Switch embedding model to Alibaba-NLP/gte-Qwen2-7B-instruct via AI Grid API

config.py

@@ -3,6 +3,7 @@ import os
 # ── API Keys ──────────────────────────────────────────────
 AIGRID_API_KEY_GPT   = os.getenv("AIGRID_API_KEY_GPT", "sk-XaDrxkmJNvrp04SfkHT2ig")
 AIGRID_API_KEY_GEMMA = os.getenv("AIGRID_API_KEY_GEMMA", "sk-eQYZ67KgWjIMcPZb6SwwKg")
+AIGRID_API_KEY_EMBEDDING = os.getenv("AIGRID_API_KEY_EMBEDDING", "sk-CNXeCrle_NG4iCre1XvcAA")
 OPENROUTER_API_KEY   = os.getenv("OPENROUTER_API_KEY", "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")
 
 # ── API Bases ─────────────────────────────────────────────
@@ -11,8 +12,7 @@ OPENROUTER_API_BASE = "https://openrouter.ai/api/v1"
 
 # ── Model Settings ────────────────────────────────────────
 LLM_MODEL         = "gpt-oss-120b"
-EMBEDDING_MODEL   = "intfloat/multilingual-e5-base"
-# EMBEDDING_MODEL   = "sentence-transformers/all-MiniLM-L6-v2"
+EMBEDDING_MODEL   = "Alibaba-NLP/gte-Qwen2-7B-instruct"
 LLM_TEMPERATURE   = 0.1                     # Low = less hallucination
 MAX_TOKENS        = 1024
 

rag_pipeline.py

@@ -4,13 +4,13 @@ from langdetect import detect
 
 from llama_index.core                        import VectorStoreIndex, Document, Settings
 from llama_index.core.retrievers             import VectorIndexRetriever
-from llama_index.embeddings.huggingface      import HuggingFaceEmbedding
+from llama_index.embeddings.openai           import OpenAIEmbedding
 from llama_index.llms.openai_like            import OpenAILike
 from llama_index.core.node_parser            import SimpleNodeParser
 
 from document_processor import load_all_documents, load_uploaded_pdf
 from config import (
-    AIGRID_API_KEY_GPT, AIGRID_API_KEY_GEMMA, AIGRID_API_BASE, 
+    AIGRID_API_KEY_GPT, AIGRID_API_KEY_GEMMA, AIGRID_API_KEY_EMBEDDING, AIGRID_API_BASE, 
     OPENROUTER_API_KEY, OPENROUTER_API_BASE,
     LLM_MODEL, EMBEDDING_MODEL,
     LLM_TEMPERATURE, MAX_TOKENS, TOP_K, SIMILARITY_CUTOFF,
@@ -62,7 +62,11 @@ def get_llm(model_id: str):
 # Initialise models
 # ─────────────────────────────────────────────────────────────────────────────
 def _init_models():
-    embed_model = HuggingFaceEmbedding(model_name=EMBEDDING_MODEL)
+    embed_model = OpenAIEmbedding(
+        model_name=EMBEDDING_MODEL,
+        api_key=AIGRID_API_KEY_EMBEDDING,
+        api_base=AIGRID_API_BASE
+    )
     
     # We use a default LLM for any global index operations if necessary
     default_llm = get_llm(LLM_MODEL)

requirements.txt

@@ -6,6 +6,7 @@ python-dotenv
 llama-index-core
 llama-index-llms-openai-like
 llama-index-embeddings-huggingface
+llama-index-embeddings-openai
 
 # PDF Processing
 pymupdf