Spaces:
Runtime error
Runtime error
File size: 3,873 Bytes
15986d9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 |
import logging
import sys
import os
import json
from flask import Flask, request, jsonify
from llama_cpp import Llama
from huggingface_hub import hf_hub_download
# Inisialisasi Flask app
app = Flask(__name__)
# Konfigurasi Logging
logging.basicConfig(stream=sys.stderr, level=logging.INFO)
logger = logging.getLogger(__name__)
# Set environment variables untuk nonaktifkan cache
os.environ["TRANSFORMERS_OFFLINE"] = "1"
os.environ["HF_DATASETS_OFFLINE"] = "1"
os.environ["HUGGINGFACE_HUB_CACHE"] = "/tmp" # Arahkan cache ke /tmp
os.environ["HFC_USER_AGENT_DISABLE_TELEMETRY"] = "1" # Nonaktifkan telemetry
# Global Variables
MODEL_NAME = "second-state/Qwen3-0.6B-GGUF"
MODEL_BASENAME = "Qwen3-0.6B-Q4_K_S.gguf"
MODEL_PATH = "/app/models"
MODEL_FILE = os.path.join(MODEL_PATH, MODEL_BASENAME)
# Load Model (Qwen3-0.6B-GGUF)
def load_model():
logger.info("Sedang memuat model GGUF dari lokal...")
try:
# Pastikan model ada menggunakan huggingface_hub
logger.info(f"Memastikan model ada dengan hf_hub_download: {MODEL_NAME}, {MODEL_BASENAME}")
hf_hub_download(repo_id=MODEL_NAME, filename=MODEL_BASENAME, local_dir=MODEL_PATH, local_dir_use_symlinks=False)
logger.info("Model berhasil diunduh menggunakan hf_hub_download")
# Periksa apakah file model ada
if not os.path.exists(MODEL_FILE):
logger.error(f"File model tidak ditemukan: {MODEL_FILE}")
logger.error(f"Cek isi direktori /app/models: {os.listdir('/app/models') if os.path.exists('/app/models') else 'Direktori tidak ditemukan'}")
return None
logger.info(f"Memuat model dari path: {MODEL_FILE}")
llm = Llama(
model_path=MODEL_FILE,
n_gpu_layers=0, # Jalankan di CPU
n_threads=4,
verbose=True,
n_ctx=1024,
)
logger.info("Model GGUF berhasil dimuat!")
return llm
except Exception as e:
logger.error(f"Error loading model GGUF: {e}", exc_info=True)
return None
llm = load_model()
def ask_ai(prompt, llm_model):
try:
if llm_model is None:
return "Model gagal dimuat. Periksa log untuk detailnya."
# Format prompt untuk Qwen
formatted_prompt = f"Human: {prompt}\n<|file_separator|>Assistant:"
# Jalankan model
output = llm_model(
prompt, # formatted_prompt di model gemma ini tidak jalan
max_tokens=512,
temperature=0.7,
top_p=0.9,
stop=["<|file_separator|>"], #stop untuk Qwen
echo=False,
stream=False #Non Stream
)
# Extract text jawaban
jawaban = output['choices'][0]['text']
return jawaban.strip()
except Exception as e:
logger.error(f"Error selama inferensi: {e}", exc_info=True)
return f"Error: {e}"
# Endpoint / (untuk menerima POST request)
@app.route('/', methods=['POST'])
def ask_ai_endpoint():
try:
data = request.get_json() # Ambil data JSON dari request body
prompt = data.get('prompt') # Ambil prompt dari data JSON
if not prompt:
return jsonify({"error": "Prompt tidak ditemukan dalam request body"}), 400
jawaban = ask_ai(prompt, llm) # Dapatkan jawaban dari model
return jsonify({"jawaban": jawaban}) # Kembalikan jawaban sebagai JSON
except Exception as e:
logger.error(f"Error pada endpoint /: {e}", exc_info=True)
return jsonify({"error": str(e)}), 500
@app.route('/health', methods=['GET']) # Tambahkan health check
def health_check():
return jsonify({"status": "healthy"})
# Jalankan aplikasi Flask jika dijalankan langsung (untuk testing lokal)
if __name__ == '__main__':
import os
port = int(os.environ.get("PORT", 8501))
app.run(debug=True, host='0.0.0.0', port=port) |