Spaces:
Running
Running
| import streamlit as st | |
| from huggingface_hub import list_models | |
| from datetime import date, timedelta, datetime, timezone | |
| import pandas as pd | |
| import asyncio | |
| # os.environ["KAGGLE_USERNAME"] = "hahunavth" | |
| # os.environ["KAGGLE_KEY"] = "" | |
| from kaggle.api.kaggle_api_extended import KaggleApi | |
| st.set_page_config(layout="wide") | |
| st.header("Experiments") | |
| def init_kaggle(): | |
| api = KaggleApi() | |
| api.authenticate() | |
| return api | |
| api = init_kaggle() | |
| # @st.cache_resource | |
| # def get_model_list(): | |
| # return [model for model in list(list_models(author="hahunavth", filter="emofs2"))] | |
| def get_ckpt_list(model): | |
| files = model.siblings | |
| file_names = [file.rfilename for file in files if "pytorch_model" in file.rfilename] | |
| return file_names | |
| with st.expander("Filter:"): | |
| with st.form("my_form"): | |
| _author = st.text_input("Author", "hahunavth") | |
| _filter = st.text_input("Slug", "emofs2") | |
| submit = st.form_submit_button("Submit") | |
| models = list( | |
| list_models(author=_author, filter=_filter, sort="last_modified", direction=-1) | |
| ) | |
| _models = [] | |
| now = datetime.now(timezone.utc) | |
| for model in models: | |
| ckpts = get_ckpt_list(model) | |
| ckpts_step = [ckpt.replace("pytorch_model.", "").replace(".bin", "") for ckpt in ckpts] | |
| ckpts_step = [int(ckpt) for ckpt in ckpts_step if ckpt.isdigit()] | |
| ckpts_step.sort() | |
| last_ckpt = "{:,}".format(ckpts_step[-1]) if ckpts_step else None | |
| diff = None | |
| if model.last_modified: | |
| now = now.replace(microsecond=0) | |
| last_modified = model.last_modified.replace(microsecond=0) | |
| diff = now - last_modified | |
| diff = str(diff) | |
| step = model.id.replace('hahunavth/emofs2-exp', '').split('_')[0] | |
| _models.append( | |
| { | |
| "id": model.id, | |
| "last_modified": diff, | |
| "ckpt_list": last_ckpt, | |
| "status": None, | |
| "huggingface": f'https://huggingface.co/{model.id}/tree/main' if model.id else None, | |
| "notebook": f"https://www.kaggle.com/code/hahunavth/ess-vlsp2023-train-{step}" if model.id else None, | |
| "config": f"/kaggle/repo/vlsp2023-ess/config/exp{step}", | |
| # "command": "st.balloons", | |
| # "is_widget": True | |
| } | |
| ) | |
| df = pd.DataFrame(_models) | |
| def get_kernel_status(step): | |
| kaggle_slug = f"ess-vlsp2023-train-{step}" | |
| try: | |
| kernel_status = api.kernel_status(user_name="hahunavth", kernel_slug=kaggle_slug)["status"] | |
| except: | |
| kernel_status = "not found" | |
| return kernel_status | |
| async def set_df_kernel_status(df, n=12): | |
| for i in range(min(len(df), n)): | |
| step = df.iloc[i]["id"].replace('hahunavth/emofs2-exp', '').split('_')[0] | |
| status = get_kernel_status(step) | |
| df.at[i, "status"] = status | |
| print(i, status) | |
| def on_update_kernel_status_click(): | |
| st.session_state["last_update"] = datetime.now().strftime('%Y-%m-%d %H:%M:%S') | |
| get_kernel_status.clear() | |
| st.button("Update kernel status", on_click=on_update_kernel_status_click) | |
| st.text(f"Last updated: {st.session_state.get('last_update', 'not yet')}") | |
| asyncio.run(set_df_kernel_status(df)) | |
| edit_df = st.data_editor( | |
| df, | |
| column_config={ | |
| "huggingface": st.column_config.LinkColumn(), | |
| "notebook": st.column_config.LinkColumn(), | |
| # "a": st.button("Click me"), | |
| }, | |
| ) | |
| # st.dataframe( | |
| # edit_df, | |
| # column_config={ | |
| # "huggingface": st.column_config.LinkColumn(), | |
| # "notebook": st.column_config.LinkColumn(), | |
| # }, | |
| # use_container_width=True | |
| # ) | |