File size: 5,376 Bytes
63fba53 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 | import os
import torch
import random
import numpy as np
from tqdm import tqdm
import json
from argparse import ArgumentParser
from LLMs import init_llm
import gc
def set_seed(seed_value):
"""
Set the seed for PyTorch (both CPU and CUDA), Python, and NumPy for reproducible results.
:param seed_value: An integer value to be used as the seed.
"""
torch.manual_seed(seed_value)
if torch.cuda.is_available():
torch.cuda.manual_seed(seed_value)
torch.cuda.manual_seed_all(seed_value) # For multi-GPU setups
random.seed(seed_value)
np.random.seed(seed_value)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
def parse_eval_datasets(datasets_str):
return datasets_str.split(',')
def main():
parser = ArgumentParser()
parser.add_argument('--eval_datasets', type=parse_eval_datasets, default=['MMMU'],
help='name of eval dataset')
parser.add_argument('--datasets_path', type=str, default="benchmarks",
help='path of eval dataset')
parser.add_argument('--output_path', type=str, default='eval_results/Qwen2-VL-7B-Instruct',
help='name of saved json')
parser.add_argument('--model_name', type=str, default='Qwen2-VL-7B-Instruct',
help='name of model')
parser.add_argument('--model_path', type=str, default="Qwen/Qwen2-VL-7B-Instruct")
parser.add_argument('--seed', type=int, default=42)
parser.add_argument('--cuda_visible_devices', type=str, default=None)
parser.add_argument('--tensor_parallel_size', type=str, default="1")
parser.add_argument('--use_vllm', type=str, default="True")
parser.add_argument('--reasoning', type=str, default="False")
parser.add_argument('--num_chunks', type=str, default="1")
parser.add_argument('--chunk_idx', type=str, default="0")
parser.add_argument('--max_image_num', type=int, default=1)
parser.add_argument('--max_new_tokens', type=int, default=1024)
parser.add_argument('--temperature', type=float, default=0.0)
parser.add_argument('--top_p', type=float, default=0.001)
parser.add_argument('--repetition_penalty', type=float, default=1)
parser.add_argument('--test_times', type=int, default=1)
parser.add_argument('--use_llm_judge', type=str, default="False")
parser.add_argument('--judge_model_type', type=str, default="openai")
parser.add_argument('--judge_model', type=str,default = "None")
parser.add_argument('--api_key', type=str,default = "None")
parser.add_argument('--base_url', type=str, default= "None")
args = parser.parse_args()
os.environ["VLLM_USE_V1"] = "0"
# llm judge setting
if args.api_key == "None" and args.use_llm_judge == "True":
raise ValueError("If you want to use llm judge, please set the openai api key")
os.environ["use_llm_judge"] = args.use_llm_judge
os.environ["judge_model_type"] = args.judge_model_type
os.environ["judge_model"] = args.judge_model
os.environ["api_key"] = args.api_key
os.environ["base_url"] = args.base_url
# eval modle setting
os.environ["REASONING"] = args.reasoning
os.environ["use_vllm"] = args.use_vllm
os.environ["max_image_num"] = str(args.max_image_num)
# vllm and torch setting
os.environ["PYTORCH_CUDA_ALLOC_CONF"] = "expandable_segments:True"
os.environ["NCCL_IGNORE_DISABLED_P2P"] = "1"
if args.cuda_visible_devices is not None:
os.environ["CUDA_VISIBLE_DEVICES"] = args.cuda_visible_devices
if args.use_vllm == "True":
os.environ["tensor_parallel_size"] = args.tensor_parallel_size
if int(args.tensor_parallel_size) > 1:
os.environ["VLLM_WORKER_MULTIPROC_METHOD"] = "spawn"
os.environ["TOKENIZERS_PARALLELISM"] = "false"
else:
torch.multiprocessing.set_start_method('spawn')
os.environ["num_chunks"] = args.num_chunks
os.environ["chunk_idx"] = args.chunk_idx
os.makedirs(args.output_path, exist_ok=True)
print('initializing LLM...')
model = init_llm(args)
total_results_path = os.path.join(args.output_path,'total_results.json')
from benchmarks import prepare_benchmark
for eval_dataset in tqdm(args.eval_datasets):
set_seed(args.seed)
print(f'evaluating on {eval_dataset}...')
eval_dataset_path = os.path.join(args.datasets_path,eval_dataset) if args.datasets_path != "hf" else None
eval_output_path = os.path.join(args.output_path,eval_dataset)
os.makedirs(eval_output_path, exist_ok=True)
benchmark = prepare_benchmark(model,eval_dataset,eval_dataset_path,eval_output_path)
benchmark.load_data()
final_results = benchmark.eval() if benchmark else {}
print(f'final results on {eval_dataset}: {final_results}')
if final_results is not None:
if os.path.exists(total_results_path):
with open (total_results_path,"r") as f:
total_results = json.load(f)
else:
total_results = {}
total_results[eval_dataset] = final_results
with open(os.path.join(args.output_path,'total_results.json'),'w') as f:
json.dump(total_results,f,indent=4)
gc.collect()
if __name__ == '__main__':
main() |