Deploy fine-tuned driver

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.gguf filter=lfs diff=lfs merge=lfs -text

Qwen3.5-0.8B.Q4_K_M.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a6160a46d97148cacded85a0d46df13fea41afca96fca6d3f14ecd71768cb5b6
+size 541903808

app.py

@@ -0,0 +1,96 @@
+import os
+from typing import List, Literal, Optional
+from pydantic import BaseModel, Field
+from fastapi import FastAPI, HTTPException, Header
+import gradio as gr
+from llama_cpp import Llama
+
+# -------------------------------------------------------------
+# Configuration & Secret
+# -------------------------------------------------------------
+MODEL_PATH = "Qwen3.5-0.8B.Q4_K_M.gguf"
+API_KEY = os.getenv("API_KEY", "commitcon-secret-2026")
+
+print("Loading GGUF model via llama-cpp ...")
+llm = Llama(
+    model_path=MODEL_PATH,
+    n_ctx=4096,
+    n_threads=2,
+    verbose=False
+)
+print("Model loaded successfully!")
+
+# -------------------------------------------------------------
+# FastAPI App
+# -------------------------------------------------------------
+app = FastAPI(title="CommitCon Sherpa AI Copilot")
+
+class HistoryMessage(BaseModel):
+    role: Literal["user", "assistant"]
+    content: str = Field(min_length=1, max_length=4000)
+
+class ChatRequest(BaseModel):
+    driver_id: Optional[str] = "DRV_DEFAULT"
+    driver_prompt: str = Field(min_length=1, max_length=2000)
+    current_zone_id: Optional[str] = None
+    history: List[HistoryMessage] = Field(default_factory=list)
+
+SYSTEM_PROMPT = (
+    "You are Sherpa, an AI driving assistant for the CommitCon ride-hailing repositioning platform in Bengaluru. "
+    "Your job is to provide concise, direct, helpful recommendations to drivers about where to reposition for high demand, "
+    "earnings, and low traffic delays. Keep your advice practical, realistic, and focused on maximizing their earnings."
+)
+
+def run_inference(prompt: str, history: List[dict] = None) -> str:
+    messages = [{"role": "system", "content": SYSTEM_PROMPT}]
+    if history:
+        for m in history:
+            messages.append({"role": m["role"], "content": m["content"]})
+    messages.append({"role": "user", "content": prompt})
+
+    response = llm.create_chat_completion(
+        messages=messages,
+        temperature=0.7,
+        max_tokens=512,
+    )
+    return response["choices"][0]["message"]["content"]
+
+@app.get("/")
+def root():
+    return {"status": "ok", "service": "CommitCon HF Driver AI Inference Service"}
+
+@app.post("/chat")
+def chat_endpoint(payload: ChatRequest, x_api_key: Optional[str] = Header(None)):
+    if API_KEY and x_api_key != API_KEY:
+        raise HTTPException(status_code=401, detail="Unauthorized: Invalid API Key")
+
+    formatted_history = [{"role": m.role, "content": m.content} for m in payload.history]
+    answer = run_inference(payload.driver_prompt, formatted_history)
+    
+    return {
+        "status": "success",
+        "answer": answer,
+        "model": "project-driver-qwen-gguf",
+        "zone_context": payload.current_zone_id
+    }
+
+# -------------------------------------------------------------
+# Gradio UI Mount (allows testing directly on Hugging Face)
+# -------------------------------------------------------------
+def gradio_chat(user_message, history):
+    hist_formatted = []
+    if history:
+        for user_msg, bot_msg in history:
+            hist_formatted.append({"role": "user", "content": user_msg})
+            if bot_msg:
+                hist_formatted.append({"role": "assistant", "content": bot_msg})
+    return run_inference(user_message, hist_formatted)
+
+demo = gr.ChatInterface(
+    fn=gradio_chat,
+    title="🧭 CommitCon Sherpa Driver Copilot",
+    description="Fine-tuned Qwen 0.8B Q4_K_M AI Driver Copilot for ride repositioning."
+)
+
+# Mount Gradio onto the FastAPI app
+app = gr.mount_gradio_app(app, demo, path="/")

requirements.txt

@@ -0,0 +1,3 @@
+llama-cpp-python>=0.2.80
+fastapi>=0.115.0
+pydantic>=2.0.0