Add link to project page
#6
by
nielsr
HF Staff
- opened
README.md
CHANGED
|
@@ -1,14 +1,14 @@
|
|
| 1 |
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
datasets:
|
| 4 |
-
- Tongyi-Zhiwen/DocQA-RL-1.6K
|
| 5 |
base_model:
|
| 6 |
- deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
tags:
|
| 8 |
- long-context
|
| 9 |
- large-reasoning-model
|
| 10 |
-
pipeline_tag: text-generation
|
| 11 |
-
library_name: transformers
|
| 12 |
---
|
| 13 |
|
| 14 |
# QwenLong-L1: Towards Long-Context Large Reasoning Models with Reinforcement Learning
|
|
@@ -23,6 +23,7 @@ library_name: transformers
|
|
| 23 |
[](https://github.com/Tongyi-Zhiwen/QwenLong-L1)
|
| 24 |
[](https://modelscope.cn/models/iic/QwenLong-L1-32B)
|
| 25 |
[](https://huggingface.co/Tongyi-Zhiwen/QwenLong-L1-32B)
|
|
|
|
| 26 |
|
| 27 |
<!-- **Authors:** -->
|
| 28 |
|
|
@@ -155,8 +156,10 @@ try:
|
|
| 155 |
except ValueError:
|
| 156 |
index = 0
|
| 157 |
|
| 158 |
-
thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("
|
| 159 |
-
|
|
|
|
|
|
|
| 160 |
|
| 161 |
print("thinking content:", thinking_content)
|
| 162 |
print("content:", content)
|
|
@@ -335,11 +338,18 @@ PROJ_DIR="<YOUR_PROJ_DIR_HERE>"
|
|
| 335 |
DATA="<YOUR_DATA_HERE>" # e.g., docmath, frames, 2wikimqa, hotpotqa, musique, narrativeqa, pasper
|
| 336 |
python ${PROJ_DIR}/eval/${DATA}_verify.py \
|
| 337 |
--save_dir "${PROJ_DIR}/results/${DATA}" \
|
| 338 |
-
--save_file "${MODEL_NAME}" \
|
| 339 |
-
--judge_model "deepseek-chat" \
|
| 340 |
--batch_size 20
|
| 341 |
```
|
| 342 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 343 |
## 📝 Citation
|
| 344 |
|
| 345 |
If you find this work is relevant with your research or applications, please feel free to cite our work!
|
|
@@ -350,4 +360,8 @@ If you find this work is relevant with your research or applications, please fee
|
|
| 350 |
journal={arXiv preprint arXiv:2505.17667},
|
| 351 |
year={2025}
|
| 352 |
}
|
| 353 |
-
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
| 2 |
base_model:
|
| 3 |
- deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
|
| 4 |
+
datasets:
|
| 5 |
+
- Tongyi-Zhiwen/DocQA-RL-1.6K
|
| 6 |
+
library_name: transformers
|
| 7 |
+
license: apache-2.0
|
| 8 |
+
pipeline_tag: text-generation
|
| 9 |
tags:
|
| 10 |
- long-context
|
| 11 |
- large-reasoning-model
|
|
|
|
|
|
|
| 12 |
---
|
| 13 |
|
| 14 |
# QwenLong-L1: Towards Long-Context Large Reasoning Models with Reinforcement Learning
|
|
|
|
| 23 |
[](https://github.com/Tongyi-Zhiwen/QwenLong-L1)
|
| 24 |
[](https://modelscope.cn/models/iic/QwenLong-L1-32B)
|
| 25 |
[](https://huggingface.co/Tongyi-Zhiwen/QwenLong-L1-32B)
|
| 26 |
+
[Project Page](https://huggingface.co/Tongyi-Zhiwen/QwenLong-L1-32B)
|
| 27 |
|
| 28 |
<!-- **Authors:** -->
|
| 29 |
|
|
|
|
| 156 |
except ValueError:
|
| 157 |
index = 0
|
| 158 |
|
| 159 |
+
thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("
|
| 160 |
+
")
|
| 161 |
+
content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("
|
| 162 |
+
")
|
| 163 |
|
| 164 |
print("thinking content:", thinking_content)
|
| 165 |
print("content:", content)
|
|
|
|
| 338 |
DATA="<YOUR_DATA_HERE>" # e.g., docmath, frames, 2wikimqa, hotpotqa, musique, narrativeqa, pasper
|
| 339 |
python ${PROJ_DIR}/eval/${DATA}_verify.py \
|
| 340 |
--save_dir "${PROJ_DIR}/results/${DATA}" \
|
| 341 |
+
--save_file \"${MODEL_NAME}\" \
|
| 342 |
+
--judge_model \"deepseek-chat\" \
|
| 343 |
--batch_size 20
|
| 344 |
```
|
| 345 |
|
| 346 |
+
## 🌐 Join the Community
|
| 347 |
+
Chinese users can scan QR codes to join WeChat/DingTalk groups.
|
| 348 |
+
|
| 349 |
+
| WeChat | DingTalk |
|
| 350 |
+
|----------|---------|
|
| 351 |
+
|  |  |
|
| 352 |
+
|
| 353 |
## 📝 Citation
|
| 354 |
|
| 355 |
If you find this work is relevant with your research or applications, please feel free to cite our work!
|
|
|
|
| 360 |
journal={arXiv preprint arXiv:2505.17667},
|
| 361 |
year={2025}
|
| 362 |
}
|
| 363 |
+
```
|
| 364 |
+
|
| 365 |
+
## ⭐️ Star History
|
| 366 |
+
|
| 367 |
+
[](https://star-history.com/#Tongyi-Zhiwen/QwenLong-L1&Timeline)
|