DeepSeek-Coder-V2

DeepSeek-Coder-V2 copied to clipboard

建议

2

1.体验下来，感觉编程能力比GPT4强，但比Claude 3.5 Sonnet差很多 2.编程领域的生态没起来，例如IDE编程插件、官网体验很差，界面看上都很就低级，哪怕拿一个开源的魔改一下也行。这些功能都可以做升级，让使用者感受到你的强大，然后推广你的收费API以及收费服务 3.作为垂直领域的编程大模型，Claude 3.5 Sonnet就是很好的目标，如果能超越他，再加上你低价token，那就是王者了。推广你的API、私有化部署、企业微调，编程插件的订阅等等变现方式

Geek-Bob

Clarification on Code Implementation in DeepSeek vs Llama

2

Dear Authors, Thank you for providing such excellent work for the community to use! I have a question regarding an implementation detail. In [Line 338](https://huggingface.co/deepseek-ai/DeepSeek-Coder-V2-Lite-Base/blob/ea9b066cee82f82906fdd58898cb3788b1c5d770/modeling_deepseek.py#L338), it appears that the code...

dog14230pp

ollama Model is configured wrong: Double BOS

2

Whenever I run `deepseek-coder-v2:latest` through `ollama`, the following Error pops up in the log for each prompt: ``` llm_tokenizer_bpe::check_double_bos_eos: Added a BOS token to the prompt as specified by the...

Snowman-25

 Code used: from transformers import AutoTokenizer from vllm import LLM, SamplingParams max_model_len, tp_size = 8192, 1 model_name = "deepseek-ai/DeepSeek-Coder-V2-Instruct" tokenizer = AutoTokenizer.from_pretrained(model_name) llm = LLM(model=model_name, tensor_parallel_size=tp_size, max_model_len=max_model_len, trust_remote_code=True, enforce_eager=True)...

iamhappytoo

Error: An unknown error was encountered while running the model

1

When generating a response, `deepseek-coder-v2` randomly stops or quits with the following error: ````Error: An unknown error was encountered while running the model.```` I have tested different versions of `deepseek-coder-v2`...

vishnunuk

Hi, thanks for your great work. I found that the number of total tokens reported in the paper (which is around 2T) is far less than the reported further training...

Brucewuzhang

dependency parsing code, deduplication script

1

Hey congrats on the great work! I wonder if you can share the dependency parsing code, deduplication script? thank you!

wentinghome

Thanks for the fantastic work you've done with the deepseek-v2-lite-base! I'm trying to verify if different chars were used as prefixes for the file names in `Repository Level Code Completion.`...

valerybugakov

add fine-tuning code with lora support

5

Closes #35 #41 #58

Muhtasham

请问有考虑发布量化版本吗？

1

ybdesire