Create app.py

app.py

@@ -0,0 +1,136 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import gradio as gr
+import torch
+
+# ==========================
+#  Load your model
+# ==========================
+
+MODEL_ID = "OSS-forge/DeepSeek-Coder-1.3B-cleaned"
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+if tokenizer.pad_token is None:
+    tokenizer.pad_token = tokenizer.eos_token
+
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_ID,
+    torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
+    device_map="auto" if torch.cuda.is_available() else None,
+)
+model.to(device)
+model.eval()
+
+
+# ==========================
+#  Prompt builder
+# ==========================
+
+def build_instruction_prompt(instruction: str) -> str:
+    return '''
+You are an AI programming assistant, utilizing the DeepSeek Coder model, developed by DeepSeek Company, and you only answer questions related to computer science.
+### Instruction:
+{}
+### Response:
+'''.format(instruction.strip()).lstrip()
+
+
+# ==========================
+#  Gradio logic
+# ==========================
+
+def generate_code(instruction, chat_history, is_first_time):
+    if chat_history is None or is_first_time:
+        chat_history = []
+
+    instruction = instruction.strip()
+    if not instruction:
+        return chat_history, gr.update(value=instruction), False
+
+    prompt = build_instruction_prompt(instruction)
+
+    inputs = tokenizer(
+        prompt,
+        return_tensors="pt",
+        padding=True,
+        truncation=True,
+        max_length=512,
+    ).to(device)
+
+    try:
+        stop_id = tokenizer.convert_tokens_to_ids("<|EOT|>")
+    except Exception:
+        stop_id = tokenizer.eos_token_id
+
+    with torch.no_grad():
+        outputs = model.generate(
+            **inputs,
+            max_new_tokens=512,
+            do_sample=False,
+            pad_token_id=stop_id,
+            eos_token_id=stop_id,
+        )
+
+    input_len = inputs["input_ids"].shape[1]
+    generated_tokens = outputs[0, input_len:]
+    code = tokenizer.decode(generated_tokens, skip_special_tokens=True).strip()
+
+    user_message = f"**Instruction**:\n{instruction}"
+    ai_message = f"**Generated code**:\n```python\n{code}\n```"
+
+    chat_history = chat_history + [
+        {"role": "user", "content": user_message},
+        {"role": "assistant", "content": ai_message},
+    ]
+
+    return chat_history, gr.update(value=""), False
+
+
+def reset_interface():
+    return [], gr.update(value=""), True
+
+
+# ==========================
+#  Gradio UI
+# ==========================
+
+with gr.Blocks(title="Python Code Generator") as demo:
+    gr.Markdown("# 🧠 Python Code Generator")
+    gr.Markdown(
+        "Generate Python code from natural language instructions using your Hugging Face model."
+    )
+
+    with gr.Row():
+        with gr.Column(scale=2):
+            instruction_input = gr.Textbox(
+                label="Instruction",
+                placeholder="Describe the code you want. E.g., 'Write a Python function that checks if a number is prime.'",
+                lines=4,
+            )
+
+            is_first = gr.State(True)
+
+            submit_btn = gr.Button("Generate Code")
+            reset_btn = gr.Button("Start Over")
+
+        with gr.Column(scale=3):
+            chat_output = gr.Chatbot(
+                label="Conversation",
+                height=500,
+            )
+
+    submit_btn.click(
+        fn=generate_code,
+        inputs=[instruction_input, chat_output, is_first],
+        outputs=[chat_output, instruction_input, is_first],
+    )
+
+    reset_btn.click(
+        fn=reset_interface,
+        outputs=[chat_output, instruction_input, is_first],
+    )
+
+if __name__ == "__main__":
+    print("Launching Gradio interface...")
+    demo.queue(max_size=10).launch()