Flaglab/esnlir-test
Viewer • Updated • 80.2k • 20
ESNLIR-LLM
Spanish NLI datasets with a causal 'reasoning' label, from ESNLIR. Used to evaluate open LLMs in github.com/Pacolas/NLI-via-LLM.
-
-
Flaglab/esnlir-human-validated
Viewer • Updated • 972 • 23 -
Flaglab/esnlir-fewshot-pool
Viewer • Updated • 144 • 22 -
Flaglab/esnlir-llm-predictions
Viewer • Updated • 326k • 35
Flaglab/esnlir-human-majority
Viewer • Updated • 1.97k • 12Note 1,971 pairs with human majority labels beside the connector-derived ones; the 999 disagreements behind the kappa = 0.33 result.