Trying with existing transformer

app.py

@@ -1,10 +1,15 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
-"""
-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 model and tokenizer directly
+model_name = "jdowling/lora_model"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
+
+# Move the model to the appropriate device (GPU if available, else CPU)
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model.to(device)
 
 
 def respond(
@@ -15,8 +20,8 @@ def respond(
     temperature,
     top_p,
 ):
+    # Prepare prompt with history
     messages = [{"role": "system", "content": system_message}]
-
     for val in history:
         if val[0]:
             messages.append({"role": "user", "content": val[0]})
@@ -25,24 +30,34 @@ def respond(
 
     messages.append({"role": "user", "content": message})
 
-    response = ""
+    # Convert conversation into a single input string
+    prompt = f"{system_message}\n"
+    for turn in messages[1:]:
+        if turn["role"] == "user":
+            prompt += f"User: {turn['content']}\n"
+        elif turn["role"] == "assistant":
+            prompt += f"Assistant: {turn['content']}\n"
+    prompt += "Assistant:"
+
+    # Tokenize input
+    inputs = tokenizer(prompt, return_tensors="pt").to(device)
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
+    # Generate response
+    output = model.generate(
+        inputs["input_ids"],
+        max_length=inputs["input_ids"].shape[1] + max_tokens,
         temperature=temperature,
         top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        pad_token_id=tokenizer.eos_token_id,
+    )
 
-        response += token
-        yield response
+    # Decode response and extract the new assistant message
+    response = tokenizer.decode(output[0], skip_special_tokens=True)
+    response = response[len(prompt):].strip()  # Strip the input part from the response
+
+    yield response
 
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
@@ -59,6 +74,5 @@ demo = gr.ChatInterface(
     ],
 )
 
-
 if __name__ == "__main__":
     demo.launch()