优化模型初始化代码

2024-09-04 15:00:38 +08:00
parent 728ee06c5a
commit 97a486e631
6 changed files with 438 additions and 229 deletions
@@ -4,34 +4,48 @@ SQL_DATABASE_URL=mysql+pymysql://zjinfo1:Dy2Bcr53Hm5xRkba@110.42.234.166:3306/zj
 #SQL_DATABASE_URL=mysql+pymysql://zjinfo2:GSKcziSdBixDXwcd@110.42.234.166:3306/zjinfo2
 SQLITE_DATABASE_URL=sqlite:///./source.db

-DASHSCOPE_API_KEY=sk-02c8540e86d84b7ca0e6f4f51bac6e60
-# The provider for the AI models to use.
-MODEL_PROVIDER=dashscope
-# The name of LLM model to use.
-MODEL=qwen-max
+# The number of similar embeddings to return when retrieving documents.
+TOP_K=10
+#--------------------------
+# 是否启用混合检索
+HYBRID_ENABLED = false
+# 混合检索阈值
+HYBRID_ALPHA = 0.6
 # 是否启用检索重排功能
-ENABLE_RERANK=true
-# Name of the embedding model to use.
-EMBEDDING_MODEL=text-embedding-v2
+RERANK_ENABLED=true

-# Dimension of the embedding model to use.
+#----------   rerank- Xinference    ----------------
+RERANK_PROVIDER=xinference
+RERANK_MODEL=bge-reranker-v2-m3
+RERANK_BASE_URL=http://10.1.16.39:9995
+RERANK_TOP_N=5
+RERANK_THRESHOLD=0.3
+
+#----------   model - Xinference    ----------------
+#MODEL_PROVIDER=xinference
+#OPENAI_API_KEY=xinference
+#BASE_URL=http://172.20.0.145:9995
+#MODEL=Qwen2-72B-Instruct-GPTQ-Int8
+## Temperature for sampling from the model.
+#LLM_TEMPERATURE=0.1
+
+#----------   model - dashscope    ----------------
+MODEL_PROVIDER=dashscope
+DASHSCOPE_API_KEY=sk-221d2d202e104618a56002ce2e7dc0d0
+MODEL=qwen-max
+
+
+
+#----------   embedding - Xinference    ----------------
+EMBEDDING_PROVIDER=xinference
+EMBEDDING_MODEL=bge-m3
+EMBEDDING_BASE_URL=http://10.1.16.39:9995
 EMBEDDING_DIM=1024

+
 # The questions to help users get started (multi-line).
 CONVERSATION_STARTERS=本工程指什么？\n总算表有哪些费用？\n项目划分哪些内容构成？\n其他费用表有哪些内容？

-# The OpenAI API key to use.
-# OPENAI_API_KEY=
-
-# Temperature for sampling from the model.
-# LLM_TEMPERATURE=
-
-# Maximum number of tokens to generate.
-# LLM_MAX_TOKENS=
-
-# The number of similar embeddings to return when retrieving documents.
-TOP_K=5
-
 # The time in milliseconds to wait for the stream to return a response.
 STREAM_TIMEOUT=60000

@@ -53,9 +67,8 @@ VECTOR_STORE_PATH=./storage_vector
 BM_RETRIEVER_PATH =./storage_bm


-
 PHOENIX_API_KEY=123456
-PHOENIX_URL=http://localhost:6006/v1/traces
+PHOENIX_URL=http://10.1.6.103:6006/v1/traces
 PHOENIX_PROJECT_NAME=ly_zjapp
 #OTEL_SERVICE_NAME=ly_zjapp
 #OTEL_RESOURCE_ATTRIBUTES=openinference.project.name=ly_zjapp
@@ -82,4 +95,5 @@ SYSTEM_PROMPT="You are a weather forecast agent. You help users to get the weath

 PRJTOJSON_URL = 'http://10.1.6.60:8092'
 PROJECT_TITLE = "您好，我是博微工程理解小助手，您可以问我有关[线路工程]工程数据的相关问题！"
+
 CHAT_UPLOAD_FILECACHE = "./output/uploaded"