Commit
•
e4e2ea5
1
Parent(s):
e5147e7
Update README.md
Browse files
README.md
CHANGED
|
@@ -13,7 +13,6 @@ license: apache-2.0
|
|
| 13 |
|
| 14 |
Qwen2.5 is the latest series of Qwen large language models. For Qwen2.5, we release a number of base language models and instruction-tuned language models ranging from 0.5 to 72 billion parameters. Qwen2.5 brings the following improvements upon Qwen2:
|
| 15 |
|
| 16 |
-
- Pretrained on our **latest large-scale dataset**, encompassing up to **18T tokens**.
|
| 17 |
- Significantly **more knowledge** and has greatly improved capabilities in **coding** and **mathematics**, thanks to our specialized expert models in these domains.
|
| 18 |
- Significant improvements in **instruction following**, **generating long texts** (over 8K tokens), **understanding structured data** (e.g, tables), and **generating structured outputs** especially JSON. **More resilient to the diversity of system prompts**, enhancing role-play implementation and condition-setting for chatbots.
|
| 19 |
- **Long-context Support** up to 128K tokens and can generate up to 8K tokens.
|
|
@@ -83,7 +82,7 @@ For deployment, we recommend using vLLM. Please refer to our [Github](https://gi
|
|
| 83 |
|
| 84 |
**Note**: Presently, vLLM only supports static YARN, which means the scaling factor remains constant regardless of input length, **potentially impacting performance on shorter texts**. We advise adding the `rope_scaling` configuration only when processing long contexts is required.
|
| 85 |
|
| 86 |
-
##
|
| 87 |
|
| 88 |
Detailed evaluation results are reported in this [📑 blog](https://qwenlm.github.io/blog/qwen2.5/).
|
| 89 |
|
|
|
|
| 13 |
|
| 14 |
Qwen2.5 is the latest series of Qwen large language models. For Qwen2.5, we release a number of base language models and instruction-tuned language models ranging from 0.5 to 72 billion parameters. Qwen2.5 brings the following improvements upon Qwen2:
|
| 15 |
|
|
|
|
| 16 |
- Significantly **more knowledge** and has greatly improved capabilities in **coding** and **mathematics**, thanks to our specialized expert models in these domains.
|
| 17 |
- Significant improvements in **instruction following**, **generating long texts** (over 8K tokens), **understanding structured data** (e.g, tables), and **generating structured outputs** especially JSON. **More resilient to the diversity of system prompts**, enhancing role-play implementation and condition-setting for chatbots.
|
| 18 |
- **Long-context Support** up to 128K tokens and can generate up to 8K tokens.
|
|
|
|
| 82 |
|
| 83 |
**Note**: Presently, vLLM only supports static YARN, which means the scaling factor remains constant regardless of input length, **potentially impacting performance on shorter texts**. We advise adding the `rope_scaling` configuration only when processing long contexts is required.
|
| 84 |
|
| 85 |
+
## Evaluation & Performance
|
| 86 |
|
| 87 |
Detailed evaluation results are reported in this [📑 blog](https://qwenlm.github.io/blog/qwen2.5/).
|
| 88 |
|