{ "baseline_metrics": { "arc_easy_validation_acc_norm": 47.36842105263158, "arc_easy_validation_accuracy": 56.666666666666664, "arc_easy_validation_n": 570, "grammar_proxy_accuracy": 75.69444444444444, "grammar_proxy_n": 288, "wiki_byte_perplexity": 2.2606790895633537, "wiki_bytes": 1148001, "wiki_token_perplexity": 30.388754136565744, "wiki_tokens": 274272 }, "baseline_validation_record": "/root/slayer149-balanced/runs/arcft-balanced/baseline.validation.json", "checkpoint": "/root/slayer149-balanced/runs/arcft-balanced/epoch-02.pt", "checkpoint_sha256": "79e4d06b1b95408c4b47e6d9fab34ed2a26609997b80d45698e6ec23db8724e8", "epoch": 2, "frozen": true, "global_step": 1118, "guards": { "max_grammar_proxy_drop_pp": 5.0, "max_wiki_byte_ppl_regression": 0.01 }, "parameters": 149333081, "selected_at_utc": "2026-10-11T12:07:11Z", "selection": "highest ARC-E validation acc_norm among completed epochs; GLINT test metrics not consulted", "validation_deltas": { "arc_easy_acc_norm_pp": 15.08771929824561, "arc_easy_accuracy_pp": 9.122807017543856, "grammar_proxy_pp": -4.861111111111114, "wiki_byte_ppl": 0.004302912611668042 }, "validation_metrics": { "arc_easy_validation_acc_norm": 62.45614035087719, "arc_easy_validation_accuracy": 65.78947368421052, "arc_easy_validation_n": 570, "grammar_proxy_accuracy": 70.83333333333333, "grammar_proxy_n": 288, "wiki_byte_perplexity": 2.2649820021750218, "wiki_bytes": 1148001, "wiki_token_perplexity": 30.631590536210258, "wiki_tokens": 274272 }, "validation_record": "/root/slayer149-balanced/runs/arcft-balanced/epoch-02.validation.json", "validation_record_sha256": "7e663bf39c97ce13dc3cc9f35e0a0a41e494c879f7de27d914413615e8a60bb9" }