|
|
--- |
|
|
language: |
|
|
- en |
|
|
tags: |
|
|
- sentence-transformers |
|
|
- cross-encoder |
|
|
- reranker |
|
|
- generated_from_trainer |
|
|
- dataset_size:9960000 |
|
|
- loss:BinaryCrossEntropyLoss |
|
|
base_model: jhu-clsp/ettin-encoder-32m |
|
|
datasets: |
|
|
- sentence-transformers/msmarco |
|
|
pipeline_tag: text-ranking |
|
|
library_name: sentence-transformers |
|
|
metrics: |
|
|
- map |
|
|
- mrr@10 |
|
|
- ndcg@10 |
|
|
model-index: |
|
|
- name: CrossEncoder based on jhu-clsp/ettin-encoder-32m |
|
|
results: |
|
|
- task: |
|
|
type: cross-encoder-reranking |
|
|
name: Cross Encoder Reranking |
|
|
dataset: |
|
|
name: NanoMSMARCO R100 |
|
|
type: NanoMSMARCO_R100 |
|
|
metrics: |
|
|
- type: map |
|
|
value: 0.6052 |
|
|
name: Map |
|
|
- type: mrr@10 |
|
|
value: 0.5998 |
|
|
name: Mrr@10 |
|
|
- type: ndcg@10 |
|
|
value: 0.6717 |
|
|
name: Ndcg@10 |
|
|
- task: |
|
|
type: cross-encoder-reranking |
|
|
name: Cross Encoder Reranking |
|
|
dataset: |
|
|
name: NanoNFCorpus R100 |
|
|
type: NanoNFCorpus_R100 |
|
|
metrics: |
|
|
- type: map |
|
|
value: 0.3112 |
|
|
name: Map |
|
|
- type: mrr@10 |
|
|
value: 0.5052 |
|
|
name: Mrr@10 |
|
|
- type: ndcg@10 |
|
|
value: 0.326 |
|
|
name: Ndcg@10 |
|
|
- task: |
|
|
type: cross-encoder-reranking |
|
|
name: Cross Encoder Reranking |
|
|
dataset: |
|
|
name: NanoNQ R100 |
|
|
type: NanoNQ_R100 |
|
|
metrics: |
|
|
- type: map |
|
|
value: 0.6232 |
|
|
name: Map |
|
|
- type: mrr@10 |
|
|
value: 0.646 |
|
|
name: Mrr@10 |
|
|
- type: ndcg@10 |
|
|
value: 0.6792 |
|
|
name: Ndcg@10 |
|
|
- task: |
|
|
type: cross-encoder-nano-beir |
|
|
name: Cross Encoder Nano BEIR |
|
|
dataset: |
|
|
name: NanoBEIR R100 mean |
|
|
type: NanoBEIR_R100_mean |
|
|
metrics: |
|
|
- type: map |
|
|
value: 0.5132 |
|
|
name: Map |
|
|
- type: mrr@10 |
|
|
value: 0.5837 |
|
|
name: Mrr@10 |
|
|
- type: ndcg@10 |
|
|
value: 0.559 |
|
|
name: Ndcg@10 |
|
|
--- |
|
|
|
|
|
# CrossEncoder based on jhu-clsp/ettin-encoder-32m |
|
|
|
|
|
This is a [Cross Encoder](https://www.sbert.net/docs/cross_encoder/usage/usage.html) model finetuned from [jhu-clsp/ettin-encoder-32m](https://huggingface.co/jhu-clsp/ettin-encoder-32m) on the [msmarco](https://huggingface.co/datasets/sentence-transformers/msmarco) dataset using the [sentence-transformers](https://www.SBERT.net) library. It computes scores for pairs of texts, which can be used for text reranking and semantic search. |
|
|
|
|
|
## Model Details |
|
|
|
|
|
### Model Description |
|
|
- **Model Type:** Cross Encoder |
|
|
- **Base model:** [jhu-clsp/ettin-encoder-32m](https://huggingface.co/jhu-clsp/ettin-encoder-32m) <!-- at revision 1b8ba06455dd44f80fc9c1ca9e22806157a57379 --> |
|
|
- **Maximum Sequence Length:** 7999 tokens |
|
|
- **Number of Output Labels:** 1 label |
|
|
- **Training Dataset:** |
|
|
- [msmarco](https://huggingface.co/datasets/sentence-transformers/msmarco) |
|
|
- **Language:** en |
|
|
<!-- - **License:** Unknown --> |
|
|
|
|
|
### Model Sources |
|
|
|
|
|
- **Documentation:** [Sentence Transformers Documentation](https://sbert.net) |
|
|
- **Documentation:** [Cross Encoder Documentation](https://www.sbert.net/docs/cross_encoder/usage/usage.html) |
|
|
- **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) |
|
|
- **Hugging Face:** [Cross Encoders on Hugging Face](https://huggingface.co/models?library=sentence-transformers&other=cross-encoder) |
|
|
|
|
|
## Usage |
|
|
|
|
|
### Direct Usage (Sentence Transformers) |
|
|
|
|
|
First install the Sentence Transformers library: |
|
|
|
|
|
```bash |
|
|
pip install -U sentence-transformers |
|
|
``` |
|
|
|
|
|
Then you can load this model and run inference. |
|
|
```python |
|
|
from sentence_transformers import CrossEncoder |
|
|
|
|
|
# Download from the 🤗 Hub |
|
|
model = CrossEncoder("rahulseetharaman/reranker-ettin-encoder-32m-msmarco-bce-10m") |
|
|
# Get scores for pairs of texts |
|
|
pairs = [ |
|
|
['select committees definition government', 'There are four types of congressional committees: 1 Standing committees, which continue from one Congress to the next, are probably the most important type because they consider and shape the vast majority of proposed laws. 2 Select committees are temporarily formed for specific purposes, often to study a particular issue.'], |
|
|
['what is a perceptual map', 'Welcome to our New Castle, Pennsylvania street map page. The street map of New Castle PA that is located below is provided by Google Maps. You can grab the New Castle Pennsylvania street map and move it around to re-centre the map. You can change between standard map view, satellite map view and hybrid map view.'], |
|
|
['what makes your skin feel cold and burn', 'When the wind blows in cold weather, you feel colder than the actual temperature because the air blows away heat from your skin faster. For instance, if the temperature is -17.8 Celsius (0 Fahrenheit) and the wind blows at 15 mph, it feels like -28.3 Celsius (-19 Fahrenheit) -- exposed skin can freeze in 30 minutes.'], |
|
|
['average act score for university of georgia', 'A graph of UB, University at Buffalo GPA, SAT score, and ACT score admissions data for students who were accepted, rejected, and waitlisted. A graph of UB, University at Buffalo GPA, SAT score, and ACT score admissions data for students who were accepted, rejected, and waitlisted. University at Buffalo GPA, SAT and ACT Data Search the site GO'], |
|
|
['when was the ontario, ca, post office established', 'In 1832 Jed Jackson had the contract for carrying mail from Brantford to London twice a week along the Old Stage Road. On October 6, 1835, a post office was established at Woodstock, Ontario, with Princeton following within two years. According to the Legislative Council Sessional Papers for 1846, a post office was established at Princeton on May 6, 1836 and Jeremiah Cowin was appointed postmaster on May 9, 1837. The sureties were George Beamer and Silas Martin to the amount of £200. The assistant was John Charles.'], |
|
|
] |
|
|
scores = model.predict(pairs) |
|
|
print(scores.shape) |
|
|
# (5,) |
|
|
|
|
|
# Or rank different texts based on similarity to a single text |
|
|
ranks = model.rank( |
|
|
'select committees definition government', |
|
|
[ |
|
|
'There are four types of congressional committees: 1 Standing committees, which continue from one Congress to the next, are probably the most important type because they consider and shape the vast majority of proposed laws. 2 Select committees are temporarily formed for specific purposes, often to study a particular issue.', |
|
|
'Welcome to our New Castle, Pennsylvania street map page. The street map of New Castle PA that is located below is provided by Google Maps. You can grab the New Castle Pennsylvania street map and move it around to re-centre the map. You can change between standard map view, satellite map view and hybrid map view.', |
|
|
'When the wind blows in cold weather, you feel colder than the actual temperature because the air blows away heat from your skin faster. For instance, if the temperature is -17.8 Celsius (0 Fahrenheit) and the wind blows at 15 mph, it feels like -28.3 Celsius (-19 Fahrenheit) -- exposed skin can freeze in 30 minutes.', |
|
|
'A graph of UB, University at Buffalo GPA, SAT score, and ACT score admissions data for students who were accepted, rejected, and waitlisted. A graph of UB, University at Buffalo GPA, SAT score, and ACT score admissions data for students who were accepted, rejected, and waitlisted. University at Buffalo GPA, SAT and ACT Data Search the site GO', |
|
|
'In 1832 Jed Jackson had the contract for carrying mail from Brantford to London twice a week along the Old Stage Road. On October 6, 1835, a post office was established at Woodstock, Ontario, with Princeton following within two years. According to the Legislative Council Sessional Papers for 1846, a post office was established at Princeton on May 6, 1836 and Jeremiah Cowin was appointed postmaster on May 9, 1837. The sureties were George Beamer and Silas Martin to the amount of £200. The assistant was John Charles.', |
|
|
] |
|
|
) |
|
|
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...] |
|
|
``` |
|
|
|
|
|
<!-- |
|
|
### Direct Usage (Transformers) |
|
|
|
|
|
<details><summary>Click to see the direct usage in Transformers</summary> |
|
|
|
|
|
</details> |
|
|
--> |
|
|
|
|
|
<!-- |
|
|
### Downstream Usage (Sentence Transformers) |
|
|
|
|
|
You can finetune this model on your own dataset. |
|
|
|
|
|
<details><summary>Click to expand</summary> |
|
|
|
|
|
</details> |
|
|
--> |
|
|
|
|
|
<!-- |
|
|
### Out-of-Scope Use |
|
|
|
|
|
*List how the model may foreseeably be misused and address what users ought not to do with the model.* |
|
|
--> |
|
|
|
|
|
## Evaluation |
|
|
|
|
|
### Metrics |
|
|
|
|
|
#### Cross Encoder Reranking |
|
|
|
|
|
* Datasets: `NanoMSMARCO_R100`, `NanoNFCorpus_R100` and `NanoNQ_R100` |
|
|
* Evaluated with [<code>CrossEncoderRerankingEvaluator</code>](https://sbert.net/docs/package_reference/cross_encoder/evaluation.html#sentence_transformers.cross_encoder.evaluation.CrossEncoderRerankingEvaluator) with these parameters: |
|
|
```json |
|
|
{ |
|
|
"at_k": 10, |
|
|
"always_rerank_positives": true |
|
|
} |
|
|
``` |
|
|
|
|
|
| Metric | NanoMSMARCO_R100 | NanoNFCorpus_R100 | NanoNQ_R100 | |
|
|
|:------------|:---------------------|:---------------------|:---------------------| |
|
|
| map | 0.6052 (+0.1156) | 0.3112 (+0.0502) | 0.6232 (+0.2036) | |
|
|
| mrr@10 | 0.5998 (+0.1223) | 0.5052 (+0.0054) | 0.6460 (+0.2194) | |
|
|
| **ndcg@10** | **0.6717 (+0.1313)** | **0.3260 (+0.0010)** | **0.6792 (+0.1786)** | |
|
|
|
|
|
#### Cross Encoder Nano BEIR |
|
|
|
|
|
* Dataset: `NanoBEIR_R100_mean` |
|
|
* Evaluated with [<code>CrossEncoderNanoBEIREvaluator</code>](https://sbert.net/docs/package_reference/cross_encoder/evaluation.html#sentence_transformers.cross_encoder.evaluation.CrossEncoderNanoBEIREvaluator) with these parameters: |
|
|
```json |
|
|
{ |
|
|
"dataset_names": [ |
|
|
"msmarco", |
|
|
"nfcorpus", |
|
|
"nq" |
|
|
], |
|
|
"rerank_k": 100, |
|
|
"at_k": 10, |
|
|
"always_rerank_positives": true |
|
|
} |
|
|
``` |
|
|
|
|
|
| Metric | Value | |
|
|
|:------------|:---------------------| |
|
|
| map | 0.5132 (+0.1231) | |
|
|
| mrr@10 | 0.5837 (+0.1157) | |
|
|
| **ndcg@10** | **0.5590 (+0.1036)** | |
|
|
|
|
|
<!-- |
|
|
## Bias, Risks and Limitations |
|
|
|
|
|
*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.* |
|
|
--> |
|
|
|
|
|
<!-- |
|
|
### Recommendations |
|
|
|
|
|
*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.* |
|
|
--> |
|
|
|
|
|
## Training Details |
|
|
|
|
|
### Training Dataset |
|
|
|
|
|
#### msmarco |
|
|
|
|
|
* Dataset: [msmarco](https://huggingface.co/datasets/sentence-transformers/msmarco) at [9e329ed](https://huggingface.co/datasets/sentence-transformers/msmarco/tree/9e329ed2e649c9d37b0d91dd6b764ff6fe671d83) |
|
|
* Size: 9,960,000 training samples |
|
|
* Columns: <code>query</code>, <code>passage</code>, and <code>score</code> |
|
|
* Approximate statistics based on the first 1000 samples: |
|
|
| | query | passage | score | |
|
|
|:--------|:-----------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------|:---------------------------------------------------------------| |
|
|
| type | string | string | float | |
|
|
| details | <ul><li>min: 9 characters</li><li>mean: 33.93 characters</li><li>max: 110 characters</li></ul> | <ul><li>min: 80 characters</li><li>mean: 348.08 characters</li><li>max: 897 characters</li></ul> | <ul><li>min: 0.0</li><li>mean: 0.49</li><li>max: 1.0</li></ul> | |
|
|
* Samples: |
|
|
| query | passage | score | |
|
|
|:------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------------| |
|
|
| <code>nap in chinese</code> | <code>continued... Most children from about 21 to 36 months of age still need one nap a day, which may range from one to three and a half hours long.They typically go to bed between 7 and 9 p.m. and wake up between 6 and 8 a.m. 3-6 Years Old: 10 - 12 hours per day.ontinued... Most children from about 21 to 36 months of age still need one nap a day, which may range from one to three and a half hours long.</code> | <code>0.0</code> | |
|
|
| <code>what abdominal organ is most frequently injured as a result of blunt trauma?</code> | <code>Bochdalek Hernia. Bochdalek hernia is a congenital posterolateral diaphragmatic defect that is a result of failed closure of the pleuroperitoneal ducts -- a primitive communications between the pleural and abdominal cavities -- at 8 weeks' gestation.ochdalek Hernia. Bochdalek hernia is a congenital posterolateral diaphragmatic defect that is a result of failed closure of the pleuroperitoneal ducts -- a primitive communications between the pleural and abdominal cavities -- at 8 weeks' gestation.</code> | <code>0.0</code> | |
|
|
| <code>where is round rock tx</code> | <code>Driving distance from Dallas, TX to Fort Worth, TX The total driving distance from Dallas, TX to Fort Worth, TX is 33 miles or 53 kilometers. Your trip begins in Dallas, Texas. It ends in Fort Worth, Texas. If you are planning a road trip, you might also want to calculate the total driving time from Dallas, TX to Fort Worth, TX so you can see when you'll arrive at your destination. You can also calculate the cost of driving from Dallas, TX to Fort Worth, TX based on current local fuel prices and an estimate of your car's best gas mileage.</code> | <code>0.0</code> | |
|
|
* Loss: [<code>BinaryCrossEntropyLoss</code>](https://sbert.net/docs/package_reference/cross_encoder/losses.html#binarycrossentropyloss) with these parameters: |
|
|
```json |
|
|
{ |
|
|
"activation_fn": "torch.nn.modules.linear.Identity", |
|
|
"pos_weight": null |
|
|
} |
|
|
``` |
|
|
|
|
|
### Evaluation Dataset |
|
|
|
|
|
#### msmarco |
|
|
|
|
|
* Dataset: [msmarco](https://huggingface.co/datasets/sentence-transformers/msmarco) at [9e329ed](https://huggingface.co/datasets/sentence-transformers/msmarco/tree/9e329ed2e649c9d37b0d91dd6b764ff6fe671d83) |
|
|
* Size: 40,000 evaluation samples |
|
|
* Columns: <code>query</code>, <code>passage</code>, and <code>score</code> |
|
|
* Approximate statistics based on the first 1000 samples: |
|
|
| | query | passage | score | |
|
|
|:--------|:----------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------|:---------------------------------------------------------------| |
|
|
| type | string | string | float | |
|
|
| details | <ul><li>min: 11 characters</li><li>mean: 34.1 characters</li><li>max: 96 characters</li></ul> | <ul><li>min: 75 characters</li><li>mean: 341.31 characters</li><li>max: 938 characters</li></ul> | <ul><li>min: 0.0</li><li>mean: 0.49</li><li>max: 1.0</li></ul> | |
|
|
* Samples: |
|
|
| query | passage | score | |
|
|
|:-----------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------------| |
|
|
| <code>select committees definition government</code> | <code>There are four types of congressional committees: 1 Standing committees, which continue from one Congress to the next, are probably the most important type because they consider and shape the vast majority of proposed laws. 2 Select committees are temporarily formed for specific purposes, often to study a particular issue.</code> | <code>1.0</code> | |
|
|
| <code>what is a perceptual map</code> | <code>Welcome to our New Castle, Pennsylvania street map page. The street map of New Castle PA that is located below is provided by Google Maps. You can grab the New Castle Pennsylvania street map and move it around to re-centre the map. You can change between standard map view, satellite map view and hybrid map view.</code> | <code>0.0</code> | |
|
|
| <code>what makes your skin feel cold and burn</code> | <code>When the wind blows in cold weather, you feel colder than the actual temperature because the air blows away heat from your skin faster. For instance, if the temperature is -17.8 Celsius (0 Fahrenheit) and the wind blows at 15 mph, it feels like -28.3 Celsius (-19 Fahrenheit) -- exposed skin can freeze in 30 minutes.</code> | <code>0.0</code> | |
|
|
* Loss: [<code>BinaryCrossEntropyLoss</code>](https://sbert.net/docs/package_reference/cross_encoder/losses.html#binarycrossentropyloss) with these parameters: |
|
|
```json |
|
|
{ |
|
|
"activation_fn": "torch.nn.modules.linear.Identity", |
|
|
"pos_weight": null |
|
|
} |
|
|
``` |
|
|
|
|
|
### Training Hyperparameters |
|
|
#### Non-Default Hyperparameters |
|
|
|
|
|
- `eval_strategy`: steps |
|
|
- `per_device_train_batch_size`: 64 |
|
|
- `per_device_eval_batch_size`: 64 |
|
|
- `learning_rate`: 2e-05 |
|
|
- `num_train_epochs`: 4 |
|
|
- `warmup_ratio`: 0.1 |
|
|
- `seed`: 12 |
|
|
- `bf16`: True |
|
|
- `dataloader_num_workers`: 4 |
|
|
- `load_best_model_at_end`: True |
|
|
|
|
|
#### All Hyperparameters |
|
|
<details><summary>Click to expand</summary> |
|
|
|
|
|
- `overwrite_output_dir`: False |
|
|
- `do_predict`: False |
|
|
- `eval_strategy`: steps |
|
|
- `prediction_loss_only`: True |
|
|
- `per_device_train_batch_size`: 64 |
|
|
- `per_device_eval_batch_size`: 64 |
|
|
- `per_gpu_train_batch_size`: None |
|
|
- `per_gpu_eval_batch_size`: None |
|
|
- `gradient_accumulation_steps`: 1 |
|
|
- `eval_accumulation_steps`: None |
|
|
- `torch_empty_cache_steps`: None |
|
|
- `learning_rate`: 2e-05 |
|
|
- `weight_decay`: 0.0 |
|
|
- `adam_beta1`: 0.9 |
|
|
- `adam_beta2`: 0.999 |
|
|
- `adam_epsilon`: 1e-08 |
|
|
- `max_grad_norm`: 1.0 |
|
|
- `num_train_epochs`: 4 |
|
|
- `max_steps`: -1 |
|
|
- `lr_scheduler_type`: linear |
|
|
- `lr_scheduler_kwargs`: {} |
|
|
- `warmup_ratio`: 0.1 |
|
|
- `warmup_steps`: 0 |
|
|
- `log_level`: passive |
|
|
- `log_level_replica`: warning |
|
|
- `log_on_each_node`: True |
|
|
- `logging_nan_inf_filter`: True |
|
|
- `save_safetensors`: True |
|
|
- `save_on_each_node`: False |
|
|
- `save_only_model`: False |
|
|
- `restore_callback_states_from_checkpoint`: False |
|
|
- `no_cuda`: False |
|
|
- `use_cpu`: False |
|
|
- `use_mps_device`: False |
|
|
- `seed`: 12 |
|
|
- `data_seed`: None |
|
|
- `jit_mode_eval`: False |
|
|
- `use_ipex`: False |
|
|
- `bf16`: True |
|
|
- `fp16`: False |
|
|
- `fp16_opt_level`: O1 |
|
|
- `half_precision_backend`: auto |
|
|
- `bf16_full_eval`: False |
|
|
- `fp16_full_eval`: False |
|
|
- `tf32`: None |
|
|
- `local_rank`: 0 |
|
|
- `ddp_backend`: None |
|
|
- `tpu_num_cores`: None |
|
|
- `tpu_metrics_debug`: False |
|
|
- `debug`: [] |
|
|
- `dataloader_drop_last`: False |
|
|
- `dataloader_num_workers`: 4 |
|
|
- `dataloader_prefetch_factor`: None |
|
|
- `past_index`: -1 |
|
|
- `disable_tqdm`: False |
|
|
- `remove_unused_columns`: True |
|
|
- `label_names`: None |
|
|
- `load_best_model_at_end`: True |
|
|
- `ignore_data_skip`: False |
|
|
- `fsdp`: [] |
|
|
- `fsdp_min_num_params`: 0 |
|
|
- `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} |
|
|
- `fsdp_transformer_layer_cls_to_wrap`: None |
|
|
- `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} |
|
|
- `deepspeed`: None |
|
|
- `label_smoothing_factor`: 0.0 |
|
|
- `optim`: adamw_torch |
|
|
- `optim_args`: None |
|
|
- `adafactor`: False |
|
|
- `group_by_length`: False |
|
|
- `length_column_name`: length |
|
|
- `ddp_find_unused_parameters`: None |
|
|
- `ddp_bucket_cap_mb`: None |
|
|
- `ddp_broadcast_buffers`: False |
|
|
- `dataloader_pin_memory`: True |
|
|
- `dataloader_persistent_workers`: False |
|
|
- `skip_memory_metrics`: True |
|
|
- `use_legacy_prediction_loop`: False |
|
|
- `push_to_hub`: False |
|
|
- `resume_from_checkpoint`: None |
|
|
- `hub_model_id`: None |
|
|
- `hub_strategy`: every_save |
|
|
- `hub_private_repo`: None |
|
|
- `hub_always_push`: False |
|
|
- `hub_revision`: None |
|
|
- `gradient_checkpointing`: False |
|
|
- `gradient_checkpointing_kwargs`: None |
|
|
- `include_inputs_for_metrics`: False |
|
|
- `include_for_metrics`: [] |
|
|
- `eval_do_concat_batches`: True |
|
|
- `fp16_backend`: auto |
|
|
- `push_to_hub_model_id`: None |
|
|
- `push_to_hub_organization`: None |
|
|
- `mp_parameters`: |
|
|
- `auto_find_batch_size`: False |
|
|
- `full_determinism`: False |
|
|
- `torchdynamo`: None |
|
|
- `ray_scope`: last |
|
|
- `ddp_timeout`: 1800 |
|
|
- `torch_compile`: False |
|
|
- `torch_compile_backend`: None |
|
|
- `torch_compile_mode`: None |
|
|
- `include_tokens_per_second`: False |
|
|
- `include_num_input_tokens_seen`: False |
|
|
- `neftune_noise_alpha`: None |
|
|
- `optim_target_modules`: None |
|
|
- `batch_eval_metrics`: False |
|
|
- `eval_on_start`: False |
|
|
- `use_liger_kernel`: False |
|
|
- `liger_kernel_config`: None |
|
|
- `eval_use_gather_object`: False |
|
|
- `average_tokens_across_devices`: False |
|
|
- `prompts`: None |
|
|
- `batch_sampler`: batch_sampler |
|
|
- `multi_dataset_batch_sampler`: proportional |
|
|
- `router_mapping`: {} |
|
|
- `learning_rate_mapping`: {} |
|
|
|
|
|
</details> |
|
|
|
|
|
### Training Logs |
|
|
| Epoch | Step | Training Loss | Validation Loss | NanoMSMARCO_R100_ndcg@10 | NanoNFCorpus_R100_ndcg@10 | NanoNQ_R100_ndcg@10 | NanoBEIR_R100_mean_ndcg@10 | |
|
|
|:----------:|:----------:|:-------------:|:---------------:|:------------------------:|:-------------------------:|:--------------------:|:--------------------------:| |
|
|
| -1 | -1 | - | - | 0.0376 (-0.5028) | 0.2282 (-0.0968) | 0.0589 (-0.4418) | 0.1082 (-0.3471) | |
|
|
| 0.0000 | 1 | 1.0627 | - | - | - | - | - | |
|
|
| 0.0643 | 10000 | 0.4545 | 0.2603 | 0.6434 (+0.1030) | 0.3663 (+0.0413) | 0.5745 (+0.0738) | 0.5281 (+0.0727) | |
|
|
| 0.1285 | 20000 | 0.2205 | 0.2018 | 0.6276 (+0.0872) | 0.3557 (+0.0307) | 0.6159 (+0.1153) | 0.5331 (+0.0777) | |
|
|
| 0.1928 | 30000 | 0.1885 | 0.1793 | 0.6659 (+0.1254) | 0.3484 (+0.0234) | 0.6520 (+0.1514) | 0.5554 (+0.1001) | |
|
|
| 0.2570 | 40000 | 0.1677 | 0.1668 | 0.6539 (+0.1134) | 0.3207 (-0.0043) | 0.6273 (+0.1267) | 0.5340 (+0.0786) | |
|
|
| 0.3213 | 50000 | 0.1548 | 0.1487 | 0.6292 (+0.0888) | 0.3225 (-0.0025) | 0.6379 (+0.1373) | 0.5299 (+0.0745) | |
|
|
| 0.3855 | 60000 | 0.1425 | 0.1414 | 0.6536 (+0.1131) | 0.3328 (+0.0077) | 0.6662 (+0.1656) | 0.5509 (+0.0955) | |
|
|
| 0.4498 | 70000 | 0.1335 | 0.1233 | 0.6351 (+0.0947) | 0.3493 (+0.0242) | 0.6836 (+0.1830) | 0.5560 (+0.1006) | |
|
|
| 0.5141 | 80000 | 0.1218 | 0.1163 | 0.6462 (+0.1058) | 0.3414 (+0.0164) | 0.6786 (+0.1779) | 0.5554 (+0.1000) | |
|
|
| 0.5783 | 90000 | 0.1134 | 0.1100 | 0.6479 (+0.1075) | 0.3330 (+0.0079) | 0.6717 (+0.1711) | 0.5509 (+0.0955) | |
|
|
| 0.6426 | 100000 | 0.1052 | 0.1079 | 0.6173 (+0.0769) | 0.3226 (-0.0024) | 0.6749 (+0.1743) | 0.5383 (+0.0829) | |
|
|
| **0.7068** | **110000** | **0.0988** | **0.099** | **0.6717 (+0.1313)** | **0.3260 (+0.0010)** | **0.6792 (+0.1786)** | **0.5590 (+0.1036)** | |
|
|
| 0.7711 | 120000 | 0.0942 | 0.0940 | 0.6363 (+0.0958) | 0.3325 (+0.0075) | 0.6437 (+0.1430) | 0.5375 (+0.0821) | |
|
|
| 0.8353 | 130000 | 0.0892 | 0.0870 | 0.6209 (+0.0804) | 0.3352 (+0.0102) | 0.6580 (+0.1574) | 0.5380 (+0.0827) | |
|
|
| 0.8996 | 140000 | 0.0857 | 0.0857 | 0.5993 (+0.0589) | 0.3067 (-0.0183) | 0.6749 (+0.1742) | 0.5270 (+0.0716) | |
|
|
| 0.9639 | 150000 | 0.0818 | 0.0866 | 0.6369 (+0.0965) | 0.3175 (-0.0075) | 0.6541 (+0.1534) | 0.5362 (+0.0808) | |
|
|
| 1.0281 | 160000 | 0.0719 | 0.0827 | 0.5952 (+0.0548) | 0.3211 (-0.0039) | 0.6516 (+0.1509) | 0.5226 (+0.0673) | |
|
|
| 1.0924 | 170000 | 0.064 | 0.0806 | 0.5911 (+0.0507) | 0.3143 (-0.0107) | 0.6243 (+0.1237) | 0.5099 (+0.0545) | |
|
|
| 1.1566 | 180000 | 0.0632 | 0.0826 | 0.5829 (+0.0424) | 0.3478 (+0.0227) | 0.6187 (+0.1180) | 0.5164 (+0.0611) | |
|
|
| 1.2209 | 190000 | 0.0607 | 0.0813 | 0.5620 (+0.0216) | 0.3552 (+0.0301) | 0.6578 (+0.1571) | 0.5250 (+0.0696) | |
|
|
| 1.2851 | 200000 | 0.0601 | 0.0726 | 0.5864 (+0.0459) | 0.3455 (+0.0205) | 0.6074 (+0.1068) | 0.5131 (+0.0577) | |
|
|
| 1.3494 | 210000 | 0.0585 | 0.0723 | 0.5932 (+0.0528) | 0.3585 (+0.0335) | 0.5941 (+0.0935) | 0.5153 (+0.0599) | |
|
|
| 1.4137 | 220000 | 0.0571 | 0.0699 | 0.5746 (+0.0342) | 0.3511 (+0.0260) | 0.6478 (+0.1472) | 0.5245 (+0.0691) | |
|
|
| 1.4779 | 230000 | 0.0561 | 0.0700 | 0.6095 (+0.0691) | 0.3361 (+0.0111) | 0.6229 (+0.1223) | 0.5229 (+0.0675) | |
|
|
| 1.5422 | 240000 | 0.0547 | 0.0686 | 0.5506 (+0.0102) | 0.3530 (+0.0280) | 0.5961 (+0.0955) | 0.4999 (+0.0446) | |
|
|
| 1.6064 | 250000 | 0.0535 | 0.0668 | 0.5916 (+0.0512) | 0.3200 (-0.0050) | 0.6123 (+0.1116) | 0.5080 (+0.0526) | |
|
|
| 1.6707 | 260000 | 0.0511 | 0.0664 | 0.5673 (+0.0269) | 0.3265 (+0.0014) | 0.6145 (+0.1139) | 0.5028 (+0.0474) | |
|
|
| 1.7349 | 270000 | 0.0509 | 0.0610 | 0.5601 (+0.0196) | 0.3419 (+0.0169) | 0.5669 (+0.0662) | 0.4896 (+0.0342) | |
|
|
| 1.7992 | 280000 | 0.0493 | 0.0603 | 0.5825 (+0.0421) | 0.3506 (+0.0256) | 0.5994 (+0.0987) | 0.5108 (+0.0555) | |
|
|
| 1.8635 | 290000 | 0.0484 | 0.0575 | 0.5813 (+0.0409) | 0.3412 (+0.0161) | 0.5948 (+0.0941) | 0.5057 (+0.0504) | |
|
|
| 1.9277 | 300000 | 0.047 | 0.0577 | 0.5941 (+0.0536) | 0.3220 (-0.0031) | 0.6069 (+0.1062) | 0.5076 (+0.0523) | |
|
|
| 1.9920 | 310000 | 0.0462 | 0.0569 | 0.5736 (+0.0332) | 0.3457 (+0.0207) | 0.6148 (+0.1141) | 0.5114 (+0.0560) | |
|
|
| 2.0562 | 320000 | 0.0313 | 0.0696 | 0.5583 (+0.0179) | 0.3437 (+0.0187) | 0.5715 (+0.0709) | 0.4912 (+0.0358) | |
|
|
| 2.1205 | 330000 | 0.0299 | 0.0600 | 0.5913 (+0.0509) | 0.3545 (+0.0295) | 0.5638 (+0.0632) | 0.5032 (+0.0479) | |
|
|
| 2.1847 | 340000 | 0.03 | 0.0649 | 0.5802 (+0.0398) | 0.3652 (+0.0401) | 0.5434 (+0.0428) | 0.4963 (+0.0409) | |
|
|
| 2.2490 | 350000 | 0.0294 | 0.0605 | 0.5719 (+0.0315) | 0.3391 (+0.0141) | 0.5393 (+0.0386) | 0.4834 (+0.0281) | |
|
|
| 2.3133 | 360000 | 0.0285 | 0.0688 | 0.5311 (-0.0094) | 0.3426 (+0.0175) | 0.5545 (+0.0538) | 0.4760 (+0.0207) | |
|
|
| 2.3775 | 370000 | 0.0285 | 0.0577 | 0.5534 (+0.0130) | 0.3276 (+0.0026) | 0.5491 (+0.0484) | 0.4767 (+0.0213) | |
|
|
| 2.4418 | 380000 | 0.0274 | 0.0579 | 0.6134 (+0.0730) | 0.3495 (+0.0245) | 0.5290 (+0.0283) | 0.4973 (+0.0419) | |
|
|
| 2.5060 | 390000 | 0.0278 | 0.0521 | 0.5600 (+0.0196) | 0.3021 (-0.0230) | 0.5493 (+0.0487) | 0.4705 (+0.0151) | |
|
|
| 2.5703 | 400000 | 0.0267 | 0.0518 | 0.5386 (-0.0019) | 0.3672 (+0.0421) | 0.5531 (+0.0525) | 0.4863 (+0.0309) | |
|
|
| 2.6345 | 410000 | 0.0251 | 0.0591 | 0.5695 (+0.0291) | 0.3365 (+0.0115) | 0.5451 (+0.0445) | 0.4837 (+0.0284) | |
|
|
| 2.6988 | 420000 | 0.0249 | 0.0528 | 0.5706 (+0.0302) | 0.3504 (+0.0253) | 0.5393 (+0.0386) | 0.4868 (+0.0314) | |
|
|
| 2.7631 | 430000 | 0.0252 | 0.0520 | 0.5621 (+0.0217) | 0.3679 (+0.0429) | 0.5691 (+0.0685) | 0.4997 (+0.0443) | |
|
|
| 2.8273 | 440000 | 0.0236 | 0.0448 | 0.5448 (+0.0044) | 0.3278 (+0.0028) | 0.5799 (+0.0792) | 0.4842 (+0.0288) | |
|
|
| 2.8916 | 450000 | 0.0231 | 0.0472 | 0.5298 (-0.0106) | 0.3355 (+0.0105) | 0.5565 (+0.0558) | 0.4739 (+0.0186) | |
|
|
| 2.9558 | 460000 | 0.0223 | 0.0439 | 0.5841 (+0.0436) | 0.3248 (-0.0002) | 0.5660 (+0.0654) | 0.4916 (+0.0362) | |
|
|
| 3.0201 | 470000 | 0.0181 | 0.0558 | 0.5471 (+0.0067) | 0.3336 (+0.0085) | 0.5550 (+0.0544) | 0.4786 (+0.0232) | |
|
|
| 3.0843 | 480000 | 0.0106 | 0.0594 | 0.5427 (+0.0022) | 0.3420 (+0.0170) | 0.5677 (+0.0670) | 0.4841 (+0.0287) | |
|
|
| 3.1486 | 490000 | 0.0104 | 0.0547 | 0.5623 (+0.0219) | 0.3496 (+0.0246) | 0.5172 (+0.0165) | 0.4764 (+0.0210) | |
|
|
| 3.2129 | 500000 | 0.0103 | 0.0531 | 0.5670 (+0.0265) | 0.3319 (+0.0068) | 0.5542 (+0.0535) | 0.4843 (+0.0290) | |
|
|
| 3.2771 | 510000 | 0.0098 | 0.0556 | 0.5275 (-0.0130) | 0.3228 (-0.0023) | 0.5291 (+0.0284) | 0.4598 (+0.0044) | |
|
|
| 3.3414 | 520000 | 0.0093 | 0.0461 | 0.5574 (+0.0170) | 0.3355 (+0.0105) | 0.5389 (+0.0382) | 0.4773 (+0.0219) | |
|
|
| 3.4056 | 530000 | 0.0096 | 0.0497 | 0.5627 (+0.0222) | 0.3359 (+0.0109) | 0.5338 (+0.0332) | 0.4775 (+0.0221) | |
|
|
| 3.4699 | 540000 | 0.0086 | 0.0485 | 0.5464 (+0.0060) | 0.3435 (+0.0185) | 0.5527 (+0.0520) | 0.4809 (+0.0255) | |
|
|
| 3.5341 | 550000 | 0.0081 | 0.0462 | 0.5699 (+0.0295) | 0.3307 (+0.0057) | 0.5512 (+0.0506) | 0.4839 (+0.0286) | |
|
|
| 3.5984 | 560000 | 0.0077 | 0.0470 | 0.5452 (+0.0047) | 0.3349 (+0.0098) | 0.5386 (+0.0379) | 0.4729 (+0.0175) | |
|
|
| 3.6627 | 570000 | 0.0072 | 0.0460 | 0.5494 (+0.0090) | 0.3268 (+0.0017) | 0.5463 (+0.0456) | 0.4742 (+0.0188) | |
|
|
| 3.7269 | 580000 | 0.007 | 0.0460 | 0.5456 (+0.0051) | 0.3423 (+0.0172) | 0.5805 (+0.0798) | 0.4894 (+0.0341) | |
|
|
| 3.7912 | 590000 | 0.0066 | 0.0448 | 0.5303 (-0.0101) | 0.3347 (+0.0097) | 0.5655 (+0.0648) | 0.4768 (+0.0215) | |
|
|
| 3.8554 | 600000 | 0.0062 | 0.0409 | 0.5529 (+0.0125) | 0.3320 (+0.0069) | 0.5678 (+0.0672) | 0.4842 (+0.0289) | |
|
|
| 3.9197 | 610000 | 0.0057 | 0.0416 | 0.5429 (+0.0025) | 0.3308 (+0.0058) | 0.5598 (+0.0592) | 0.4779 (+0.0225) | |
|
|
| 3.9839 | 620000 | 0.0055 | 0.0398 | 0.5429 (+0.0025) | 0.3277 (+0.0026) | 0.5666 (+0.0660) | 0.4791 (+0.0237) | |
|
|
| -1 | -1 | - | - | 0.6717 (+0.1313) | 0.3260 (+0.0010) | 0.6792 (+0.1786) | 0.5590 (+0.1036) | |
|
|
|
|
|
* The bold row denotes the saved checkpoint. |
|
|
|
|
|
### Framework Versions |
|
|
- Python: 3.10.18 |
|
|
- Sentence Transformers: 5.0.0 |
|
|
- Transformers: 4.56.0.dev0 |
|
|
- PyTorch: 2.7.1+cu126 |
|
|
- Accelerate: 1.9.0 |
|
|
- Datasets: 4.0.0 |
|
|
- Tokenizers: 0.21.4 |
|
|
|
|
|
## Citation |
|
|
|
|
|
### BibTeX |
|
|
|
|
|
#### Sentence Transformers |
|
|
```bibtex |
|
|
@inproceedings{reimers-2019-sentence-bert, |
|
|
title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", |
|
|
author = "Reimers, Nils and Gurevych, Iryna", |
|
|
booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", |
|
|
month = "11", |
|
|
year = "2019", |
|
|
publisher = "Association for Computational Linguistics", |
|
|
url = "https://arxiv.org/abs/1908.10084", |
|
|
} |
|
|
``` |
|
|
|
|
|
<!-- |
|
|
## Glossary |
|
|
|
|
|
*Clearly define terms in order to be accessible across audiences.* |
|
|
--> |
|
|
|
|
|
<!-- |
|
|
## Model Card Authors |
|
|
|
|
|
*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.* |
|
|
--> |
|
|
|
|
|
<!-- |
|
|
## Model Card Contact |
|
|
|
|
|
*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.* |
|
|
--> |
