import pandas as pd import requests from tqdm.auto import tqdm from huggingface_hub import HfApi, hf_hub_download from huggingface_hub.repocard import metadata_load # Based on Omar Sanseviero work # Make model clickable link def make_clickable_model(model_name): # remove user from model name model_name_show = ' '.join(model_name.split('/')[1:]) link = "https://huggingface.co/" + model_name return f'{model_name_show}' # Make user clickable link def make_clickable_user(user_id): link = "https://huggingface.co/" + user_id return f'{user_id}' def get_model_ids(rl_env): api = HfApi() models = api.list_models(filter=rl_env) model_ids = [x.modelId for x in models] return model_ids def get_metadata(model_id): try: readme_path = hf_hub_download(model_id, filename="README.md") return metadata_load(readme_path) except requests.exceptions.HTTPError: # 404 README.md not found return None def parse_metrics_accuracy(meta): if "model-index" not in meta: return None result = meta["model-index"][0]["results"] metrics = result[0]["metrics"] accuracy = metrics[0]["value"] return accuracy # We keep the worst case episode def parse_rewards(accuracy): default_std = -1000 default_reward=-1000 if accuracy != None: parsed = accuracy.split(' +/- ') if len(parsed)>1: mean_reward = float(parsed[0]) std_reward = float(parsed[1]) else: mean_reward = default_std std_reward = default_reward else: mean_reward = default_std std_reward = default_reward return mean_reward, std_reward