msr2000 commited on
Commit
2b9af6b
2 Parent(s): 413b83f cfa9095

Merge branch 'main' of https://huggingface.co/deepseek-ai/DeepSeek-V2-Chat into main

Browse files
Files changed (1) hide show
  1. README.md +18 -17
README.md CHANGED
@@ -70,7 +70,7 @@ We pretrained DeepSeek-V2 on a diverse and high-quality corpus comprising 8.1 tr
70
  | **Model** | **Context Length** | **Download** |
71
  | :------------: | :------------: | :------------: |
72
  | DeepSeek-V2 | 128k | [🤗 HuggingFace](https://huggingface.co/deepseek-ai/DeepSeek-V2) |
73
- | DeepSeek-V2-Chat(RL) | 128k | [🤗 HuggingFace](https://huggingface.co/deepseek-ai/DeepSeek-V2-Chat) |
74
 
75
  </div>
76
 
@@ -82,7 +82,7 @@ Due to the constraints of HuggingFace, the open-source code currently experience
82
 
83
  <div align="center">
84
 
85
- | **Benchmark** | **Domain** | **LLaMA3 70B** | **Mixtral 8x22B** | **DeepSeek V1 (Dense-67B)** | **DeepSeek V2 (MoE-236B)** |
86
  |:-----------:|:--------:|:------------:|:---------------:|:-------------------------:|:------------------------:|
87
  | **MMLU** | English | 78.9 | 77.6 | 71.3 | 78.5 |
88
  | **BBH** | English | 81.0 | 78.9 | 68.7 | 78.9 |
@@ -107,7 +107,7 @@ Evaluation results on the ``Needle In A Haystack`` (NIAH) tests. DeepSeek-V2 pe
107
  #### Standard Benchmark
108
  <div align="center">
109
 
110
- | Benchmark | Domain | QWen1.5 72B Chat | Mixtral 8x22B | LLaMA3 70B Instruct | DeepSeek V1 Chat (SFT) | DeepSeek V2 Chat(SFT) | DeepSeek V2 Chat(RL) |
111
  |:-----------:|:----------------:|:------------------:|:---------------:|:---------------------:|:-------------:|:-----------------------:|:----------------------:|
112
  | **MMLU** | English | 76.2 | 77.8 | 80.3 | 71.1 | 78.4 | 77.8 |
113
  | **BBH** | English | 65.9 | 78.4 | 80.1 | 71.7 | 81.3 | 79.7 |
@@ -134,15 +134,15 @@ We evaluate our model on AlpacaEval 2.0 and MTBench, showing the competitive per
134
  | **模型** | **开源/闭源** | **总分** | **中文推理** | **中文语言** |
135
  | :---: | :---: | :---: | :---: | :---: |
136
  | gpt-4-1106-preview | 闭源 | 8.01 | 7.73 | 8.29 |
137
- | DeepSeek-V2 Chat(RL) | 开源 | 7.91 | 7.45 | 8.35 |
138
- | erniebot-4.0-202404(文心一言) | 闭源 | 7.89 | 7.61 | 8.17 |
139
- | DeepSeek-V2 Chat(SFT) | 开源 | 7.74 | 7.30 | 8.17 |
140
  | gpt-4-0613 | 闭源 | 7.53 | 7.47 | 7.59 |
141
- | erniebot-4.0-202312(文心一言) | 闭源 | 7.36 | 6.84 | 7.88 |
142
- | moonshot-v1-32k-202404(月之暗面) | 闭源 | 7.22 | 6.42 | 8.02 |
143
- | Qwen1.5-72B-Chat(通义千问) | 开源 | 7.19 | 6.45 | 7.93 |
144
  | DeepSeek-67B-Chat | 开源 | 6.43 | 5.75 | 7.11 |
145
- | Yi-34B-Chat(零一万物) | 开源 | 6.12 | 4.86 | 7.38 |
146
  | gpt-3.5-turbo-0613 | 闭源 | 6.08 | 5.35 | 6.71 |
147
 
148
  </div>
@@ -282,13 +282,14 @@ This code repository is licensed under [the MIT License](LICENSE-CODE). The use
282
 
283
  ## 9. Citation
284
  ```
285
- @misc{deepseek-v2,
286
- author = {DeepSeek-AI},
287
- title = {DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model},
288
- year = {2024},
289
- note = {GitHub repository},
290
- url = {https://github.com/deepseek-ai/deepseek-v2}
291
- }
 
292
  ```
293
 
294
  ## 10. Contact
 
70
  | **Model** | **Context Length** | **Download** |
71
  | :------------: | :------------: | :------------: |
72
  | DeepSeek-V2 | 128k | [🤗 HuggingFace](https://huggingface.co/deepseek-ai/DeepSeek-V2) |
73
+ | DeepSeek-V2-Chat (RL) | 128k | [🤗 HuggingFace](https://huggingface.co/deepseek-ai/DeepSeek-V2-Chat) |
74
 
75
  </div>
76
 
 
82
 
83
  <div align="center">
84
 
85
+ | **Benchmark** | **Domain** | **LLaMA3 70B** | **Mixtral 8x22B** | **DeepSeek-V1 (Dense-67B)** | **DeepSeek-V2 (MoE-236B)** |
86
  |:-----------:|:--------:|:------------:|:---------------:|:-------------------------:|:------------------------:|
87
  | **MMLU** | English | 78.9 | 77.6 | 71.3 | 78.5 |
88
  | **BBH** | English | 81.0 | 78.9 | 68.7 | 78.9 |
 
107
  #### Standard Benchmark
108
  <div align="center">
109
 
110
+ | Benchmark | Domain | QWen1.5 72B Chat | Mixtral 8x22B | LLaMA3 70B Instruct | DeepSeek-V1 Chat (SFT) | DeepSeek-V2 Chat (SFT) | DeepSeek-V2 Chat (RL) |
111
  |:-----------:|:----------------:|:------------------:|:---------------:|:---------------------:|:-------------:|:-----------------------:|:----------------------:|
112
  | **MMLU** | English | 76.2 | 77.8 | 80.3 | 71.1 | 78.4 | 77.8 |
113
  | **BBH** | English | 65.9 | 78.4 | 80.1 | 71.7 | 81.3 | 79.7 |
 
134
  | **模型** | **开源/闭源** | **总分** | **中文推理** | **中文语言** |
135
  | :---: | :---: | :---: | :---: | :---: |
136
  | gpt-4-1106-preview | 闭源 | 8.01 | 7.73 | 8.29 |
137
+ | DeepSeek-V2 Chat (RL) | 开源 | 7.91 | 7.45 | 8.35 |
138
+ | erniebot-4.0-202404 (文心一言) | 闭源 | 7.89 | 7.61 | 8.17 |
139
+ | DeepSeek-V2 Chat (SFT) | 开源 | 7.74 | 7.30 | 8.17 |
140
  | gpt-4-0613 | 闭源 | 7.53 | 7.47 | 7.59 |
141
+ | erniebot-4.0-202312 (文心一言) | 闭源 | 7.36 | 6.84 | 7.88 |
142
+ | moonshot-v1-32k-202404 (月之暗面) | 闭源 | 7.22 | 6.42 | 8.02 |
143
+ | Qwen1.5-72B-Chat (通义千问) | 开源 | 7.19 | 6.45 | 7.93 |
144
  | DeepSeek-67B-Chat | 开源 | 6.43 | 5.75 | 7.11 |
145
+ | Yi-34B-Chat (零一万物) | 开源 | 6.12 | 4.86 | 7.38 |
146
  | gpt-3.5-turbo-0613 | 闭源 | 6.08 | 5.35 | 6.71 |
147
 
148
  </div>
 
282
 
283
  ## 9. Citation
284
  ```
285
+ @misc{deepseekv2,
286
+ title={DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model},
287
+ author={DeepSeek-AI},
288
+ year={2024},
289
+ eprint={2405.04434},
290
+ archivePrefix={arXiv},
291
+ primaryClass={cs.CL}
292
+ }
293
  ```
294
 
295
  ## 10. Contact