Add gradeio app

app.py

@@ -5,8 +5,7 @@ import gradio as gr
 import numpy as np
 import soundfile as sf
 import torch
-from huggingface_hub import hf_hub_download
-from llama_cpp import Llama
+from huggingface_hub import InferenceClient
 from transformers import pipeline
 from TTS.api import TTS
 
@@ -18,19 +17,13 @@ print("Loading Whisper...")
 STT_MODEL_NAME = "openai/whisper-tiny.en"
 stt_pipe = pipeline("automatic-speech-recognition", model=STT_MODEL_NAME, device=device)
 
-# --- 2. LLM Setup (YOUR GGUF MODEL) ---
-print("Downloading and loading your GGUF model...")
-
-# Download your specific model file
-model_path = hf_hub_download(repo_id="gzsol/model_1b", filename="model-Q8_0.gguf")
-
-# Initialize Llama-cpp (Optimized for CPU)
-llm = Llama(
-    model_path=model_path,
-    n_ctx=2048,  # Context window
-    n_threads=2,  # CPU threads
-    verbose=False,
-)
+# --- 2. LLM Setup (HuggingFace Inference API) ---
+print("Setting up HuggingFace Inference API...")
+HF_API_TOKEN = os.getenv("HF_TOKEN")
+if not HF_API_TOKEN:
+    print("Warning: HF_TOKEN not set. Set it as a secret in HF Spaces.")
+llm_client = InferenceClient(token=HF_API_TOKEN)
+LLM_MODEL = "gzsol/model_1b"
 
 # --- 3. TTS Setup (Coqui) ---
 print("Loading TTS...")

requirements.txt

@@ -5,4 +5,3 @@ TTS
 soundfile
 numpy
 huggingface-hub
-llama-cpp-python