更新LlamaIndex版本库

backend/app/settings.py

@@ -4,7 +4,8 @@ from abc import abstractmethod
 from llama_index.core.constants import DEFAULT_TEMPERATURE
 from llama_index.core.settings import Settings
 from llama_index.embeddings.xinference import XinferenceEmbedding
-from llama_index.llms.xinference import Xinference
+#from llama_index.llms.xinference import Xinference
+from app.engine.model.xinfeng import XinfengModel
 #from llama_index.embeddings.xinference import XinferenceEmbedding
 from llama_index.llms.xinference.base import DEFAULT_XINFERENCE_TEMP
 from llama_index.postprocessor.xinference_rerank import XinferenceRerank
@@ -96,7 +97,7 @@ class XinferencePlatform(ModelPlatform):
         model = os.getenv("MODEL")
         max_tokens = int(os.getenv("LLM_MAX_TOKENS")) if os.getenv("LLM_MAX_TOKENS") is not None else None
         temperature = float(os.getenv("LLM_TEMPERATURE", DEFAULT_XINFERENCE_TEMP))
-        return Xinference(model, base_url, temperature, max_tokens)
+        return XinfengModel(model_uid = model,endpoint = base_url,temperature = temperature,max_tokens = max_tokens)
     
     def embedding(self):
         base_url = os.getenv("BASE_URL")
@@ -115,7 +116,7 @@ class XinferencePlatform(ModelPlatform):
         rerank_threshold = os.getenv("RERANK_THRESHOLD")
         postprocess = None
         if rerank_model is not None:
-            postprocess = [XinferenceRerank(rerank_model, rerank_url, top_n=rerank_top_n, threshold=rerank_threshold)]
+            postprocess = [XinferenceRerank(model = rerank_model, base_url = rerank_url, top_n=rerank_top_n)]
         return postprocess
 
 @register(ModelPlateCategory,'openai')