humaneval+: by models

Home Paper Code


SE predicted by accuracy

The typical standard errors between pairs of models on this dataset as a function of the absolute accuracy.

SE vs accuracy difference

The standard error of each model pair against their observed accuracy difference. Pairs below a reference line differ by more than the corresponding number of standard errors, i.e. they are statistically distinguishable at that level.

CDF of question level accuracy

Results table by model

model pass1 pass@count win_rate count SE(A) SE_x(A) SE_pred(A)
claude-3-opus-20240229 77.4 77.4 25.3 1 3.3 NaN NaN
deepseek-coder-33b-instruct 76.2 76.2 24.1 1 3.3 NaN NaN
opencodeinterpreter-ds-33b 74.4 74.4 23.8 1 3.4 NaN NaN
mixtral-8x22b-instruct-v0.1 73.8 73.8 23.4 1 3.4 NaN NaN
speechless-codellama-34b 72.6 72.6 22 1 3.5 NaN NaN
HuggingFaceH4--starchat2-15b-v0.1 72 72 21.9 1 3.5 NaN NaN
code-millenials-34b 72 72 22.3 1 3.5 NaN NaN
deepseek-coder-6.7b-instruct 72 72 23.2 1 3.5 NaN NaN
meta-llama-3-70b-instruct 72 72 21.9 1 3.5 NaN NaN
deepseek-coder-7b-instruct-v1.5 71.3 71.3 21.6 1 3.5 NaN NaN
gpt-3.5-turbo 70.7 70.7 20.8 1 3.6 NaN NaN
opencodeinterpreter-ds-6.7b 70.7 70.7 21.4 1 3.6 NaN NaN
xwincoder-34b 70.1 70.1 21.2 1 3.6 NaN NaN
claude-3-haiku-20240307 68.9 68.9 20.5 1 3.6 NaN NaN
openchat 68.9 68.9 20.5 1 3.6 NaN NaN
speechless-coder-ds-6.7b 66.5 66.5 17.9 1 3.7 NaN NaN
code-llama-70b-instruct 66.5 66.5 19.6 1 3.7 NaN NaN
white-rabbit-neo-33b-v1 65.9 65.9 19 1 3.7 NaN NaN
codebooga-34b 65.9 65.9 17.7 1 3.7 NaN NaN
claude-3-sonnet-20240229 64.6 64.6 18.8 1 3.7 NaN NaN
mistral-large-latest 63.4 63.4 18.1 1 3.8 NaN NaN
speechless-starcoder2-15b 63.4 63.4 16.8 1 3.8 NaN NaN
deepseek-coder-1.3b-instruct 61.6 61.6 16.2 1 3.8 NaN NaN
bigcode--starcoder2-15b-instruct-v0.1 61 61 15.8 1 3.8 NaN NaN
Qwen--Qwen1.5-72B-Chat 59.8 59.8 15.8 1 3.8 NaN NaN
microsoft--Phi-3-mini-4k-instruct 59.8 59.8 16 1 3.8 NaN NaN
code-13b 53.7 53.7 13.1 1 3.9 NaN NaN
codegemma-7b-it 53 53 11.7 1 3.9 NaN NaN
speechless-coding-7b-16k-tora 52.4 52.4 12.2 1 3.9 NaN NaN
speechless-starcoder2-7b 51.8 51.8 11.8 1 3.9 NaN NaN
wizardcoder-15b 50.6 50.6 11 1 3.9 NaN NaN
open-hermes-2.5-code-290k-13b 50.6 50.6 10.8 1 3.9 NaN NaN
code-33b 50 50 11.8 1 3.9 NaN NaN
phi-2 45.7 45.7 10.6 1 3.9 NaN NaN
wizardcoder-7b 45.7 45.7 9.72 1 3.9 NaN NaN
code-llama-multi-34b 43.9 43.9 8.78 1 3.9 NaN NaN
deepseek-coder-33b 43.9 43.9 10.6 1 3.9 NaN NaN
mistral-7b-codealpaca 43.3 43.3 9.46 1 3.9 NaN NaN
starcoder2-15b-oci 43.3 43.3 8.89 1 3.9 NaN NaN
speechless-mistral-7b 42.7 42.7 7.81 1 3.9 NaN NaN
codegemma-7b 42.1 42.1 11.4 1 3.9 NaN NaN
mixtral-8x7b-instruct 40.9 40.9 9.08 1 3.8 NaN NaN
solar-10.7b-instruct 37.8 37.8 7.08 1 3.8 NaN NaN
mistralai--Mistral-7B-Instruct-v0.2 36.6 36.6 6.99 1 3.8 NaN NaN
gemma-1.1-7b-it 36 36 5.83 1 3.7 NaN NaN
code-llama-multi-13b 34.8 34.8 6.12 1 3.7 NaN NaN
octocoder 33.5 33.5 6.44 1 3.7 NaN NaN
xdan-l1-chat 32.9 32.9 5.92 1 3.7 NaN NaN
python-code-13b 31.7 31.7 5.74 1 3.6 NaN NaN