ref: ‍⁠​‌​‌​‬‍​​​​‍​​​⁠⁠​⁠​​‬‍‍​​‌​​‬​​《MetaGPT智能体开发入门》教程 - 飞书云文档

【多Agent】MetaGPT学习教程 - 飞书云文档

一、配置本地部署的llm和embedding

llm的配置以及embedding配置：

# Full Example: https://github.com/geekan/MetaGPT/blob/main/config/config2.example.yaml
# Reflected Code: https://github.com/geekan/MetaGPT/blob/main/metagpt/config2.py
# Config Docs: https://docs.deepwisdom.ai/main/en/guide/get_started/configuration.html
llm:
  api_type: 'openai'
  base_url : 'http://0.0.0.0:8000/v1'
  model: 'llama'


# RAG Embedding.
# For backward compatibility, if the embedding is not set and the llm's api_type is either openai or azure, the llm's config will be used. 
embedding:
  api_type: "openai"  "ollama" # openai / azure / gemini / ollama etc. Check EmbeddingType for more options.
  base_url: "http://0.0.0.0:8011/"
  api_key: ""
  model: "bge-large-en-v1.5"
  api_version: "v1"
  embed_batch_size: 100
  dimensions: # output dimension of embedding model

对于llm 如果max_model_len不能承受更多的话，需要修改metagpt\provider\openrouter_reasoning.py

    def _get_max_tokens(self, messages: list[dict]):
        if not self.auto_max_tokens:
            return 1024
            return self.config.max_token
        # FIXME
        # https://community.openai.com/t/why-is-gpt-3-5-turbo-1106-max-tokens-limited-to-4096/494973/3
        return min(get_max_completion_tokens(messages, self.model, self.config.max_token), 1024) #4096)

对于rag案例：

需要修改metagpt\rag\schema.py

class FAISSRetrieverConfig(IndexRetrieverConfig):
    """Config for FAISS-based retrievers."""

    dimensions: int = Field(default=0, description="Dimensionality of the vectors for FAISS index construction.")

    _embedding_type_to_dimensions: ClassVar[dict[EmbeddingType, int]] = {
        EmbeddingType.GEMINI: 768,
        EmbeddingType.OLLAMA: 1024， #4096,
    }

二、用Discord 发送 github trending 跑通

做好配置

整个过程就是配置

export DISCORD_TOKEN=
export DISCORD_CHANNEL_ID=

三、用Discord 发送 github trending动态  进阶版

读取readme总结

对应whole_run2.py

四、用Discord发送 huggingface paper动态 

对应whole_run-huggingface.py

五、用邮件发送github trending动态

对应whole_run-email.py

六、多智能体：你画我猜 multi-ones.py