+
"rstrip": false,
|
| 58 |
+
"single_word": false,
|
| 59 |
+
"special": true
|
| 60 |
+
},
|
| 61 |
+
"151650": {
|
| 62 |
+
"content": "<|quad_start|>",
|
| 63 |
+
"lstrip": false,
|
| 64 |
+
"normalized": false,
|
| 65 |
+
"rstrip": false,
|
| 66 |
+
"single_word": false,
|
| 67 |
+
"special": true
|
| 68 |
+
},
|
| 69 |
+
"151651": {
|
| 70 |
+
"content": "<|quad_end|>",
|
| 71 |
+
"lstrip": false,
|
| 72 |
+
"normalized": false,
|
| 73 |
+
"rstrip": false,
|
| 74 |
+
"single_word": false,
|
| 75 |
+
"special": true
|
| 76 |
+
},
|
| 77 |
+
"151652": {
|
| 78 |
+
"content": "<|vision_start|>",
|
| 79 |
+
"lstrip": false,
|
| 80 |
+
"normalized": false,
|
| 81 |
+
"rstrip": false,
|
| 82 |
+
"single_word": false,
|
| 83 |
+
"special": true
|
| 84 |
+
},
|
| 85 |
+
"151653": {
|
| 86 |
+
"content": "<|vision_end|>",
|
| 87 |
+
"lstrip": false,
|
| 88 |
+
"normalized": false,
|
| 89 |
+
"rstrip": false,
|
| 90 |
+
"single_word": false,
|
| 91 |
+
"special": true
|
| 92 |
+
},
|
| 93 |
+
"151654": {
|
| 94 |
+
"content": "<|vision_pad|>",
|
| 95 |
+
"lstrip": false,
|
| 96 |
+
"normalized": false,
|
| 97 |
+
"rstrip": false,
|
| 98 |
+
"single_word": false,
|
| 99 |
+
"special": true
|
| 100 |
+
},
|
| 101 |
+
"151655": {
|
| 102 |
+
"content": "<|image_pad|>",
|
| 103 |
+
"lstrip": false,
|
| 104 |
+
"normalized": false,
|
| 105 |
+
"rstrip": false,
|
| 106 |
+
"single_word": false,
|
| 107 |
+
"special": true
|
| 108 |
+
},
|
| 109 |
+
"151656": {
|
| 110 |
+
"content": "<|video_pad|>",
|
| 111 |
+
"lstrip": false,
|
| 112 |
+
"normalized": false,
|
| 113 |
+
"rstrip": false,
|
| 114 |
+
"single_word": false,
|
| 115 |
+
"special": true
|
| 116 |
+
},
|
| 117 |
+
"151657": {
|
| 118 |
+
"content": "<tool_call>",
|
| 119 |
+
"lstrip": false,
|
| 120 |
+
"normalized": false,
|
| 121 |
+
"rstrip": false,
|
| 122 |
+
"single_word": false,
|
| 123 |
+
"special": false
|
| 124 |
+
},
|
| 125 |
+
"151658": {
|
| 126 |
+
"content": "</tool_call>",
|
| 127 |
+
"lstrip": false,
|
| 128 |
+
"normalized": false,
|
| 129 |
+
"rstrip": false,
|
| 130 |
+
"single_word": false,
|
| 131 |
+
"special": false
|
| 132 |
+
},
|
| 133 |
+
"151659": {
|
| 134 |
+
"content": "<|fim_prefix|>",
|
| 135 |
+
"lstrip": false,
|
| 136 |
+
"normalized": false,
|
| 137 |
+
"rstrip": false,
|
| 138 |
+
"single_word": false,
|
| 139 |
+
"special": false
|
| 140 |
+
},
|
| 141 |
+
"151660": {
|
| 142 |
+
"content": "<|fim_middle|>",
|
| 143 |
+
"lstrip": false,
|
| 144 |
+
"normalized": false,
|
| 145 |
+
"rstrip": false,
|
| 146 |
+
"single_word": false,
|
| 147 |
+
"special": false
|
| 148 |
+
},
|
| 149 |
+
"151661": {
|
| 150 |
+
"content": "<|fim_suffix|>",
|
| 151 |
+
"lstrip": false,
|
| 152 |
+
"normalized": false,
|
| 153 |
+
"rstrip": false,
|
| 154 |
+
"single_word": false,
|
| 155 |
+
"special": false
|
| 156 |
+
},
|
| 157 |
+
"151662": {
|
| 158 |
+
"content": "<|fim_pad|>",
|
| 159 |
+
"lstrip": false,
|
| 160 |
+
"normalized": false,
|
| 161 |
+
"rstrip": false,
|
| 162 |
+
"single_word": false,
|
| 163 |
+
"special": false
|
| 164 |
+
},
|
| 165 |
+
"151663": {
|
| 166 |
+
"content": "<|repo_name|>",
|
| 167 |
+
"lstrip": false,
|
| 168 |
+
"normalized": false,
|
| 169 |
+
"rstrip": false,
|
| 170 |
+
"single_word": false,
|
| 171 |
+
"special": false
|
| 172 |
+
},
|
| 173 |
+
"151664": {
|
| 174 |
+
"content": "<|file_sep|>",
|
| 175 |
+
"lstrip": false,
|
| 176 |
+
"normalized": false,
|
| 177 |
+
"rstrip": false,
|
| 178 |
+
"single_word": false,
|
| 179 |
+
"special": false
|
| 180 |
+
}
|
| 181 |
+
},
|
| 182 |
+
"additional_special_tokens": [
|
| 183 |
+
"<|im_start|>",
|
| 184 |
+
"<|im_end|>",
|
| 185 |
+
"<|object_ref_start|>",
|
| 186 |
+
"<|object_ref_end|>",
|
| 187 |
+
"<|box_start|>",
|
| 188 |
+
"<|box_end|>",
|
| 189 |
+
"<|quad_start|>",
|
| 190 |
+
"<|quad_end|>",
|
| 191 |
+
"<|vision_start|>",
|
| 192 |
+
"<|vision_end|>",
|
| 193 |
+
"<|vision_pad|>",
|
| 194 |
+
"<|image_pad|>",
|
| 195 |
+
"<|video_pad|>"
|
| 196 |
+
],
|
| 197 |
+
"bos_token": null,
|
| 198 |
+
"chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0]['role'] == 'system' %}\n {{- messages[0]['content'] }}\n {%- else %}\n {{- 'You are a helpful assistant.' }}\n {%- endif %}\n {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n {%- if messages[0]['role'] == 'system' %}\n {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n {%- else %}\n {{- '<|im_start|>system\\nYou are a helpful assistant.<|im_end|>\\n' }}\n {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {{- '<|im_start|>' + message.role }}\n {%- if message.content %}\n {{- '\\n' + message.content }}\n {%- endif %}\n {%- for tool_call in message.tool_calls %}\n {%- if tool_call.function is defined %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {{- '\\n<tool_call>\\n{\"name\": \"' }}\n {{- tool_call.name }}\n {{- '\", \"arguments\": ' }}\n {{- tool_call.arguments | tojson }}\n {{- '}\\n</tool_call>' }}\n {%- endfor %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- message.content }}\n {{- '\\n</tool_response>' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
|
| 199 |
+
"clean_up_tokenization_spaces": false,
|
| 200 |
+
"eos_token": "<|im_end|>",
|
| 201 |
+
"errors": "replace",
|
| 202 |
+
"extra_special_tokens": {},
|
| 203 |
+
"model_max_length": 131072,
|
| 204 |
+
"pad_token": "<|endoftext|>",
|
| 205 |
+
"processor_class": "JinaEmbeddingsV4Processor",
|
| 206 |
+
"split_special_tokens": false,
|
| 207 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 208 |
+
"unk_token": null
|
| 209 |
+
}
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|