Reflection 70B-AI界的新星来袭：深度解析Reflection-Llama3.1-70b模型，揭秘其超强推理能力与自我纠错技术，带你体验AI思维的革命性突破

hugging face

https://huggingface.co/mattshumer/Reflection-Llama-3.1-70B

算法测试

https://leetcode.com/problems/text-justification/

推理测试

Which is bigger -- 9.11 or 9.9?

How many Rs are in strawberry?

小A去商店买了90元的东西,但发现自己只带了20元。
商店老板借给他80元,小A用这100元付款后,商店老板找回10元给他。
回家后拿到钱后，小A来到商店把80元还给超市老板。
但超市老板总感觉下A给的少了，请问小A应该给多少钱？

ollama

curl -fsSL https://ollama.com/install.sh | sh

ollama run reflection:70b

👉👉👉如有问题请联系我的徽信 stoeng

🔥🔥🔥本项目代码由AI超元域频道制作，观看更多大模型微调视频请访问我的频道⬇

👉👉👉我的哔哩哔哩频道

👉👉👉我的YouTube频道

👉👉👉我的开源项目 https://github.com/win4r/AISuperDomain

api

https://openrouter.ai/chat?models=mattshumer/reflection-70b:free

curl https://openrouter.ai/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -d '{
  "model": "mattshumer/reflection-70b:free",
  "messages": [
    {"role": "user", "content": "What is the meaning of life?"}
  ]
  
}'

LM Studio

https://huggingface.co/lmstudio-community/Reflection-Llama-3.1-70B-GGUF

curl https://api.deepseek.com/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-9d626022379a49a29e4147f593853732" \
  -d '{
        "model": "deepseek-chat",
        "messages": [
          {"role": "system", "content": "You are a helpful assistant."},
          {"role": "user", "content": "Hello!"}
        ],
        "stream": false
      }'

AutoGen+LlamaIndex

!pip install pyautogen llama-index-vector-stores-chroma llama-index llama-index-embeddings-huggingface llama-index-llms-together

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader
from llama_index.vector_stores.chroma import ChromaVectorStore
from llama_index.core import StorageContext
from llama_index.embeddings.huggingface import HuggingFaceEmbedding
from llama_index.core import Settings
from llama_index.llms.together import TogetherLLM

import chromadb
import autogen
from autogen import ConversableAgent



# 创建目录（如果不存在）
!mkdir -p ./documents

# 下载文件到 ./documents 目录
!wget -P ./documents https://raw.githubusercontent.com/win4r/mytest/main/book.txt




def initialize_index():
    # 初始化 Chroma 数据库客户端
    db = chromadb.PersistentClient(path="./chroma_db")
    # 获取或创建一个名为 "my-docs" 的集合
    chroma_collection = db.get_or_create_collection("my-docs")
    # 创建 ChromaVectorStore 实例
    vector_store = ChromaVectorStore(chroma_collection=chroma_collection)
    # 创建存储上下文
    storage_context = StorageContext.from_defaults(vector_store=vector_store)

    # 使用 BAAI/bge-large-en-v1.5 嵌入模型
    embed_model = HuggingFaceEmbedding(model_name="BAAI/bge-large-en-v1.5")

    # 设置全局嵌入模型
    Settings.embed_model = embed_model
    Settings.llm = TogetherLLM( model="meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo", api_key="sk-" )

    # 检查集合是否已存在数据
    if chroma_collection.count() > 0:
        print("Loading existing index...")
        # 如果存在，从向量存储加载索引
        return VectorStoreIndex.from_vector_store(
            vector_store, storage_context=storage_context
        )
    else:
        print("Creating new index...")
        # 如果不存在，从文档目录加载数据并创建新索引
        documents = SimpleDirectoryReader("./documents").load_data()
        return VectorStoreIndex.from_documents(
            documents, storage_context=storage_context
        )

# 初始化索引
index = initialize_index()
# 创建查询引擎
query_engine = index.as_query_engine()



def create_prompt(user_input):
    result = query_engine.query(user_input)

    prompt = f"""
    Your Task: Provide a concise and informative response to the user's query, drawing on the provided context.

    Context: {result}

    User Query: {user_input}

    Guidelines:
    1. Relevance: Focus directly on the user's question.
    2. Conciseness: Avoid unnecessary details.
    3. Accuracy: Ensure factual correctness.
    4. Clarity: Use clear language.
    5. Contextual Awareness: Use general knowledge if context is insufficient.
    6. Honesty: State if you lack information.

    Response Format:
    - Direct answer
    - Brief explanation (if necessary)
    - Citation (if relevant)
    - Conclusion
    """

    return prompt




# 配置LLM(语言模型)
llm_config = {
    "config_list": [
        # {
        #     "model": "llama-3.1-8b-instant",
        #     "api_key": os.getenv("GROQ_API_KEY"),
        #     "api_type": "groq",
        # }
        {
          "model": "mattshumer/reflection-70b:free",
          "base_url": "https://openrouter.ai/api/v1",
          "api_key": "sk-or-v1-",
          "cache_seed": 42
        },
    ]
}

# 创建RAG机器人代理
rag_agent = ConversableAgent(
    name="RAGbot",
    system_message="You are a RAG chatbot",
    llm_config=llm_config,
    code_execution_config=False,
    human_input_mode="NEVER",
)





prompt = create_prompt("Show me some samples of Knowledge Integration prompts")

reply = rag_agent.generate_reply(messages=[{"content": prompt, "role": "user"}])

print("\nType of reply:", type(reply))
print("\nContent of reply:", reply)

if isinstance(reply, dict) and 'content' in reply:
    print(f"\nRAGbot: {reply['content']}")
elif isinstance(reply, str):
    print(f"\nRAGbot: {reply}")
else:
    print("\nUnexpected reply format")