Spaces:
Running
Running
File size: 3,602 Bytes
bd710e9 c015721 bd710e9 c015721 bd710e9 1ccb133 b282fec bd710e9 b282fec 1ccb133 b282fec 1ccb133 bd710e9 1ccb133 b282fec bd710e9 1ccb133 bd710e9 87de695 bd710e9 87de695 bd710e9 87de695 bd710e9 |
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 |
#!/usr/bin/env python3
"""
Hugging Face Spaces entry point for NanoChat.
This file is automatically detected and run by HF Spaces.
"""
import os
import sys
# Set environment variables for HF Spaces BEFORE any other imports
os.environ.setdefault("NANOCHAT_BASE_DIR", "/data")
os.environ.setdefault("TORCHINDUCTOR_CACHE_DIR", "/data/.cache/torch")
os.environ.setdefault("HF_HOME", "/data/.cache/huggingface")
os.environ.setdefault("TORCH_HOME", "/data/.cache/torch")
os.environ.setdefault("XDG_CACHE_HOME", "/data/.cache")
# Download model from HF if not present
def download_model():
"""Download model weights and metadata from Hugging Face."""
# Create all necessary directories
os.makedirs("/data/chatsft_checkpoints/d20", exist_ok=True)
os.makedirs("/data/.cache/huggingface", exist_ok=True)
checkpoint_path = "/data/chatsft_checkpoints/d20/model_000650.pt"
meta_path = "/data/chatsft_checkpoints/d20/meta_000650.json"
if os.path.exists(checkpoint_path) and os.path.exists(meta_path):
print(f"Model checkpoint and metadata found, skipping download")
return
print("Downloading model files from BrianGuo/nanochat-d20-chat...")
from huggingface_hub import hf_hub_download
# Download the checkpoint file
if not os.path.exists(checkpoint_path):
print(" - Downloading model_000650.pt...")
hf_hub_download(
repo_id="BrianGuo/nanochat-d20-chat",
filename="chatsft_checkpoints/d20/model_000650.pt",
local_dir="/data",
local_dir_use_symlinks=False
)
# Download the metadata file
if not os.path.exists(meta_path):
print(" - Downloading meta_000650.json...")
hf_hub_download(
repo_id="BrianGuo/nanochat-d20-chat",
filename="chatsft_checkpoints/d20/meta_000650.json",
local_dir="/data",
local_dir_use_symlinks=False
)
print("Model files downloaded successfully!")
def download_tokenizer():
"""Download tokenizer files from Hugging Face."""
tokenizer_dir = "/data/tokenizer"
if os.path.exists(tokenizer_dir) and os.listdir(tokenizer_dir):
print(f"Tokenizer found, skipping download")
return
print("Downloading tokenizer from BrianGuo/nanochat-d20-chat...")
from huggingface_hub import hf_hub_download
os.makedirs(tokenizer_dir, exist_ok=True)
# Download tokenizer files
hf_hub_download(
repo_id="BrianGuo/nanochat-d20-chat",
filename="tokenizer/token_bytes.pt",
local_dir="/data",
local_dir_use_symlinks=False
)
hf_hub_download(
repo_id="BrianGuo/nanochat-d20-chat",
filename="tokenizer/tokenizer.pkl",
local_dir="/data",
local_dir_use_symlinks=False
)
print("Tokenizer downloaded successfully!")
if __name__ == "__main__":
# Download model and tokenizer before starting
download_model()
download_tokenizer()
# Override sys.argv to pass default arguments for HF Spaces
sys.argv = [
"app.py",
"--port", "7860", # HF Spaces default port
"--host", "0.0.0.0",
"--source", "sft",
"--model-tag", os.environ.get("MODEL_TAG", "d20"),
"--step", os.environ.get("MODEL_STEP", "650"),
]
# Import and run the web server
from scripts.chat_web import app
import uvicorn
print("Starting NanoChat on Hugging Face Spaces...")
print(f"Model: {os.environ.get('MODEL_TAG', 'd20')} - Step: {os.environ.get('MODEL_STEP', '650')}")
uvicorn.run(app, host="0.0.0.0", port=7860)
|