Create app.py

app.py

@@ -0,0 +1,95 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
+import torch
+from threading import Thread
+import gradio as gr
+import spaces
+
+model_id = "ByteDance-Seed/Seed-Coder-8B-Instruct"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    torch_dtype=torch.bfloat16,
+    device_map="auto"
+).eval()
+
+def format_conversation_history(chat_history):
+    messages = []
+    for item in chat_history:
+        role = item["role"]
+        content = item["content"]
+        if isinstance(content, list):
+            content = content[0]["text"] if content and "text" in content[0] else str(content)
+        messages.append({"role": role, "content": content})
+    return messages
+    
+@spaces.GPU()
+def generate_response(input_data, chat_history, max_new_tokens, system_prompt, temperature, top_p, top_k, repetition_penalty):
+    new_message = {"role": "user", "content": input_data}
+    system_message = [{"role": "system", "content": system_prompt}] if system_prompt else []
+    processed_history = format_conversation_history(chat_history)
+    messages = system_message + processed_history + [new_message]
+    inputs = tokenizer.apply_chat_template(
+        messages,
+        add_generation_prompt=True,
+        tokenize=True,
+        return_tensors="pt"
+    ).to(model.device)
+    streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
+    generation_kwargs = {
+        "input_ids": inputs,
+        "streamer": streamer,
+        "max_new_tokens": max_new_tokens,
+        "do_sample": True,
+        "temperature": temperature,
+        "top_p": top_p,
+        "top_k": top_k,
+        "repetition_penalty": repetition_penalty
+    }
+
+    thread = Thread(target=model.generate, kwargs=generation_kwargs)
+    thread.start()
+    outputs = []
+    for text_chunk in streamer:
+        outputs.append(text_chunk)
+        yield "".join(outputs)
+
+demo = gr.ChatInterface(
+    fn=generate_response,
+    additional_inputs=[
+        gr.Slider(label="Max new tokens", minimum=64, maximum=4096, step=1, value=1024),
+        gr.Textbox(
+            label="System Prompt",
+            value="You are a helpful coding assistant specializing in generating accurate and efficient code.",
+            lines=4,
+            placeholder="Change system prompt"
+        ),
+        gr.Slider(label="Temperature", minimum=0.1, maximum=2.0, step=0.1, value=0.7),
+        gr.Slider(label="Top-p", minimum=0.05, maximum=1.0, step=0.05, value=0.9),
+        gr.Slider(label="Top-k", minimum=1, maximum=100, step=1, value=50),
+        gr.Slider(label="Repetition Penalty", minimum=1.0, maximum=2.0, step=0.05, value=1.0)
+    ],
+    examples=[
+        [{"text": "Develop a Python Dijkstra’s algorithm to find the shortest path between nodes in a weighted graph for a navigation app"}],
+        [{"text": "Write an SQL query to retrieve the top 5 most-accessed files in a cloud storage system by download count, including file type and size"}],
+        [{"text": "Write a JavaScript function to validate email address and telephone number using regular expressions."}],
+        [{"text": "Write an HTML/CSS stylesheet to style a multi-level navigation menu with hover effects and mobile compatibility"}],
+    ],
+    cache_examples=False,
+    type="messages",
+    description="""
+    # Seed-Coder-8B-Instruct
+    This model excelling in code generation, code completion, code editing and software engineering tasks and developed by ByteDance Seed team. 
+    It pre-trained on 6 trillion token dataset supporting 89 programming languages.
+    """,
+    fill_height=True,
+    textbox=gr.Textbox(
+        label="Query Input",
+        placeholder="Type your prompt"
+    ),
+    stop_btn="Stop Generation",
+    multimodal=False,
+    theme=gr.themes.Soft()
+)
+
+if __name__ == "__main__":
+    demo.launch()