llm

src/use_llm.py

@@ -31,14 +31,20 @@ def main_generate(prompt, model=DEFAULT_MODEL, system_prompt="You are a helpful
 
 
 
-# Use your own token securely via Space secrets or local env
-HF_TOKEN = os.getenv("HF_TOKEN")  # define this in Hugging Face Space Secrets
-MODEL_ID = "thenlper/gte-large"  # or another embedding model like BAAI/bge-base-en
+import os
+from huggingface_hub import InferenceClient
+
+HF_TOKEN = os.getenv("HF_TOKEN")
+MODEL_ID = "thenlper/gte-large"  # embedding model
 
 client = InferenceClient(model=MODEL_ID, token=HF_TOKEN)
 
 def get_embeddings(texts):
     if isinstance(texts, str):
         texts = [texts]
-    embeddings = [client.embed(text) for text in texts]
+    embeddings = []
+    for text in texts:
+        response = client.text_to_vector(text)
+        # response is usually a list of floats (the embedding vector)
+        embeddings.append(response)
     return embeddings