import glob import json import os from dataclasses import dataclass import dateutil from src.display.formatting import make_clickable_model from src.display.utils import AutoEvalColumn, EvalDimensions from src.submission.check_validity import is_model_on_hub @dataclass class EvalResult: """Represents one full evaluation. Built from a combination of the result and request file for a given run. """ eval_name: str # org_model_precision (uid) full_model: str # org/model (path on hub) org: str model: str results: dict model_source: str = "" # HF, API, ... model_category: str = "" #Nano, Small, Medium, Large date: str = "" # submission date of request file still_on_hub: bool = False @classmethod def init_from_json_file(self, json_filepath): """Inits the result from the specific model result file""" with open(json_filepath) as fp: data = json.load(fp) config = data.get("config") # Get model and org org_and_model = config.get("model", config.get("model_args", None)) org_and_model = org_and_model.split("/", 1) if len(org_and_model) == 1: org = None model = org_and_model[0] result_key = f"{model}" else: org = org_and_model[0] model = org_and_model[1] result_key = f"{org}_{model}" full_model = "/".join(org_and_model) still_on_hub, _, _ = is_model_on_hub( full_model, config.get("model_sha", "main"), trust_remote_code=True, test_tokenizer=False ) # Extract results available in this file (some results are split in several files) results = {} results_obj = data.get("results") results["average_score"] = results_obj.get("average_score") results["speed"] = results_obj.get("speed") results["contamination_score"] = results_obj.get("contamination_score") scores_by_category = results_obj.get("scores_by_category") for category_obj in scores_by_category: category = category_obj["category"] average_score = category_obj["average_score"] results[category.lower()] = average_score return self( eval_name=result_key, full_model=full_model, org=org, model=model, model_source=config.get("model_source", ""), model_category=config.get("model_category", ""), results=results, still_on_hub=still_on_hub, ) def update_with_request_file(self, requests_path): """Finds the relevant request file for the current model and updates info with it""" request_file = get_request_file_for_model(requests_path, self.full_model) try: with open(request_file, "r") as f: request = json.load(f) self.date = request.get("submitted_time", "") except Exception: print(f"Could not find request file for {self.org}/{self.model}") def to_dict(self): """Converts the Eval Result to a dict compatible with our dataframe display""" average_score = self.results["average_score"] data_dict = { "eval_name": self.eval_name, # not a column, just a save name, AutoEvalColumn.model_source.name: self.model_source, AutoEvalColumn.model_category.name: self.model_category, AutoEvalColumn.model.name: make_clickable_model(self.full_model), AutoEvalColumn.average_score.name: average_score, } for eval_dim in EvalDimensions: dimension_name = eval_dim.value.col_name try: dimension_value = self.results[eval_dim.value.metric] except KeyError: dimension_value = 0 if dimension_name == "Contamination Score": dimension_value = 0 if dimension_value < 0 else round(dimension_value,2) data_dict[dimension_name] = dimension_value return data_dict def get_request_file_for_model(requests_path, model_name): """Selects the correct request file for a given model. Only keeps runs tagged as FINISHED""" request_files = os.path.join( requests_path, f"{model_name}_eval_request.json", ) request_files = glob.glob(request_files) # Select correct request file (precision) request_file = "" request_files = sorted(request_files, reverse=True) for tmp_request_file in request_files: with open(tmp_request_file, "r") as f: req_content = json.load(f) if ( req_content["status"] in ["FINISHED"] ): request_file = tmp_request_file return request_file def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResult]: """From the path of the results folder root, extract all needed info for results""" model_result_filepaths = [] for root, _, files in os.walk(results_path): ## we allow HTML files now #if len(files) == 0 or any([not f.endswith(".json") for f in files]): # continue files = [f for f in files if f.endswith(".json")] # Sort the files by date try: files.sort(key=lambda x: x.removesuffix(".json").removeprefix("results_")[:-7]) except dateutil.parser._parser.ParserError as e: print("Error",e) files = [files[-1]] for file in files: model_result_filepaths.append(os.path.join(root, file)) eval_results = {} for model_result_filepath in model_result_filepaths: # Creation of result eval_result = EvalResult.init_from_json_file(model_result_filepath) #eval_result.update_with_request_file(requests_path) ##not needed, save processing time # Store results of same eval together eval_name = eval_result.eval_name if eval_name in eval_results.keys(): eval_results[eval_name].results.update({k: v for k, v in eval_result.results.items() if v is not None}) else: eval_results[eval_name] = eval_result results = [] for v in eval_results.values(): try: v.to_dict() # we test if the dict version is complete results.append(v) except KeyError: # not all eval values present print("Key error in eval result, skipping") continue return results def get_model_answers_html_file(results_path, model_name): model_org,model_name_only = model_name.split("/") model_answers_prefix = f"{results_path}/{model_org}/" html_file_content = "EMPTY" download_file_path = "https://huggingface.co/spaces/silma-ai/Arabic-LLM-Broad-Leaderboard/raw/main/" for root, _, files in os.walk(model_answers_prefix): for file_name in files: if file_name.startswith(f"{model_name_only}_abb_benchmark_answers_"): file_path = os.path.join(root, file_name) with open(file_path, "r") as f: html_file_content = f.read() download_file_path = download_file_path + file_path.replace("./", "") break return html_file_content,download_file_path