zqh11's picture
Update README.md
0a5828f verified
|
raw
history blame
3.23 kB
---
license: other
license_name: deepseek
license_link: https://github.com/deepseek-ai/DeepSeek-Math/blob/main/LICENSE-MODEL
---
<p align="center">
<img width="500px" alt="DeepSeek Chat" src="https://github.com/deepseek-ai/DeepSeek-LLM/blob/main/images/logo.png?raw=true">
</p>
<p align="center"><a href="https://www.deepseek.com/">[🏠Homepage]</a> | <a href="https://chat.deepseek.com/">[🤖 Chat with DeepSeek LLM]</a> | <a href="https://discord.gg/Tc7c45Zzu5">[Discord]</a> | <a href="https://github.com/deepseek-ai/DeepSeek-LLM/blob/main/images/qr.jpeg">[Wechat(微信)]</a> </p>
<p align="center">
<a href="https://arxiv.org/pdf/2402.03300.pdf"><b>Paper Link</b>👁️</a>
</p>
<hr>
### 1. Introduction to DeepSeekMath
See the [Introduction](https://github.com/deepseek-ai/DeepSeek-Math) for more details.
### 2. How to Use
Here give some examples of how to use our model.
**Chat Completion**
❗❗❗ **Please use chain-of-thought prompt to test DeepSeekMath-Instruct and DeepSeekMath-RL:**
- English questions: **{question}\nPlease reason step by step, and put your final answer within \\boxed{}.**
- Chinese questions: **{question}\n请通过逐步推理来解答问题,并把最终答案放置于\\boxed{}中。**
```python
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig
model_name = "deepseek-ai/deepseek-math-7b-instruct"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
model.generation_config = GenerationConfig.from_pretrained(model_name)
model.generation_config.pad_token_id = model.generation_config.eos_token_id
messages = [
{"role": "user", "content": "what is the integral of x^2 from 0 to 2?\nPlease reason step by step, and put your final answer within \\boxed{}."}
]
input_tensor = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt")
outputs = model.generate(input_tensor.to(model.device), max_new_tokens=100)
result = tokenizer.decode(outputs[0][input_tensor.shape[1]:], skip_special_tokens=True)
print(result)
```
Avoiding the use of the provided function `apply_chat_template`, you can also interact with our model following the sample template. Note that `messages` should be replaced by your input.
```
User: {messages[0]['content']}
Assistant: {messages[1]['content']}<|end▁of▁sentence|>User: {messages[2]['content']}
Assistant:
```
**Note:** By default (`add_special_tokens=True`), our tokenizer automatically adds a `bos_token` (`<|begin▁of▁sentence|>`) before the input text. Additionally, since the system prompt is not compatible with this version of our models, we DO NOT RECOMMEND including the system prompt in your input.
### 3. License
This code repository is licensed under the MIT License. The use of DeepSeekMath models is subject to the Model License. DeepSeekMath supports commercial use.
See the [LICENSE-MODEL](https://github.com/deepseek-ai/DeepSeek-Math/blob/main/LICENSE-MODEL) for more details.
### 4. Contact
If you have any questions, please raise an issue or contact us at [service@deepseek.com](mailto:service@deepseek.com).