| import torch |
| import os |
| import json |
|
|
| from tqdm import tqdm |
| import gc |
|
|
| from .utils import save_json |
|
|
| class BaseDataset: |
| def __init__(self): |
| self.chunk_idx = int(os.environ.get("chunk_idx",0)) |
| self.num_chunks = int(os.environ.get("num_chunks",1)) |
|
|
| def run(self,samples,model,batch_size = 2000): |
| out_samples = [] |
| with torch.no_grad(): |
| messages_list = [] |
| current_messages = [] |
| current_samples = [] |
| for sample in tqdm(samples): |
| messages = sample["messages"] |
| current_messages.append(messages) |
| current_samples.append(sample) |
| if len(current_messages) >= batch_size: |
| messages_list.append([current_messages,current_samples]) |
| current_messages = [] |
| current_samples = [] |
| if current_messages: |
| messages_list.append([current_messages,current_samples]) |
| |
| for current_messages,current_samples in tqdm(messages_list): |
| outputs = model.generate_outputs(current_messages) |
| try: |
| for sample,response in zip(current_samples,outputs): |
| del sample["messages"] |
| sample["response"] = response |
| out_samples.append(sample) |
| except Exception as e: |
| from pdb import set_trace;set_trace() |
| print(e) |
| gc.collect() |
| return out_samples |
|
|
| def cal_matrics(self): |
| pass |
|
|
| def init_dataset(self): |
| pass |
|
|
| def construct_messages(self): |
| pass |
|
|
| def eval(self): |
| model = self.model |
| dataset_path = self.dataset_path |
| output_path = self.output_path |
| num_chunks = self.num_chunks |
| chunk_idx = self.chunk_idx |
| if num_chunks == 1: |
| results_path = os.path.join(output_path,"results.json") |
| matric_path = os.path.join(output_path,"metrics.json") |
| out_samples = self.run(self.samples,model) |
| save_json(results_path,out_samples) |
|
|
| metrics,out_samples = self.cal_metrics(out_samples) |
| save_json(matric_path,metrics) |
| save_json(results_path,out_samples) |
| return metrics |
|
|
| elif num_chunks > 1: |
| results_path = os.path.join(output_path,f"results_{chunk_idx}.json") |
| final_results_path = os.path.join(output_path,"results.json") |
| out_samples = self.run(self.samples,model) |
| save_json(results_path,out_samples) |
|
|
| total_results_path = os.listdir(output_path) |
| total_results_path = [result for result in total_results_path if result.startswith("results_")] |
| if len(total_results_path) == num_chunks: |
| total_results = [] |
| for result in total_results_path: |
| results_path = os.path.join(output_path,result) |
| with open(results_path,"r") as f: |
| total_results.extend(json.load(f)) |
|
|
| save_json(final_results_path,total_results) |
| metrics,out_samples = self.cal_metrics(total_results) |
| matric_path = os.path.join(output_path,"metrics.json") |
| save_json(matric_path,metrics) |
| save_json(final_results_path,out_samples) |
| return metrics |
| else: |
| return None |
| else: |
| raise ValueError("num_chunks must be greater than 0") |
| |
| def _download_file_local(self,local_path,url): |
| |
| |
| os.makedirs(local_path,exist_ok=True) |
| |
| |
| filename = url.split("/")[-1] |
| file_path = os.path.join(local_path, filename) |
| |
| |
| if os.system("which wget > /dev/null 2>&1") == 0: |
| download_cmd = f"wget {url} -O {file_path}" |
| elif os.system("which curl > /dev/null 2>&1") == 0: |
| download_cmd = f"curl -L {url} -o {file_path}" |
| else: |
| raise RuntimeError("Neither wget nor curl is available for downloading") |
|
|
| |
| if os.system(download_cmd) != 0: |
| if os.path.exists(file_path): |
| os.remove(file_path) |
| raise RuntimeError("Failed to download dataset") |
| |
| def _unzip_img_zip_local(self, local_path, zip_filename): |
| |
| zip_file_path = os.path.join(local_path, zip_filename) |
| |
| if zip_filename.endswith('.zip'): |
| if os.system(f"unzip -q {zip_file_path} -d {local_path}") != 0: |
| if os.path.exists(zip_file_path): |
| os.remove(zip_file_path) |
| raise RuntimeError("Failed to extract dataset") |
| elif zip_filename.endswith('.tgz') or zip_filename.endswith('.tar.gz'): |
| if os.system(f"tar -xzf {zip_file_path} -C {local_path}") != 0: |
| if os.path.exists(zip_file_path): |
| os.remove(zip_file_path) |
| raise RuntimeError("Failed to extract dataset") |
| else: |
| if os.path.exists(zip_file_path): |
| os.remove(zip_file_path) |
| raise RuntimeError(f"Unsupported file format: {zip_filename}") |
| |
| if os.path.exists(zip_file_path): |
| os.remove(zip_file_path) |
