important changes

agent.py

@@ -14,11 +14,20 @@ from basic_tools import *
 from typing import List
 import numpy as np
 from datetime import datetime, timedelta
+from sentence_transformers import SentenceTransformer
+import torch
 import heapq
 
-# build a retriever
+os.environ['HF_HOME'] = os.path.join(
+    os.path.expanduser('~'), '.cache', "huggingface")
+
 embeddings = HuggingFaceEmbeddings(
-    model_name="sentence-transformers/all-mpnet-base-v2")  # dim=768
+    model_name="sentence-transformers/all-mpnet-base-v2",
+    # hugging_face_api_key=os.getenv("HF_TOKEN"),
+    model_kwargs={"device": "gpu" if torch.cuda.is_available() else "cpu",
+                  "token": os.getenv("HF_TOKEN")},
+    show_progress=True,
+)
 vector_store: FAISS = FAISS.from_texts(
     texts=[],
     embedding=embeddings)
@@ -208,7 +217,8 @@ class BasicAgent:
             llm = ChatOpenAI(
                 base_url="http://localhost:11432/v1",  # default LM Studio endpoint
                 api_key="not-used",  # required by interface but ignored #type: ignore
-                model="mistral-nemo-instruct-2407",
+                # model="mistral-nemo-instruct-2407",
+                model="llama-3.1-8b-claude-3.7-sonnet-reasoning-distilled",
                 temperature=0.2
             )
         elif provider == "openai":

app.py

@@ -3,7 +3,7 @@ import gradio as gr
 import requests
 import inspect
 import pandas as pd
-from agent import BasicAgent, embeddings, vector_store, 
+from agent import BasicAgent, embeddings, vector_store
 from dotenv import load_dotenv
 import os
 

basic_tools.py

@@ -13,7 +13,7 @@ from langchain_community.vectorstores import SupabaseVectorStore
 from langchain_core.messages import SystemMessage, HumanMessage
 from langchain_core.tools import tool
 from langchain.tools.retriever import create_retriever_tool
-from langchain.requests import RequestsWrapper
+from langchain_community.utilities import RequestsWrapper
 
 # from supabase.client import Client, create_client
 # from langchain.tools.requests.toolkit import RequestsToolkit  # Added for RequestsToolKit
@@ -140,8 +140,8 @@ def arxiv_search(query: str) -> str:
 #     return response.text
 
 # Adding request toolkits
-requests_get = RequestsGetTool(requests_wrapper=requests_wrapper)
-requests_post = RequestsPostTool(requests_wrapper=requests_wrapper)
+requests_get = RequestsGetTool(requests_wrapper=requests_wrapper, allow_dangerous_requests=True)
+requests_post = RequestsPostTool(requests_wrapper=requests_wrapper, allow_dangerous_requests=True)
 
 # @tool
 # def requests_post(url: str, data: dict = {}, json: dict = {}, headers: dict = {}) -> str:

local_gpu_embeddings.py

@@ -0,0 +1,62 @@
+# Using local sentence transformers with GPU
+# from langchain_community.embeddings import HuggingFaceEmbeddings
+import torch
+from sentence_transformers import SentenceTransformer
+from langchain.embeddings.base import Embeddings
+from typing import List
+
+
+class LocalHuggingFaceEmbeddings(Embeddings):
+    """Use local SentenceTransformer embeddings"""
+
+    def __init__(self, model_name: str = "sentence-transformers/all-mpnet-base-v2"):
+        """Initialize the embeddings"""
+
+        # Determine device
+        if torch.cuda.is_available():
+            self.device = "cuda"
+            print("Using CUDA for embeddings")
+        elif hasattr(torch.backends, "mps") and torch.backends.mps.is_available():
+            self.device = "mps"
+            print("Using MPS for embeddings")
+        else:
+            self.device = "cpu"
+            print("Using CPU for embeddings")
+
+        # Load the model
+        self.model = SentenceTransformer(model_name, device=self.device)
+        self.model_name = model_name
+
+    def embed_query(self, text: str) -> List[float]:
+        """Embed a query"""
+        # Creates embedding for a single query
+        embedding = self.model.encode(text, show_progress_bar=False)
+        return embedding.tolist()  # Convert numpy array to list
+
+    def embed_documents(self, texts: List[str]) -> List[List[float]]:
+        """Embed documents"""
+        # Creates embeddings for a list of documents
+        embeddings = self.model.encode(texts, show_progress_bar=True)
+        return embeddings.tolist()  # Convert numpy arrays to lists
+
+
+if __name__ == "__main__":
+    # Create the local embeddings model
+    embeddings = LocalHuggingFaceEmbeddings(
+        model_name="sentence-transformers/all-mpnet-base-v2")
+
+    print("Is CUDA Available? ", torch.cuda.is_available())
+    print("Is MPS Available? ", torch.backends.mps.is_available())
+
+    # Test embeddings
+    sentence = "Hello, how are you?"
+    embed = embeddings.embed_query(sentence)
+    print(f"Embedding length: {len(embed)}")
+    print(f"First few values: {embed[:5]}")
+
+    # Test with multiple sentences
+    sentences = ["Hello, how are you?", "I am fine, thank you.",
+                 "What is the weather like today?"]
+    embeds = embeddings.embed_documents(sentences)
+    print(f"Number of embeddings: {len(embeds)}")
+    print(f"Embedding dimensions: {len(embeds[0])}")