convert_large / app.py
Narsil's picture
Narsil HF staff
Update app.py
73de75c
import csv
from datetime import datetime
import os
from typing import Optional
import gradio as gr
from convert import convert
from huggingface_hub import HfApi, Repository
DATASET_REPO_URL = "https://huggingface.co/datasets/safetensors/conversions"
DATA_FILENAME = "data.csv"
DATA_FILE = os.path.join("data", DATA_FILENAME)
HF_TOKEN = os.environ.get("HF_TOKEN")
repo: Optional[Repository] = None
# TODO
if False and HF_TOKEN:
repo = Repository(local_dir="data", clone_from=DATASET_REPO_URL, token=HF_TOKEN)
def run(token: str, model_id: str) -> str:
if token == "" or model_id == "":
return """
### Invalid input 🐞
Please fill a token and model_id.
"""
try:
api = HfApi(token=token)
is_private = api.model_info(repo_id=model_id).private
print("is_private", is_private)
commit_info, errors = convert(api=api, model_id=model_id)
print("[commit_info]", commit_info)
# save in a (public) dataset:
# TODO False because of LFS bug.
if False and repo is not None and not is_private:
repo.git_pull(rebase=True)
print("pulled")
with open(DATA_FILE, "a") as csvfile:
writer = csv.DictWriter(
csvfile, fieldnames=["model_id", "pr_url", "time"]
)
writer.writerow(
{
"model_id": model_id,
"pr_url": commit_info.pr_url,
"time": str(datetime.now()),
}
)
commit_url = repo.push_to_hub()
print("[dataset]", commit_url)
string = f"""
### Success πŸ”₯
Yay! This model was successfully converted and a PR was open using your token, here:
[{commit_info.pr_url}]({commit_info.pr_url})
"""
if errors:
string += "\nErrors during conversion:\n"
string += "\n".join(f"Error while converting {filename}: {e}, skipped conversion" for filename, e in errors)
return string
except Exception as e:
return f"""
### Error 😒😒😒
{e}
"""
DESCRIPTION = """
EXPERIMENTAL convertion which should use less disk to enable converting larger checkpoints.
But potential issues on some weights sharing
"""
demo = gr.Interface(
title="Convert any model to Safetensors and open a PR",
description=DESCRIPTION,
allow_flagging="never",
article="Check out the [Safetensors repo on GitHub](https://github.com/huggingface/safetensors)",
inputs=[
gr.Text(max_lines=1, label="your_hf_token"),
gr.Text(max_lines=1, label="model_id"),
],
outputs=[gr.Markdown(label="output")],
fn=run,
).queue(max_size=10, concurrency_count=1)
demo.launch(show_api=True)