OpenCompass/configs/datasets/winogrande
2024-05-30 00:06:39 +08:00
..
deprecated_winogrande_gen_a9ede5.py [Format] Add config lints (#892) 2024-05-14 15:35:58 +08:00
README.md [Doc] Update running command in README (#1206) 2024-05-30 00:06:39 +08:00
winogrande_5shot_gen_6447e6.py [Format] Add config lints (#892) 2024-05-14 15:35:58 +08:00
winogrande_5shot_gen_b36770.py [Sync] update evaluator (#1175) 2024-05-21 14:22:46 +08:00
winogrande_5shot_ll_252f01.py [Sync] update evaluator (#1175) 2024-05-21 14:22:46 +08:00
winogrande_gen_458220.py [Format] Add config lints (#892) 2024-05-14 15:35:58 +08:00
winogrande_gen_a027b6.py [Format] Add config lints (#892) 2024-05-14 15:35:58 +08:00
winogrande_gen.py [Feature] Add huggingface apply_chat_template (#1098) 2024-05-14 14:50:16 +08:00
winogrande_ll_c5cf57.py [Format] Add config lints (#892) 2024-05-14 15:35:58 +08:00
winogrande_ll.py [Sync] Sync with internal codes 2023.01.08 (#777) 2024-01-08 14:07:24 +00:00
winogrande_ppl_55a66e.py [Format] Add config lints (#892) 2024-05-14 15:35:58 +08:00
winogrande_ppl_9307fd.py [Format] Add config lints (#892) 2024-05-14 15:35:58 +08:00

WinoGrande

python3 run.py --models hf_internlm2_7b --datasets winogrande_5shot_ll_252f01 --debug
python3 run.py --models hf_internlm2_chat_7b --datasets winogrande_5shot_gen_b36770 --debug

Base Models

model winogrande
llama-7b-turbomind 71.19
llama-13b-turbomind 76.16
llama-30b-turbomind 80.66
llama-65b-turbomind 82.16
llama-2-7b-turbomind 74.03
llama-2-13b-turbomind 76.48
llama-2-70b-turbomind 83.98
llama-3-8b-turbomind 77.82
llama-3-70b-turbomind 83.43
internlm2-1.8b-turbomind 66.77
internlm2-7b-turbomind 83.50
internlm2-20b-turbomind 84.69
qwen-1.8b-turbomind 61.25
qwen-7b-turbomind 72.06
qwen-14b-turbomind 72.45
qwen-72b-turbomind 82.56
qwen1.5-0.5b-hf 57.38
qwen1.5-1.8b-hf 60.46
qwen1.5-4b-hf 65.90
qwen1.5-7b-hf 70.01
qwen1.5-14b-hf 72.93
qwen1.5-32b-hf 78.69
qwen1.5-72b-hf 80.74
qwen1.5-moe-a2-7b-hf 71.43
mistral-7b-v0.1-hf 78.30
mistral-7b-v0.2-hf 77.51
mixtral-8x7b-v0.1-hf 81.53
mixtral-8x22b-v0.1-hf 86.50
yi-6b-hf 74.35
yi-34b-hf 79.01
deepseek-7b-base-hf 74.11
deepseek-67b-base-hf 79.32

Chat Models

model winogrande
qwen1.5-0.5b-chat-hf 50.51
qwen1.5-1.8b-chat-hf 51.07
qwen1.5-4b-chat-hf 57.54
qwen1.5-7b-chat-hf 65.27
qwen1.5-14b-chat-hf 70.09
qwen1.5-32b-chat-hf 77.90
qwen1.5-72b-chat-hf 80.82
qwen1.5-110b-chat-hf 82.32
internlm2-chat-1.8b-hf 57.62
internlm2-chat-1.8b-sft-hf 57.93
internlm2-chat-7b-hf 73.56
internlm2-chat-7b-sft-hf 73.80
internlm2-chat-20b-hf 81.06
internlm2-chat-20b-sft-hf 81.37
llama-3-8b-instruct-hf 66.22
llama-3-70b-instruct-hf 81.29
llama-3-8b-instruct-lmdeploy 66.93
llama-3-70b-instruct-lmdeploy 81.22
mistral-7b-instruct-v0.1-hf 58.56
mistral-7b-instruct-v0.2-hf 59.43
mixtral-8x7b-instruct-v0.1-hf 65.75