update app.py

app.py

@@ -1,64 +1,79 @@
+import torch
+import transformers
 import gradio as gr
-from huggingface_hub import InferenceClient
+from unsloth import FastLanguageModel
 
-"""
-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("emeses/lab2_model")
+# Load the fine-tuned Unsloth model
+max_seq_length = 2048  # Adjust based on your training
+dtype = None  # None for auto detection
 
+def load_model():
+    model, tokenizer = FastLanguageModel.from_pretrained(
+        model_name="ivwhy/lora_model",  # Your fine-tuned model path
+        max_seq_length=max_seq_length,
+        dtype=dtype,
+        load_in_4bit=True  # Optional: load in 4-bit for efficiency
+    )
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+    # Optional: Add special tokens for chat if needed
+    tokenizer.pad_token = tokenizer.eos_token
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    # Create the pipeline
+    pipeline = transformers.pipeline(
+        "text-generation",
+        model=model,
+        tokenizer=tokenizer,
+        device=0 if torch.cuda.is_available() else -1  # Use GPU if available
+    )
+    
+    return pipeline, tokenizer
 
-    messages.append({"role": "user", "content": message})
+# Load model globally
+generation_pipeline, tokenizer = load_model()
 
-    response = ""
-
-    for message in client.chat_completion(
+def chat_function(message, history, system_prompt, max_new_tokens, temperature):
+    messages = [
+        {"role": "system", "content": system_prompt},
+        {"role": "user", "content": message}
+    ]
+    
+    # Apply chat template
+    prompt = tokenizer.apply_chat_template(
         messages,
-        max_tokens=max_tokens,
-        stream=True,
+        tokenize=False,
+        add_generation_prompt=True,
+    )
+    
+    # Define terminators
+    terminators = [
+        tokenizer.eos_token_id,
+        tokenizer.convert_tokens_to_ids("<|eot_id|>")
+    ]
+    
+    # Generate response
+    outputs = generation_pipeline(
+        prompt,
+        max_new_tokens=max_new_tokens,
+        eos_token_id=terminators,
+        do_sample=True,
         temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
+        top_p=0.9,
+    )
+    
+    # Extract and return just the generated text
+    return outputs[0]["generated_text"][len(prompt):]
 
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+# Create Gradio interface
 demo = gr.ChatInterface(
-    respond,
+    chat_function,
+    textbox=gr.Textbox(placeholder="Enter message here", container=False, scale=7),
+    chatbot=gr.Chatbot(height=400),
     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)",
-        ),
-    ],
+        gr.Textbox("You are helpful AI", label="System Prompt"),
+        gr.Slider(minimum=1, maximum=4000, value=500, label="Max New Tokens"),
+        gr.Slider(minimum=0, maximum=1, value=0.7, label="Temperature")
+    ]
 )
 
-
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()

app_old.py

@@ -0,0 +1,64 @@
+import gradio as gr
+from huggingface_hub import InferenceClient
+
+"""
+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("emeses/lab2_model")
+
+
+def respond(
+    message,
+    history: list[tuple[str, str]],
+    system_message,
+    max_tokens,
+    temperature,
+    top_p,
+):
+    messages = [{"role": "system", "content": system_message}]
+
+    for val in history:
+        if val[0]:
+            messages.append({"role": "user", "content": val[0]})
+        if val[1]:
+            messages.append({"role": "assistant", "content": val[1]})
+
+    messages.append({"role": "user", "content": message})
+
+    response = ""
+
+    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
+
+        response += token
+        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=[
+        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)",
+        ),
+    ],
+)
+
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -2,4 +2,5 @@ huggingface_hub==0.25.2
 transformers
 torch
 gradio
-python-dotenv
+python-dotenv
+accelerate