update

app.py

@@ -1,25 +1,68 @@
+import os
+import tempfile
+import subprocess
+import shutil
 import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch
 
-# Load model and tokenizer
-model_id = "prasenjeethowlader099/zetallm_4"
+# Load Qwen code model
+model_name = "Qwen/Qwen2.5-Coder-32B-Instruct"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", device_map="auto")
 
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda" if torch.cuda.is_available() else "cpu")
+def clone_and_read_repo(repo_url):
+    """Clone the repo and return concatenated text of selected files."""
+    temp_dir = tempfile.mkdtemp()
+    try:
+        subprocess.run(["git", "clone", "--depth", "1", repo_url, temp_dir], check=True)
+        repo_text = ""
+        for root, dirs, files in os.walk(temp_dir):
+            for file in files:
+                if file.endswith((".py", ".js", ".ts", ".md", ".txt", ".java", ".c", ".cpp")):
+                    file_path = os.path.join(root, file)
+                    try:
+                        with open(file_path, "r", encoding="utf-8") as f:
+                            repo_text += f"\n\n# File: {os.path.relpath(file_path, temp_dir)}\n" + f.read()
+                    except:
+                        pass
+        return repo_text[:50000]  # limit for model context
+    finally:
+        shutil.rmtree(temp_dir)
 
-# Chat function
-def chat(prompt, history=[]):
-    input_ids = tokenizer.encode(prompt, return_tensors="pt").to(model.device)
-    output = model.generate(input_ids, max_new_tokens=200, do_sample=True, top_k=50, top_p=0.95, temperature=0.7)
-    response = tokenizer.decode(output[0], skip_special_tokens=True)
-    return response
+def generate_code(user_code, repo_url):
+    # Add repo content as extra context
+    context = ""
+    if repo_url and repo_url.strip():
+        try:
+            context = clone_and_read_repo(repo_url)
+        except Exception as e:
+            context = f"[Error fetching repo: {e}]"
+    
+    full_prompt = ""
+    if context:
+        full_prompt += f"Here is the repository context:\n{context}\n\n"
+    full_prompt += f"User request/code:\n{user_code}"
 
-# Gradio UI
-gr.Interface(
-    fn=chat,
-    inputs=gr.Textbox(lines=5, label="Ask Zeeta"),
-    outputs=gr.Textbox(label="Zeeta's Response"),
-    title="CodeAgent Zeeta LLM",
-    description="Chat with prasenjeethowlader099/zetallm_4"
-).launch()
+    messages = [
+        {"role": "system", "content": "You are a helpful coding assistant."},
+        {"role": "user", "content": full_prompt}
+    ]
+    text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    inputs = tokenizer([text], return_tensors="pt").to(model.device)
+    outputs = model.generate(**inputs, max_new_tokens=512)
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+iface = gr.Interface(
+    fn=generate_code,
+    inputs=[
+        gr.Code(label="Write Code or Prompt", language="python", placeholder="# Type your code or question here"),
+        gr.Textbox(label="GitHub Repo URL (optional)", placeholder="https://github.com/user/repo")
+    ],
+    outputs=gr.Code(label="Generated Answer", language="python"),
+    title="Qwen-Coder with Git Repo Context & Monaco Editor",
+    description="Ask questions or provide code, optionally fetch a GitHub repo as context. Uses a Monaco code editor for input and output."
+)
+
+if __name__ == "__main__":
+    iface.launch()

requirements.txt

@@ -1,3 +1,4 @@
-transformers
-torch
-gradio
+transformers>=4.37.0
+gradio>=4.21.0
+torch>=2.1.0
+gitpython