initial

app.py

@@ -1,64 +1,132 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel
+from dotenv import load_dotenv
+import torch
+import cohere
+from pinecone import Pinecone, ServerlessSpec
+import os
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+# Load the environment variables
+load_dotenv()
+# Model configuration
+max_seq_length = 2048
+dtype = None
+load_in_4bit = True
+PINECONE_API_KEY = os.getenv("PINECONE_API_KEY")
 
+class ModelInterface:
+    def __init__(self):
+        # Model names and paths
+        model_name = "Aradhya15/Mistral7b_hypertuned"  # Replace with your Hugging Face model path
+        base_model_name = "mistralai/Mistral-7B-v0.1"
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+        base_model = AutoModelForCausalLM.from_pretrained(base_model_name)
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+        # Load the tokenizer
+        self.tokenizer = AutoTokenizer.from_pretrained(base_model_name)
+        self.tokenizer.pad_token = self.tokenizer.eos_token
+        
+        # Load the PEFT adapter
+        self.model = PeftModel.from_pretrained(base_model, model_name)
+        if torch.cuda.is_available():
+            print("GPU is available and ready to use:", torch.cuda.get_device_name(0))
+        else:
+            print("GPU not detected; using CPU.")
+        # Check for GPU availability
+        if torch.cuda.is_available():
+            device = torch.device("cuda")  # Use the first available GPU
+        else:
+            device = torch.device("cpu")    # Fallback to CPU
 
-    messages.append({"role": "user", "content": message})
+        # Convert the model to half-precision (optional)
+        self.model = self.model.half()
 
-    response = ""
+        # Move the model to the chosen device (GPU or CPU)
+        self.model = self.model.to(device)
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        print(f"Model is moved to {device}")
+        
+        # Initialize Cohere
+        self.cohere_client = cohere.Client(api_key=os.getenv("COHERE_API_KEY"))
+        
+        # Initialize Pinecone with your API key
+        pc = Pinecone(api_key=os.getenv("PINECONE_API_KEY"))
+        self.index = pc.Index("cohere-pinecone-tree")
 
-        response += token
-        yield response
+    def generate_response(self, query):
+        try:
+            # Generate query embedding
+            response = self.cohere_client.embed(
+                texts=[query], 
+                model="embed-english-light-v2.0"
+            )
+            query_embedding = response.embeddings[0]
 
+            # Retrieve documents
+            results = self.index.query(
+                vector=query_embedding, 
+                top_k=5, 
+                include_metadata=True
+            )
+            retrieved_context = "\n".join(
+                [result["metadata"]["text"] for result in results["matches"]]
+            )
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
+            # Prepare input for model
+            messages = [
+                {"role": "system", "content": f"Context: {retrieved_context}"},
+                {"role": "user", "content": query},
+            ]
+            
+            # Tokenize input
+            inputs = self.tokenizer(
+                f"Context: {retrieved_context}\nUser: {query}",
+                return_tensors="pt",
+                truncation=True,
+                padding=True,
+                max_length=max_seq_length
+            ).to("cuda")
 
+            # Generate response from model
+            outputs = self.model.generate(
+                inputs["input_ids"],
+                max_new_tokens=64,
+                temperature=0.3,
+                top_p=0.9,
+                repetition_penalty=1.2
+            )
+            
+            response_text = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
+            return response_text
 
+        except Exception as e:
+            return f"Error generating response: {str(e)}"
+
+# Create Gradio interface
+def create_interface():
+    interface = ModelInterface()
+    
+    def predict(message):
+        return interface.generate_response(message)
+    
+    iface = gr.Interface(
+        fn=predict,
+        inputs=gr.Textbox(label="Enter your question"),
+        outputs=gr.Textbox(label="Response"),
+        title="RAG-Enhanced LLM Assistant",
+        description="Ask a question and get a response enhanced with retrieved context.",
+        examples=[["What are the best practices for tree planting?"], ["How can I improve soil quality in my garden?"]]
+    )
+    
+    return iface
+
+# Launch the interface
 if __name__ == "__main__":
-    demo.launch()
+    iface = create_interface()
+    iface.launch()
+
+# import torch
+# print(torch.__version__)
+# print(torch.cuda.is_available())
+# print(torch.version.cuda)  # This should show "12.6" or compatible

requirements.txt

@@ -1 +1,11 @@
-huggingface_hub==0.25.2
+huggingface_hub==0.25.2
+gradio
+torch>=2.0.0
+transformers>=4.36.0
+unsloth
+cohere
+pinecone-client
+accelerate
+bitsandbytes
+peft
+python-dotenv