Create app.py

app.py

@@ -0,0 +1,92 @@
+import base64
+import gradio as gr
+import torch
+from fastapi import FastAPI, HTTPException
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel
+from typing import List, Optional
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+MODEL_ID = "ApolloAIs/phoebus-3.1-1.5b-merged"
+
+# Load model on CPU (Free Tier)
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_ID,
+    torch_dtype=torch.float32,
+    device_map="cpu"
+)
+
+app = FastAPI()
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+class Message(BaseModel):
+    role: str
+    content: str
+
+class ChatRequest(BaseModel):
+    messages: List[Message]
+    temperature: Optional[float] = 0.7
+    max_tokens: Optional[int] = 256
+
+def format_emem_cell64(claim: str) -> str:
+    encoded = base64.b64encode(claim.encode("utf-8")).decode("utf-8")[:12]
+    return f"emem:fact:cell64_{encoded}"
+
+@app.get("/status")
+def status():
+    return {"status": "online", "model": "Phoebus 3.1"}
+
+@app.post("/v1/chat/completions")
+def chat_api(req: ChatRequest):
+    try:
+        user_msg = req.messages[-1].content
+        context = ""
+
+        if "search" in user_msg.lower():
+            context += "\n\n[TinyFish]: Telemetry and spatial validation indexed."
+
+        if "location" in user_msg.lower() or "coordinates" in user_msg.lower():
+            receipt = format_emem_cell64(user_msg)
+            context += f"\n\n[Vortx emem]: Attach spatial receipt `{receipt}`."
+
+        formatted_messages = [m.model_dump() for m in req.messages]
+        if context:
+            formatted_messages[-1]["content"] += context
+
+        prompt = tokenizer.apply_chat_template(
+            formatted_messages,
+            tokenize=False,
+            add_generation_prompt=True
+        )
+
+        inputs = tokenizer(prompt, return_tensors="pt").to("cpu")
+        with torch.no_grad():
+            outputs = model.generate(
+                **inputs,
+                max_new_tokens=req.max_tokens,
+                temperature=req.temperature,
+                do_sample=True,
+                pad_token_id=tokenizer.eos_token_id
+            )
+
+        reply = tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True)
+        return {"choices": [{"message": {"role": "assistant", "content": reply}}]}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+# Mount FastAPI inside Gradio
+def gradio_chat(user_message, history):
+    req = ChatRequest(messages=[{"role": "user", "content": user_message}])
+    res = chat_api(req)
+    return res["choices"][0]["message"]["content"]
+
+demo = gr.ChatInterface(fn=gradio_chat, title="Phoebus 3.1 API Engine")
+app = gr.mount_gradio_app(app, demo, path="/")