fix local ollama embedding now working (#188)

Co-authored-by: chen zhang <[email protected]>

Author: zcuder

api/config/embedder.json

@@ -8,12 +8,6 @@
       "encoding_format": "float"
     }
   },
-  "embedder_ollama": {
-    "client_class": "OllamaClient",
-    "model_kwargs": {
-      "model": "nomic-embed-text"
-    }
-  },
   "retriever": {
     "top_k": 20
   },

api/config/embedder.json.bak renamed to api/config/embedder.ollama.json

@@ -1,14 +1,5 @@
 {
   "embedder": {
-    "client_class": "OpenAIClient",
-    "batch_size": 500,
-    "model_kwargs": {
-      "model": "text-embedding-3-small",
-      "dimensions": 256,
-      "encoding_format": "float"
-    }
-  },
-  "embedder_ollama": {
     "client_class": "OllamaClient",
     "model_kwargs": {
       "model": "nomic-embed-text"

api/data_pipeline.py

@@ -370,7 +370,7 @@ def prepare_data_pipeline(is_ollama_embedder: bool = None):
     splitter = TextSplitter(**configs["text_splitter"])
     embedder_config = get_embedder_config()
 
-    embedder = get_embedder(is_local_ollama=is_ollama_embedder)
+    embedder = get_embedder()
 
     if is_ollama_embedder:
         # Use Ollama document processor for single-document processing

api/rag.py

@@ -229,7 +229,7 @@ def __init__(self, provider="google", model=None, use_s3: bool = False):  # noqa
 
         # Initialize components
         self.memory = Memory()
-        self.embedder = get_embedder(is_local_ollama=self.is_ollama_embedder)
+        self.embedder = get_embedder()
 
         # Patch: ensure query embedding is always single string for Ollama
         def single_string_embedder(query):

api/tools/embedder.py

@@ -3,11 +3,8 @@
 from api.config import configs
 
 
-def get_embedder(is_local_ollama: bool = False) -> adal.Embedder:
-    if is_local_ollama:
-        embedder_config = configs["embedder_ollama"]
-    else:
-        embedder_config = configs["embedder"]
+def get_embedder() -> adal.Embedder:
+    embedder_config = configs["embedder"]
 
     # --- Initialize Embedder ---
     model_client_class = embedder_config["model_client"]