testbot
try in docker
1fc2558
raw history blame
No virus
2.53 kB
from pathlib import Path
import gradio as gr
from huggingface_hub import HfApi, Repository
from huggingface_hub.utils import RepositoryNotFoundError
from convert import convert
REPO_PATH = Path("bloomz.cpp")
repo = Repository(local_dir="bloomz.cpp", clone_from="https://github.com/NouamaneTazi/bloomz.cpp")
def run(token: str, model_id: str, precision: str, quantization: bool) -> str:
if token == "" or model_id == "":
return """
### Invalid input 🐞
Please fill a token and model_id.
"""
api = HfApi(token=token)
try:
# TODO: make a PR to bloomz.cpp to be able to pass a token
api.model_info(repo_id=model_id, token=False) # only public repos are accessible
except RepositoryNotFoundError:
return f"""
### Error 😒😒😒
Repository {model_id} not found. Only public models are convertible at the moment.
"""
try:
model_path = convert(model_id=model_id, precision=precision, quantization=quantization)
print("[commit_info]", model_path)
return f"""
### Success πŸ”₯
"""
return 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})
"""
except Exception as e:
return f"""
### Error 😒😒😒
{e}
"""
DESCRIPTION = """
The steps are the following:
- Paste a read-access token from hf.co/settings/tokens. Read access is enough given that we will open a PR against the source repo.
- Input a model id from the Hub
- Click "Submit"
- That's it! You'll get feedback if it works or not, and if it worked, you'll get the URL of the opened PR πŸ”₯
⚠️ For now only `pytorch_model.bin` files are supported but we'll extend in the future.
"""
demo = gr.Interface(
title="Convert any BLOOM-like model to be compatible with bloomz.cpp",
description=DESCRIPTION,
allow_flagging="never",
article="Check out the [bloomz.cpp](https://github.com/NouamaneTazi/bloomz.cpp) repo on GitHub",
inputs=[
gr.Text(max_lines=1, label="your hf_token"),
gr.Text(max_lines=1, label="model_id (e.g.: bigscience/bloomz-7b1)"),
gr.Radio(choices=["FP16", "FP32"], label="Precision", value="FP16"),
gr.Checkbox(value=False, label="4-bits quantization"),
],
outputs=[gr.Markdown(label="output")],
fn=run,
).queue()
demo.launch()