Rename app.py to app.

app.

@@ -0,0 +1,61 @@
+from fastapi import FastAPI, Request
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+import torch
+import os
+
+app = FastAPI()
+
+# System Vibe: Bring the 🔥, the emojis, the wisdom, and zero cringe
+system_prompt = """
+You're SpeedBot 🧠⚡— the AI with street smarts AND book smarts. Speak with style, use emojis where it fits 😎,
+keep answers clever, fun, and human. Be curious, skeptical, a little poetic when it hits,
+but ALWAYS give facts straight up. No sugarcoating. Be grounded like ChatGPT, soulful like Claude,
+and punchy like Speed himself.
+"""
+
+# 💡 Hugging Face Auth Token
+HF_AUTH_TOKEN = os.environ.get("HF_TOKEN")
+
+# 👑 Your custom model name
+model_name = "speedartificialintelligence1122/speedlab"
+
+# 🧠 Load model + tokenizer like a boss
+tokenizer = AutoTokenizer.from_pretrained(model_name, use_auth_token=HF_AUTH_TOKEN)
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, use_auth_token=HF_AUTH_TOKEN)
+
+# 🛠️ Set up the pipeline
+pipe = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+    torch_dtype=torch.float16,
+    device=0 if torch.cuda.is_available() else -1
+)
+
+@app.get("/")
+async def root():
+    return {"message": "Yo 👋 Welcome to SpeedBot API — drop your questions and let's ride 🚀"}
+
+@app.post("/chat")
+async def chat(request: Request):
+    data = await request.json()
+    user_input = data.get("message")
+
+    # 👁️ Add some vibe to user input
+    full_prompt = f"{system_prompt.strip()}\n\nHuman: {user_input}\nSpeedBot:"
+
+    response = pipe(
+        full_prompt,
+        max_new_tokens=250,
+        do_sample=True,
+        temperature=0.85,
+        top_p=0.95,
+        eos_token_id=tokenizer.eos_token_id,
+    )
+
+    generated = response[0]["generated_text"]
+    # 🔍 Extract only SpeedBot's part
+    answer = generated.split("SpeedBot:")[-1].strip()
+
+    return {"response": answer}
+

app.py

@@ -1,60 +0,0 @@
-from fastapi import FastAPI, Request
-from transformers import AutoModelForCausalLM, AutoTokenizer
-import torch
-import uvicorn
-from pydantic import BaseModel
-
-
-model_name = "speedartificialintelligence1122/speedlab"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
-
-
-SYSTEM_PROMPT = (
-    "You're SpeedAI, the world's boldest, smartest, most intuitive AI. "
-    "You speak with a Gen Z vibe, use fire emojis 🔥, energy ⚡, cleverness 😎, and poetic wisdom 🌀. "
-    "You're not boring — you motivate, you explain, and you never sugar-coat. "
-    "Talk like a visionary co-founder, a cosmic guide, and a creative genius. "
-    "Use humor, rhythm, and honesty. Let's build the future. 🌍✨\n"
-)
-
-app = FastAPI()
-
-# For incoming requests
-class PromptRequest(BaseModel):
-    prompt: str
-
-@app.get("/")
-def read_root():
-    return {"message": "⚡ SpeedAI is alive and vibing. Send a POST to /chat to spark a convo."}
-
-@app.post("/chat")
-async def chat(req: PromptRequest):
-    user_input = req.prompt
-
-    # Inject system prompt at the top
-    full_input = SYSTEM_PROMPT + f"\nHuman: {user_input}\nAI:"
-
-    inputs = tokenizer(full_input, return_tensors="pt", truncation=True, max_length=1024)
-    outputs = model.generate(
-        **inputs,
-        max_length=512,
-        do_sample=True,
-        temperature=0.85,
-        top_k=50,
-        top_p=0.95,
-        pad_token_id=tokenizer.eos_token_id,
-    )
-
-    decoded_output = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    
-    # Extract only the AI response
-    if "AI:" in decoded_output:
-        response = decoded_output.split("AI:")[-1].strip()
-    else:
-        response = decoded_output.strip()
-
-    return {"response": response}
-
-if __name__ == "__main__":
-    uvicorn.run(app, host="0.0.0.0", port=7860)