Improve language tag
Browse filesHi! As the model is multilingual, this is a PR to add other languages than English to the language tag to improve the referencing. Note that 29 languages are announced in the README, but only 13 are explicitly listed. I was therefore only able to add these 13 languages.
README.md
CHANGED
|
@@ -1,96 +1,110 @@
|
|
| 1 |
-
---
|
| 2 |
-
library_name: transformers
|
| 3 |
-
license: apache-2.0
|
| 4 |
-
base_model: Qwen/Qwen2.5-32B-Instruct
|
| 5 |
-
tags:
|
| 6 |
-
- llama-factory
|
| 7 |
-
- full
|
| 8 |
-
- generated_from_trainer
|
| 9 |
-
|
| 10 |
-
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
This model was trained on
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
###
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
-
|
| 75 |
-
-
|
| 76 |
-
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: Qwen/Qwen2.5-32B-Instruct
|
| 5 |
+
tags:
|
| 6 |
+
- llama-factory
|
| 7 |
+
- full
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
datasets:
|
| 10 |
+
- open-thoughts/OpenThoughts2-1M
|
| 11 |
+
language:
|
| 12 |
+
- zho
|
| 13 |
+
- eng
|
| 14 |
+
- fra
|
| 15 |
+
- spa
|
| 16 |
+
- por
|
| 17 |
+
- deu
|
| 18 |
+
- ita
|
| 19 |
+
- rus
|
| 20 |
+
- jpn
|
| 21 |
+
- kor
|
| 22 |
+
- vie
|
| 23 |
+
- tha
|
| 24 |
+
- ara
|
| 25 |
+
model-index:
|
| 26 |
+
- name: OpenThinker2-32B
|
| 27 |
+
results: []
|
| 28 |
+
---
|
| 29 |
+
|
| 30 |
+
<p align="center">
|
| 31 |
+
<img src="https://huggingface.co/datasets/open-thoughts/open-thoughts-114k/resolve/main/open_thoughts.png" width="50%">
|
| 32 |
+
</p>
|
| 33 |
+
|
| 34 |
+
# OpenThinker2-32B
|
| 35 |
+
|
| 36 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct) on the
|
| 37 |
+
[OpenThoughts2-1M](https://huggingface.co/datasets/open-thoughts/OpenThoughts2-1M) dataset.
|
| 38 |
+
|
| 39 |
+
The [OpenThinker2-32B](https://huggingface.co/open-thoughts/OpenThinker2-32B) model is the highest performing open-data model.
|
| 40 |
+
This model improves upon our previous [OpenThinker-32B](https://huggingface.co/open-thoughts/OpenThinker-32B) model, which was trained on 114k examples from [OpenThoughts-114k](https://huggingface.co/datasets/open-thoughts/open-thoughts-114k).
|
| 41 |
+
The numbers reported in the table below are evaluated with our open-source tool [Evalchemy](https://github.com/mlfoundations/Evalchemy).
|
| 42 |
+
|
| 43 |
+
| Model | Data | AIME24 | AIME25 | AMC23 | MATH500 | GPQA-D | LCBv2 |
|
| 44 |
+
| ----------------------------------------------------------------------------------------------- | ---- | ------ | ------ | ----- | ------- | ------ | ----- |
|
| 45 |
+
| [OpenThinker2-32B](https://huggingface.co/open-thoughts/OpenThinker2-32B) | ✅ | 76.7 | 58.7 | 94.0 | 90.8 | 64.1 | 72.5 |
|
| 46 |
+
| [OpenThinker-32B](https://huggingface.co/open-thoughts/OpenThinker-32B) | ✅ | 68.0 | 49.3 | 95.5 | 90.6 | 63.5 | 68.6 |
|
| 47 |
+
| [DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B) | ❌ | 74.7 | 50.0 | 96.5 | 90.0 | 65.8 | 72.3 |
|
| 48 |
+
| [Light-R1-32B](https://huggingface.co/qihoo360/Light-R1-32B) | ✅ | 74.7 | 58.0 | 96.0 | 90.4 | 62.0 | 56.0 |
|
| 49 |
+
| [S1.1-32B](https://huggingface.co/simplescaling/s1.1-32B) | ✅ | 59.3 | 42.7 | 91.5 | 87.4 | 62.0 | 58.7 |
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
## Data
|
| 53 |
+
|
| 54 |
+
This model was trained on the [OpenThoughts2-1M](https://huggingface.co/datasets/open-thoughts/OpenThoughts2-1M) dataset.
|
| 55 |
+
|
| 56 |
+
The [OpenThoughts2-1M](https://huggingface.co/datasets/open-thoughts/OpenThoughts2-1M) dataset was constructed by augmenting [OpenThoughts-114k](https://huggingface.co/datasets/open-thoughts/open-thoughts-114k) with existing datasets like [OpenR1](https://huggingface.co/open-r1), as well as additional math and code reasoning data.
|
| 57 |
+
We generate the additional math and code data by ablating over 26 different question generation methodologies and sampling from the highest performing ones.
|
| 58 |
+
|
| 59 |
+
See the [OpenThoughts2-1M](https://huggingface.co/datasets/open-thoughts/OpenThoughts2-1M) dataset page or our [blog post](https://www.open-thoughts.ai/blog/thinkagain) for additional information.
|
| 60 |
+
|
| 61 |
+
|
| 62 |
+
## Intended uses & limitations
|
| 63 |
+
|
| 64 |
+
Apache 2.0 License
|
| 65 |
+
|
| 66 |
+
|
| 67 |
+
## Training procedure
|
| 68 |
+
|
| 69 |
+
We used 128 4xA100 nodes to train the model for 50 hours.
|
| 70 |
+
|
| 71 |
+
### Training hyperparameters
|
| 72 |
+
|
| 73 |
+
The following hyperparameters were used during training:
|
| 74 |
+
- learning_rate: 8e-05
|
| 75 |
+
- seed: 42
|
| 76 |
+
- distributed_type: multi-GPU
|
| 77 |
+
- num_devices: 512
|
| 78 |
+
- gradient_accumulation_steps: 1
|
| 79 |
+
- total_train_batch_size: 512
|
| 80 |
+
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 81 |
+
- lr_scheduler_type: cosine
|
| 82 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 83 |
+
- num_epochs: 5.0
|
| 84 |
+
|
| 85 |
+
### Framework versions
|
| 86 |
+
|
| 87 |
+
- Transformers 4.46.1
|
| 88 |
+
- Pytorch 2.3.0
|
| 89 |
+
- Datasets 3.1.0
|
| 90 |
+
- Tokenizers 0.20.3
|
| 91 |
+
|
| 92 |
+
More info can be found in our repository: [https://github.com/open-thoughts/open-thoughts](https://github.com/open-thoughts/open-thoughts).
|
| 93 |
+
|
| 94 |
+
# Citation
|
| 95 |
+
```
|
| 96 |
+
@misc{openthoughts,
|
| 97 |
+
author = {Team, OpenThoughts},
|
| 98 |
+
month = jan,
|
| 99 |
+
title = {{Open Thoughts}},
|
| 100 |
+
howpublished = {https://open-thoughts.ai},
|
| 101 |
+
year = {2025}
|
| 102 |
+
}
|
| 103 |
+
```
|
| 104 |
+
|
| 105 |
+
# Links
|
| 106 |
+
- 📊 [OpenThoughts2 and OpenThinker2 Blog Post](https://www.open-thoughts.ai/blog/thinkagain)
|
| 107 |
+
- 💻 [Open Thoughts GitHub Repository](https://github.com/open-thoughts/open-thoughts)
|
| 108 |
+
- 🧠 [OpenThoughts2-1M dataset](https://huggingface.co/datasets/open-thoughts/OpenThoughts2-1M)
|
| 109 |
+
- 🤖 [OpenThinker2-7B model](https://huggingface.co/open-thoughts/OpenThinker2-7B)
|
| 110 |
+
- 🤖 [OpenThinker2-32B model](https://huggingface.co/open-thoughts/OpenThinker2-32B) - this model.
|