{"version":1,"run_id":"run_40338c73ff","files":[{"name":"launch-instructions-9671ae2b3b79.md","kind":"instructions","sha256":"9671ae2b3b79c49d342b16d2da266b23771992ff92215779aca6c67cf91693e6","size":39030,"url":"/api/runs/run_40338c73ff/evidence/9671ae2b3b79c49d342b16d2da266b23771992ff92215779aca6c67cf91693e6"},{"name":"assemble.py","kind":"script","sha256":"5e5f4578397a6b24a67a28e4838646b2d8c1cb34b59c31d285016d170c05619c","size":3567,"url":"/api/runs/run_40338c73ff/evidence/5e5f4578397a6b24a67a28e4838646b2d8c1cb34b59c31d285016d170c05619c"},{"name":"build_report.py","kind":"script","sha256":"763e7cd89fbe6e08720830f8ac1af91bce821e194ca1c766de072cd9ab568e19","size":6964,"url":"/api/runs/run_40338c73ff/evidence/763e7cd89fbe6e08720830f8ac1af91bce821e194ca1c766de072cd9ab568e19"},{"name":"code-revision.json","kind":"code_revision","sha256":"a05da5568122e85406959517f5ae4b5a7cebff6e4ab11fff71fe092edc22b289","size":1007,"url":"/api/runs/run_40338c73ff/evidence/a05da5568122e85406959517f5ae4b5a7cebff6e4ab11fff71fe092edc22b289"},{"name":"common.py","kind":"script","sha256":"f6918e60cb26a8c0792f0f2f191538c60b8ef5364917d0e9954d735cab23b878","size":4028,"url":"/api/runs/run_40338c73ff/evidence/f6918e60cb26a8c0792f0f2f191538c60b8ef5364917d0e9954d735cab23b878"},{"name":"eval-scores-and-selection.json","kind":"metrics","sha256":"13eedf41edc6ad919f3b1a1c51442f04f7b02ccfceb6549f23108826544fba77","size":6578,"url":"/api/runs/run_40338c73ff/evidence/13eedf41edc6ad919f3b1a1c51442f04f7b02ccfceb6549f23108826544fba77"},{"name":"eval_gsm8k.py","kind":"script","sha256":"c30bf146eab81c6b8cb27d8495c1b49f585b9242b9a98fafbbf4b88ded4843bb","size":5112,"url":"/api/runs/run_40338c73ff/evidence/c30bf146eab81c6b8cb27d8495c1b49f585b9242b9a98fafbbf4b88ded4843bb"},{"name":"final.py","kind":"script","sha256":"dc26cb3f41369b127b3d4dece9287fc520be9a36e829db968d90d40ad678fd2a","size":11575,"url":"/api/runs/run_40338c73ff/evidence/dc26cb3f41369b127b3d4dece9287fc520be9a36e829db968d90d40ad678fd2a"},{"name":"probe-results.json","kind":"metrics","sha256":"fe9d306e6d3ecc36c0b509d73071a5ef7a9e8960a88ff60b57d9cc576e2c408e","size":4315,"url":"/api/runs/run_40338c73ff/evidence/fe9d306e6d3ecc36c0b509d73071a5ef7a9e8960a88ff60b57d9cc576e2c408e"},{"name":"probe.py","kind":"script","sha256":"10ccfc225fb989e84419e17776f25f25e8ed3b53a9847f159b0f741f3b1de9a5","size":3382,"url":"/api/runs/run_40338c73ff/evidence/10ccfc225fb989e84419e17776f25f25e8ed3b53a9847f159b0f741f3b1de9a5"},{"name":"select_lr.py","kind":"script","sha256":"829d732b07dc6fbf887980ff0e7eaa4268bcf6525f6f47d6128ecc666b7098f0","size":1435,"url":"/api/runs/run_40338c73ff/evidence/829d732b07dc6fbf887980ff0e7eaa4268bcf6525f6f47d6128ecc666b7098f0"},{"name":"train_arm.py","kind":"script","sha256":"e5b6586647ce2a34cc254e97cc14b01c6546be4aa08328cb4705008b7cfff466","size":5897,"url":"/api/runs/run_40338c73ff/evidence/e5b6586647ce2a34cc254e97cc14b01c6546be4aa08328cb4705008b7cfff466"},{"name":"effective-config.json","kind":"effective_config","sha256":"b070056890cc7a9c97333e4e987279b1bc0aa1ffe89b07beee38fa127d69eed6","size":1421,"url":"/api/runs/run_40338c73ff/evidence/b070056890cc7a9c97333e4e987279b1bc0aa1ffe89b07beee38fa127d69eed6"}],"checkpoints":[{"arm":"rank1_lr1e-3","uri":"tinker://b1503a4ce8731de97e8dc361cde0a152:train:0/sampler_weights/final","step":40},{"arm":"rank1_lr1e-4","uri":"tinker://36d478cb73c0d4af71be9e84880ce6fc:train:0/sampler_weights/final","step":40},{"arm":"rank1_lr3e-4","uri":"tinker://1293cfb24019e3fc1998a35d984bddf0:train:0/sampler_weights/final","step":40},{"arm":"rank32_lr1e-3","uri":"tinker://fa768cbf5c7fd3a1554d3f4319b2478f:train:0/sampler_weights/final","step":40},{"arm":"rank32_lr3e-4","uri":"tinker://2aa88565-ba3e-4d6b-abc3-c0f9777792da:train:0/sampler_weights/final","step":40},{"arm":"rank32_lr1e-4","uri":"tinker://4d7ec04e-f22a-4ee5-9c17-64cb108c82e2:train:0/sampler_weights/final","step":40},{"arm":"rank1_lr3e-5","uri":"tinker://080365acb1f7ca92af4613520ce33213:train:0/sampler_weights/final","step":40}],"per_arm_metrics":{"rank1_lr1e-4":{"train_steps_completed":40,"train_correct_step0":0.6875,"train_correct_last":0.875,"frac_mixed_step0":1.0,"learning_rate":0.0001,"lora_rank":1,"selection_accuracy":0.82,"final_sampler_checkpoint_step":40},"rank1_lr3e-4":{"train_steps_completed":40,"train_correct_step0":0.703125,"train_correct_last":0.8125,"frac_mixed_step0":1.0,"learning_rate":0.0003,"lora_rank":1,"selection_accuracy":0.82,"final_sampler_checkpoint_step":40},"rank1_lr1e-3":{"train_steps_completed":40,"train_correct_step0":0.6875,"train_correct_last":0.0,"frac_mixed_step0":1.0,"learning_rate":0.001,"lora_rank":1,"selection_accuracy":0.0,"final_sampler_checkpoint_step":40},"rank32_lr1e-4":{"train_steps_completed":40,"train_correct_step0":0.75,"train_correct_last":0.828125,"frac_mixed_step0":1.0,"learning_rate":0.0001,"lora_rank":32,"selection_accuracy":0.815,"final_sampler_checkpoint_step":40},"rank32_lr3e-4":{"train_steps_completed":40,"train_correct_step0":0.65625,"train_correct_last":0.828125,"frac_mixed_step0":1.0,"learning_rate":0.0003,"lora_rank":32,"confirmation_accuracy":0.8105451295799821,"selection_accuracy":0.84,"final_sampler_checkpoint_step":40},"rank32_lr1e-3":{"train_steps_completed":40,"train_correct_step0":0.640625,"train_correct_last":0.0,"frac_mixed_step0":1.0,"learning_rate":0.001,"lora_rank":32,"selection_accuracy":0.0,"final_sampler_checkpoint_step":40},"rank1_lr3e-5":{"train_steps_completed":40,"train_correct_step0":0.75,"train_correct_last":0.671875,"frac_mixed_step0":1.0,"learning_rate":3e-05,"lora_rank":1,"confirmation_accuracy":0.8239499553172476,"selection_accuracy":0.82,"final_sampler_checkpoint_step":40},"base":{"confirmation_accuracy":0.7345844504021448,"selection_accuracy":0.71},"treatment":{"confirmation_accuracy":0.8239499553172476},"baseline":{"confirmation_accuracy":0.8105451295799821}},"missing":["predictions"],"errors":["file 2: invalid file name or kind","file 9: attachment retrieval returned HTTP 401","file 10: attachment retrieval returned HTTP 401","file 11: attachment retrieval returned HTTP 401","file 12: attachment retrieval returned HTTP 401","file 13: attachment retrieval returned HTTP 401","file 14: attachment retrieval returned HTTP 401","file 15: attachment retrieval returned HTTP 401","file 16: attachment retrieval returned HTTP 401","file 17: attachment retrieval returned HTTP 401","file 18: attachment retrieval returned HTTP 401","file 19: attachment retrieval returned HTTP 401","file 20: attachment retrieval returned HTTP 401","file 21: attachment retrieval returned HTTP 401","file 22: attachment retrieval returned HTTP 401","file 25: attachment retrieval returned HTTP 401","file 28: attachment retrieval returned HTTP 401"],"status":"partial"}