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
@@ -14,27 +14,28 @@ HYBRID_ALPHA = 0.6
 #--------------------------
 # 是否启用检索重排功能
 RERANK_ENABLED=true
-# Rerank model
+
+#----------   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
-#----------   Xinference    ----------------
-# The provider for the AI models to use.
-MODEL_PROVIDER=xinference
-# The OpenAI API key to use.
-OPENAI_API_KEY=xinference
+
+#----------   model - Xinference    ----------------
+MODEL_PROVIDER=xinference  # The provider for the AI models to use.
+OPENAI_API_KEY=xinference # The OpenAI API key to use.
 BASE_URL=http://10.1.0.142:9995
 MODEL=Qwen2-72B-Instruct-GPTQ-Int8
-# Temperature for sampling from the model.
-LLM_TEMPERATURE=0.1
-# Maximum number of tokens to generate.
-#LLM_MAX_TOKENS=
-# Name of the embedding model to use.
+LLM_TEMPERATURE=0.1  # Temperature for sampling from the model.
+#LLM_MAX_TOKENS=   # Maximum number of tokens to generate.
+
+
+#----------   embedding - Xinference    ----------------
+EMBEDDING_PROVIDER=xinference
 EMBEDDING_MODEL=bge-m3
 EMBEDDING_BASE_URL=http://10.1.16.39:9995
-# Dimension of the embedding model to use.
-EMBEDDING_DIM=1024
+EMBEDDING_DIM=1024  # Dimension of the embedding model to use.

 ##----------   OpenAI    ----------------
 ## The provider for the AI models to use.