{ "base": { "arc_easy_test_accuracy_pct": 53.7037037037037, "arc_easy_test_correct": 1276, "arc_easy_validation_accuracy_pct": 52.28070175438597, "blimp_accuracy_pct": 77.70597014925373, "wikitext_byte_ppl": 2.312318507477825 }, "experiment": { "arc_easy_test_accuracy_mean_pct": 60.22727272727273, "arc_easy_test_accuracy_sample_sd_pp": 0.2946127946127959, "arc_easy_test_delta_mean_pp": 6.523569023569024, "blimp_accuracy_mean_pct": 77.35373134328358, "blimp_delta_mean_pp": -0.35223880597014556, "wikitext_byte_ppl_mean": 2.314352502083215, "wikitext_byte_ppl_delta": 0.0020339946053899993, "selection": "epoch with highest validation accuracy; all three seeds selected epoch 3", "test_policy": "one pass per selected checkpoint; test scores did not select epoch, seed, or recipe" }, "seeds": [ { "seed": 45500, "selected_epoch": 3, "validation_accuracy_pct": 59.473684210526315, "arc_easy_test_accuracy_pct": 60.22727272727273, "arc_easy_test_correct": 1431, "delta_vs_base_pp": 6.523569023569024, "blimp_accuracy_pct": 77.36119402985075, "wikitext_byte_ppl": 2.314460052543862, "selected_checkpoint_sha256": "8e62ae4f586db75b8c76396f716676f462fb3b89ec1ed80b34620d1a1e54c939" }, { "seed": 45501, "selected_epoch": 3, "validation_accuracy_pct": 59.29824561403509, "arc_easy_test_accuracy_pct": 60.52188552188552, "arc_easy_test_correct": 1438, "delta_vs_base_pp": 6.81818181818182, "blimp_accuracy_pct": 77.31492537313433, "wikitext_byte_ppl": 2.31431533842126, "selected_checkpoint_sha256": "4aebb0bed57a4829ee03697e9f55ac8a3685763ce661795c5e0a092119b747c6" }, { "seed": 45502, "selected_epoch": 3, "validation_accuracy_pct": 59.29824561403509, "arc_easy_test_accuracy_pct": 59.93265993265993, "arc_easy_test_correct": 1424, "delta_vs_base_pp": 6.228956228956228, "blimp_accuracy_pct": 77.38507462686567, "wikitext_byte_ppl": 2.314282115284523, "selected_checkpoint_sha256": "155d94496f1c312922800f8d18ba394b35863decfd2e648e1bd373ccaff6b338" } ] }