Natural Language Inference Distribution Estimation on ChaosNLI 1.0 (test) benchmark leaderboard · SOTA2 Research

Natural Language Inference Distribution Estimation on ChaosNLI 1.0 (test)

0KL Div (Dist)ChaosNLI HJD

Evaluation Results

Method
ChaosNLI HJD2024.12	0	0	0	0.077	0.974	60.9	0.06	0.922	65.3	1
Mixtral MJD + MNLI Label-Guided model explanations2024.12	0.237	0.223	0.253	0.095	1.028	53.3	0.085	0.997	55.9	0.726
Mixtral MJD + VariErr Label-Guided model explanations2024.12	0.238	0.224	0.255	0.097	1.032	51.3	0.089	1.01	55.7	0.642
Mixtral MJD + human explanations2024.12	0.239	0.225	0.257	0.109	1.075	51.9	0.085	0.998	55.7	0.656
Mixtral MJD + Label-Free model explanations2024.12	0.361	0.299	0.343	0.222	1.407	29.6	0.237	1.452	30.2	0.483
Mixtral MJD2024.12	0.437	0.293	0.324	0.129	1.13	43.2	0.125	1.118	47.2	0.593
MNLI dist.2024.12	1.242	0.281	0.295	0.1	1.042	55.5	0.093	1.02	60.7	0.795
VariErr dist.2024.12	3.604	0.282	0.296	0.179	1.279	52.2	0.173	1.261	59.4	0.688