docs: compact README + viewer config (eval_edges)

feat: add data/eval_edges.jsonl for the dataset viewer

docs: add ODbL v1.0 LICENSE

chore: relicense dataset under ODbL v1.0 (was CC-BY-4.0)

Remove old top-level cell dirs (batch 4)

Remove old top-level cell dirs (batch 3)

Remove old top-level cell dirs (batch 2)

Remove old top-level cell dirs (batch 1)

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Move agent results under agent_runs/ (batch 5)

Move agent results under agent_runs/ (batch 4)

Move agent results under agent_runs/ (batch 3)

Move agent results under agent_runs/ (batch 2)

Move agent results under agent_runs/ (batch 1)

Drop agent_runs_summary.json and batch_summary.json — keep only bench.json

Rename all_results_summary.json -> agent_runs_summary.json (distinguish from bench spec)

Add verification_bench benchmark spec (263 ground-truth (model, dataset, metric) tuples)

Add verification_bench summary JSON (263 cells × model/dataset/metric/results)

Move skills_multiagent contents directly under verification_bench/

Remove nested skills_multiagent_gpt-5.2_metadatatool/ (batch 3)

Remove nested skills_multiagent_gpt-5.2_metadatatool/ (batch 2)

Remove nested skills_multiagent_gpt-5.2_metadatatool/ (batch 1)

Add verification_bench section to README

Add verification_bench: skills_multiagent_gpt-5.2 (263 agent eval reproductions)

Update NLI case study README: drop scripts section, note reproducibility caveat

Remove partial run_eval.py (119 cells) and outdated shared/ loaders — keep only results+predictions

Add shared dataset_loaders + model_loaders for NLI case study

Add NLI case study section

Add NLI case study: 576 raw evals + aggregate + scripts + figures

Upload README.md with huggingface_hub

Add full split

Upload README.md with huggingface_hub

Add inductive split

Add transductive split

initial commit