first commit

README.md

@@ -12,3 +12,26 @@ short_description: 'This model is fine tuned on PHI-2 model with OASST1 dataset
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+
+
+# Phi-2 Assistant
+
+This is a fine-tuned version of Microsoft's Phi-2 model on the OpenAssistant dataset. The model has been trained to be helpful and provide informative responses while maintaining a conversational tone.
+
+## Features
+- Pre-loaded with example prompts
+- Adjustable generation parameters:
+  - Maximum Length: Control the length of the generated response
+  - Temperature: Control the randomness of the response
+  - Top P: Control the diversity of the response
+- User-friendly interface
+
+## Usage
+1. Type your prompt in the text box or select one of the example prompts
+2. Adjust the generation parameters if desired
+3. Click "Submit" to generate a response
+
+## Model Details
+- Base Model: microsoft/phi-2
+- Fine-tuned on: OpenAssistant/oasst1
+- Training Repository: [PHI2-SFT-OASST1](https://huggingface.co/satyanayak/PHI2-SFT-OASST1) 

app.py

@@ -0,0 +1,96 @@
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel
+import torch
+
+# Load the base model and tokenizer
+def load_model():
+    base_model = AutoModelForCausalLM.from_pretrained(
+        "microsoft/phi-2",
+        torch_dtype=torch.float16,
+        device_map="auto",
+        trust_remote_code=True
+    )
+    
+    # Load the fine-tuned adapter
+    model = PeftModel.from_pretrained(
+        base_model,
+        "satyanayak/PHI2-SFT-OASST1",
+        torch_dtype=torch.float16,
+        device_map="auto"
+    )
+    
+    tokenizer = AutoTokenizer.from_pretrained(
+        "microsoft/phi-2",
+        trust_remote_code=True
+    )
+    return model, tokenizer
+
+# Generate response
+def generate_response(prompt, max_length=512, temperature=0.7, top_p=0.9):
+    inputs = tokenizer(f"Human: {prompt}\nAssistant:", return_tensors="pt").to(model.device)
+    
+    outputs = model.generate(
+        **inputs,
+        max_length=max_length,
+        temperature=temperature,
+        top_p=top_p,
+        do_sample=True,
+        pad_token_id=tokenizer.eos_token_id
+    )
+    
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    # Extract only the Assistant's response
+    response = response.split("Assistant:")[-1].strip()
+    return response
+
+# Example prompts
+EXAMPLE_PROMPTS = [
+    "What is the capital of France?",
+    "Write a short poem about autumn.",
+    "Explain quantum computing in simple terms.",
+    "Give me a recipe for chocolate chip cookies.",
+    "What are the benefits of regular exercise?"
+]
+
+# Load model and tokenizer
+print("Loading model...")
+model, tokenizer = load_model()
+print("Model loaded!")
+
+# Create Gradio interface
+demo = gr.Interface(
+    fn=generate_response,
+    inputs=[
+        gr.Textbox(
+            label="Enter your prompt",
+            placeholder="Type your message here...",
+            lines=4
+        ),
+        gr.Slider(
+            minimum=64,
+            maximum=1024,
+            value=512,
+            step=64,
+            label="Maximum Length"
+        ),
+        gr.Slider(
+            minimum=0.1,
+            maximum=1.0,
+            value=0.7,
+            step=0.1,
+            label="Temperature"
+        ),
+        gr.Slider(
+            minimum=0.1,
+            maximum=1.0,
+            value=0.9,
+            step=0.1,
+            label="Top P"
+        )
+    ],
+    outputs=gr.Textbox(label="Response", lines=10),
+    examples=EXAMPLE_PROMPTS,
+    title="Phi-2 Assistant",
+    description="This is a fine-tuned version of Phi-2 on the OpenAssistant dataset. Enter your prompt and adjust generation parameters as needed.",
+) 

requirements.txt

@@ -0,0 +1,5 @@
+gradio>=4.0.0
+torch>=2.0.1
+transformers>=4.31.0
+peft>=0.4.0
+accelerate>=0.21.0