Barbados / evidence /evaluation /shared /summary.json
Vykos's picture
Release Barbados looped MoE research prototype
d5147ba verified
Raw History Blame Contribute Delete
40.7 kB
{
"arm": "shared",
"backward_called": false,
"buffers_unchanged": true,
"elapsed_seconds": 740.7935931779211,
"gradients_absent": true,
"load_receipt": {
"arm": "shared",
"backbone_parameters": 110465280,
"baseline": {
"ancestor_receipt_sha256": "8851a1a4690ef7d7b119db914f6ecf3cf4a6877731f49019a14d5f836bce915e",
"artifact_manifest_sha256": "1a72c590e35aa8dd61a0ffa526e2e26d62ab737d517859c1558925129dc911ff",
"backbone_and_buffers_bit_equal": true,
"backbone_parameters": 110465280,
"continuation_from_step": 6000,
"device": "cpu",
"dtype": "float32",
"export_verification_sha256": "5f740ed2ebcfc88b894543049406b37f8dfea6fd638bbc44a289199f1729afe2",
"final_checkpoint_receipt_sha256": "5de189db4e6d7cac839afb4bc421124e523ab1df19c5b3640b11d63cd2823837",
"gate_parameters": 1771776,
"initialized_gate_tensors": [
"model.layers.0.attn.melt_gate.bias",
"model.layers.0.attn.melt_gate.input_weight",
"model.layers.0.attn.melt_gate.memory_weight",
"model.layers.1.attn.melt_gate.bias",
"model.layers.1.attn.melt_gate.input_weight",
"model.layers.1.attn.melt_gate.memory_weight",
"model.layers.2.attn.melt_gate.bias",
"model.layers.2.attn.melt_gate.input_weight",
"model.layers.2.attn.melt_gate.memory_weight",
"model.layers.3.attn.melt_gate.bias",
"model.layers.3.attn.melt_gate.input_weight",
"model.layers.3.attn.melt_gate.memory_weight",
"model.layers.4.attn.melt_gate.bias",
"model.layers.4.attn.melt_gate.input_weight",
"model.layers.4.attn.melt_gate.memory_weight",
"model.layers.5.attn.melt_gate.bias",
"model.layers.5.attn.melt_gate.input_weight",
"model.layers.5.attn.melt_gate.memory_weight"
],
"loader_sha256": "082f1e59d0ab8e0240e5902ba85a9720e3f645836a7322da40fa6b79f496a3a3",
"mode": "completed_scratch",
"optimizer_created": false,
"origin_manifest_sha256": "efb3804850e4fa6fb4b6f88b8f71c55a329b9d57e40368a3df9b52a56cbbd4cf",
"original_origin_manifest_sha256": "42cf14afaae1d23201cc4ca56de3441c9257a7f22ffd882359b1806c76d12c93",
"parent_receipt_sha256": "6476628a108b6a87380726f074fae9b84070ae7873d206bcf2a03f92accf3d9e",
"retrieval_index_sha256": "ab8b46387f547e1a6b162cfba3d04efe94c7633882846d0d416901b7115323e2",
"run_manifest_sha256": "af9c994f4a162f05870bd191f84a4322865366fb34ba72bbd6878e9816e34e17",
"source_integrity_sha256": "9074c5ea1d86c3c20892d099a8aedc2b561d08c07dda3d9887ac3631568d4713",
"source_step": 7217,
"trained_provenance": true,
"trained_run": "/home/moloch/Projects/mistral-application/runs/scratch-cloud-continuation-6000-v1",
"unique_parameters": 112237056
},
"cuda_initialized": false,
"gates_executed": true,
"model_fingerprint": "18c3eac60d7b51d7211cd943eb646bf1cb50d0dacfb587bd182cd9fd9d88b935",
"pair_bundle_manifest_sha256": "809a3836e2ad6a62c0bc95ae95431b4a18ba50bc839d25998a03bfa62eadd915",
"pair_summary_sha256": "306b576f59993d0caa6f08f458cb289bebe805a1eb395191de1d5dfd541fbe3b",
"parameter_tensors": 92,
"source_files": {
"research/probes/evaluate_loom_finish.py": "cfce8bc406b4bfae28a504731d06b6e6c013e66e8361981afdf553939d05ba18",
"research/probes/evaluate_loom_melt_pair_v2.py": "d155251dad16b797d26449e08df9ad5d9b9de993f3b6557a0c99fabb9b5d93d1",
"research/probes/load_loom_melt_v3.py": "082f1e59d0ab8e0240e5902ba85a9720e3f645836a7322da40fa6b79f496a3a3",
"research/probes/loom_melt_cuda.py": "7dad5d3355bbd71c5db6df5a7ad4632f29f7dcc1f139010da21f97b167c73e81",
"research/probes/loom_melt_cuda_v2.py": "a20865b78c47f5d2aadb31f8d573f5623bca419fa3c6c6d60a439ff03b4cd9d7",
"research/probes/loom_melt_inference.py": "07d5f127e1dd8d78416ac2488b77577b98efc27220ee443dad01d6f6d3b52b25",
"research/probes/pack_loom_adaptation.py": "fd8f4c0814777cb677afd134354ad360a46cfb7ebef5f4e53c36f5cf312d4f3a",
"research/probes/prepare_loom_finish_eval.py": "b684178e8ff35bff3322de585b5926122b28460785e54cc20f1121273f250e57",
"research/probes/train_loom_melt.py": "7cf0da22488108b5e5e35410b26a31bb2a33b9d4a3a55efdced42880112f3880",
"research/probes/train_loom_melt_v2.py": "cf52f529563d40edc10723433820e7a21d9b3083a6c0d6f7d835948591e6a609"
},
"storage_parameters": 112237056,
"trained_adaptation_updates": 32
},
"memory": {
"peak_allocated_bytes": 936485888,
"peak_reserved_bytes": 1090519040
},
"mode": "score",
"model_fingerprint": "18c3eac60d7b51d7211cd943eb646bf1cb50d0dacfb587bd182cd9fd9d88b935",
"new_chunks": 10242,
"new_groups": 948,
"new_microbatches": 62,
"new_raw_targets": 992326,
"optimizer_created": false,
"pack_receipt_sha256": "06b0e290a4a02f5565ec4508e68deadbae19b939ca3fa995dd57718315614546",
"precision": {
"attention_backend": "math",
"batch": 1,
"buffers_preserved_fp32": true,
"chunks": 3,
"compute_contract": {
"autocast": false,
"buffers": "float32",
"cast_scope": "one independent unit through backward",
"parameter_views": "bfloat16",
"source_policy": "FSDP2 param_dtype=bfloat16/reduce_dtype=float32"
},
"length": 144,
"live_kv_bytes": 1124352,
"loss_dtype": "torch.float32",
"master_dtype": "torch.float32",
"master_parameter_tensors": 92,
"observed_dtypes": {
"embedding_core": [
"torch.bfloat16"
],
"embedding_source": [
"torch.bfloat16"
],
"gate_input": [
"torch.bfloat16"
],
"key": [
"torch.bfloat16"
],
"loop_memory": [
"torch.bfloat16"
],
"parameter_view": [
"torch.bfloat16"
],
"query": [
"torch.bfloat16"
],
"route_scores": [
"torch.bfloat16"
],
"routed_compute": [
"torch.bfloat16"
],
"routed_input": [
"torch.bfloat16"
],
"router_logits": [
"torch.float32"
],
"router_matmul_input": [
"torch.float32"
],
"router_matmul_weight": [
"torch.float32"
],
"value": [
"torch.bfloat16"
]
},
"precision": "bf16",
"readout_dtypes": [
"torch.bfloat16"
],
"referenced_kv_storage_bytes": 1124352,
"routed_compute_basis": "source_grouped_swiglu_uses_input_dtype",
"router_compute_dtype": "torch.float32",
"routing_adapter": "original_route_no_autocast_with_actual_logits_observation",
"storage_is_total_training_memory": false
},
"primary_ce": 3.785574605887762,
"primary_targets": 1021448,
"raw_ce": 3.786990511118591,
"raw_targets": 1024890,
"rng_preserved": true,
"routing_stats_clear": true,
"rows_file": {
"bytes": 604527,
"path": "unit-losses.jsonl",
"sha256": "9c4a51b4ff1036dd92fb05d5d909d173fd89ca3c1b1bbcec71fb5b169ddf0f37"
},
"runtime": {
"bf16_matmul": true,
"capability": [
8,
6
],
"device": "cuda",
"gpu": "NVIDIA A40",
"memory_bytes": 50897289216
},
"seconds_forward": 714.6343508122955,
"source_files": {
"research/probes/evaluate_loom_finish.py": "cfce8bc406b4bfae28a504731d06b6e6c013e66e8361981afdf553939d05ba18",
"research/probes/evaluate_loom_melt_pair_v2.py": "d155251dad16b797d26449e08df9ad5d9b9de993f3b6557a0c99fabb9b5d93d1",
"research/probes/load_loom_melt_v3.py": "082f1e59d0ab8e0240e5902ba85a9720e3f645836a7322da40fa6b79f496a3a3",
"research/probes/loom_melt_cuda.py": "7dad5d3355bbd71c5db6df5a7ad4632f29f7dcc1f139010da21f97b167c73e81",
"research/probes/loom_melt_cuda_v2.py": "a20865b78c47f5d2aadb31f8d573f5623bca419fa3c6c6d60a439ff03b4cd9d7",
"research/probes/loom_melt_inference.py": "07d5f127e1dd8d78416ac2488b77577b98efc27220ee443dad01d6f6d3b52b25",
"research/probes/pack_loom_adaptation.py": "fd8f4c0814777cb677afd134354ad360a46cfb7ebef5f4e53c36f5cf312d4f3a",
"research/probes/prepare_loom_finish_eval.py": "b684178e8ff35bff3322de585b5926122b28460785e54cc20f1121273f250e57",
"research/probes/train_loom_melt.py": "7cf0da22488108b5e5e35410b26a31bb2a33b9d4a3a55efdced42880112f3880",
"research/probes/train_loom_melt_v2.py": "cf52f529563d40edc10723433820e7a21d9b3083a6c0d6f7d835948591e6a609"
},
"state_unchanged": true,
"status": "complete",
"tasks": {
"accuracy": 0.65625,
"backward_called": false,
"buffers_unchanged": true,
"cases": [
{
"candidates": [
{
"answer_targets": 16,
"loss_sum": 68.5475298166275,
"mean_answer_nll": 4.284220613539219
},
{
"answer_targets": 16,
"loss_sum": 65.64686036109924,
"mean_answer_nll": 4.102928772568703
}
],
"correct": true,
"id": "piqa:validation:513",
"label": 1,
"predicted": 1,
"source_index": 0
},
{
"candidates": [
{
"answer_targets": 36,
"loss_sum": 126.57257795333862,
"mean_answer_nll": 3.515904943148295
},
{
"answer_targets": 36,
"loss_sum": 125.6378527879715,
"mean_answer_nll": 3.4899403552214303
}
],
"correct": true,
"id": "piqa:validation:1634",
"label": 1,
"predicted": 1,
"source_index": 1
},
{
"candidates": [
{
"answer_targets": 6,
"loss_sum": 20.711718440055847,
"mean_answer_nll": 3.4519530733426413
},
{
"answer_targets": 5,
"loss_sum": 25.911905527114868,
"mean_answer_nll": 5.182381105422974
}
],
"correct": false,
"id": "piqa:validation:1089",
"label": 1,
"predicted": 0,
"source_index": 2
},
{
"candidates": [
{
"answer_targets": 21,
"loss_sum": 87.10708074644208,
"mean_answer_nll": 4.147956226021051
},
{
"answer_targets": 21,
"loss_sum": 84.81375490874052,
"mean_answer_nll": 4.038750233749549
}
],
"correct": true,
"id": "piqa:validation:411",
"label": 1,
"predicted": 1,
"source_index": 3
},
{
"candidates": [
{
"answer_targets": 60,
"loss_sum": 257.73825839161873,
"mean_answer_nll": 4.295637639860312
},
{
"answer_targets": 107,
"loss_sum": 442.0394263714552,
"mean_answer_nll": 4.131209592256591
}
],
"correct": true,
"id": "piqa:validation:771",
"label": 1,
"predicted": 1,
"source_index": 4
},
{
"candidates": [
{
"answer_targets": 35,
"loss_sum": 91.77772015333176,
"mean_answer_nll": 2.6222205758094788
},
{
"answer_targets": 40,
"loss_sum": 125.31447923183441,
"mean_answer_nll": 3.13286198079586
}
],
"correct": true,
"id": "piqa:validation:1742",
"label": 0,
"predicted": 0,
"source_index": 5
},
{
"candidates": [
{
"answer_targets": 32,
"loss_sum": 156.62766665406525,
"mean_answer_nll": 4.894614582939539
},
{
"answer_targets": 32,
"loss_sum": 157.38724090717733,
"mean_answer_nll": 4.9183512783492915
}
],
"correct": false,
"id": "piqa:validation:1194",
"label": 1,
"predicted": 0,
"source_index": 6
},
{
"candidates": [
{
"answer_targets": 3,
"loss_sum": 21.68057870864868,
"mean_answer_nll": 7.2268595695495605
},
{
"answer_targets": 5,
"loss_sum": 15.259928695857525,
"mean_answer_nll": 3.051985739171505
}
],
"correct": true,
"id": "piqa:validation:919",
"label": 1,
"predicted": 1,
"source_index": 7
},
{
"candidates": [
{
"answer_targets": 27,
"loss_sum": 131.91240333020687,
"mean_answer_nll": 4.88564456778544
},
{
"answer_targets": 25,
"loss_sum": 128.79841957986355,
"mean_answer_nll": 5.151936783194542
}
],
"correct": true,
"id": "piqa:validation:613",
"label": 0,
"predicted": 0,
"source_index": 8
},
{
"candidates": [
{
"answer_targets": 27,
"loss_sum": 112.20432268828154,
"mean_answer_nll": 4.155715655121538
},
{
"answer_targets": 28,
"loss_sum": 113.8412959203124,
"mean_answer_nll": 4.0657605685825855
}
],
"correct": false,
"id": "piqa:validation:1138",
"label": 0,
"predicted": 1,
"source_index": 9
},
{
"candidates": [
{
"answer_targets": 17,
"loss_sum": 72.6776533126831,
"mean_answer_nll": 4.2751560772166535
},
{
"answer_targets": 17,
"loss_sum": 67.08504620194435,
"mean_answer_nll": 3.9461791883496677
}
],
"correct": false,
"id": "piqa:validation:356",
"label": 0,
"predicted": 1,
"source_index": 10
},
{
"candidates": [
{
"answer_targets": 46,
"loss_sum": 214.16678887605667,
"mean_answer_nll": 4.655799758175145
},
{
"answer_targets": 45,
"loss_sum": 200.59760043025017,
"mean_answer_nll": 4.457724454005559
}
],
"correct": false,
"id": "piqa:validation:208",
"label": 0,
"predicted": 1,
"source_index": 11
},
{
"candidates": [
{
"answer_targets": 62,
"loss_sum": 262.97786244750023,
"mean_answer_nll": 4.241578426572584
},
{
"answer_targets": 62,
"loss_sum": 251.71729150414467,
"mean_answer_nll": 4.059956314582979
}
],
"correct": true,
"id": "piqa:validation:1162",
"label": 1,
"predicted": 1,
"source_index": 12
},
{
"candidates": [
{
"answer_targets": 14,
"loss_sum": 51.7253004014492,
"mean_answer_nll": 3.694664314389229
},
{
"answer_targets": 13,
"loss_sum": 40.81349352002144,
"mean_answer_nll": 3.1394995015401106
}
],
"correct": false,
"id": "piqa:validation:485",
"label": 0,
"predicted": 1,
"source_index": 13
},
{
"candidates": [
{
"answer_targets": 12,
"loss_sum": 51.268762826919556,
"mean_answer_nll": 4.272396902243297
},
{
"answer_targets": 6,
"loss_sum": 28.0722878575325,
"mean_answer_nll": 4.678714642922084
}
],
"correct": true,
"id": "piqa:validation:1250",
"label": 0,
"predicted": 0,
"source_index": 14
},
{
"candidates": [
{
"answer_targets": 14,
"loss_sum": 47.090190559625626,
"mean_answer_nll": 3.363585039973259
},
{
"answer_targets": 14,
"loss_sum": 51.64306902885437,
"mean_answer_nll": 3.688790644918169
}
],
"correct": true,
"id": "piqa:validation:992",
"label": 0,
"predicted": 0,
"source_index": 15
},
{
"candidates": [
{
"answer_targets": 3,
"loss_sum": 30.38958168029785,
"mean_answer_nll": 10.129860560099283
},
{
"answer_targets": 3,
"loss_sum": 26.16910409927368,
"mean_answer_nll": 8.723034699757894
}
],
"correct": true,
"id": "piqa:validation:479",
"label": 1,
"predicted": 1,
"source_index": 16
},
{
"candidates": [
{
"answer_targets": 14,
"loss_sum": 50.62457346916199,
"mean_answer_nll": 3.616040962082999
},
{
"answer_targets": 12,
"loss_sum": 45.61664766073227,
"mean_answer_nll": 3.8013873050610223
}
],
"correct": false,
"id": "piqa:validation:242",
"label": 1,
"predicted": 0,
"source_index": 17
},
{
"candidates": [
{
"answer_targets": 23,
"loss_sum": 84.15969115495682,
"mean_answer_nll": 3.659117006737253
},
{
"answer_targets": 18,
"loss_sum": 84.8076197206974,
"mean_answer_nll": 4.711534428927633
}
],
"correct": true,
"id": "piqa:validation:1215",
"label": 0,
"predicted": 0,
"source_index": 18
},
{
"candidates": [
{
"answer_targets": 132,
"loss_sum": 443.08306596428156,
"mean_answer_nll": 3.3566898936687997
},
{
"answer_targets": 133,
"loss_sum": 450.8020041026175,
"mean_answer_nll": 3.3894887526512596
}
],
"correct": true,
"id": "piqa:validation:569",
"label": 0,
"predicted": 0,
"source_index": 19
},
{
"candidates": [
{
"answer_targets": 5,
"loss_sum": 26.740032196044922,
"mean_answer_nll": 5.3480064392089846
},
{
"answer_targets": 7,
"loss_sum": 41.806596755981445,
"mean_answer_nll": 5.972370965140207
}
],
"correct": true,
"id": "piqa:validation:1655",
"label": 0,
"predicted": 0,
"source_index": 20
},
{
"candidates": [
{
"answer_targets": 23,
"loss_sum": 85.59374976158142,
"mean_answer_nll": 3.7214673809383227
},
{
"answer_targets": 24,
"loss_sum": 89.42527055740356,
"mean_answer_nll": 3.726052939891815
}
],
"correct": false,
"id": "piqa:validation:1495",
"label": 1,
"predicted": 0,
"source_index": 21
},
{
"candidates": [
{
"answer_targets": 16,
"loss_sum": 59.04576522111893,
"mean_answer_nll": 3.690360326319933
},
{
"answer_targets": 19,
"loss_sum": 71.80738918483257,
"mean_answer_nll": 3.7793362728859248
}
],
"correct": false,
"id": "piqa:validation:838",
"label": 1,
"predicted": 0,
"source_index": 22
},
{
"candidates": [
{
"answer_targets": 10,
"loss_sum": 59.85038113594055,
"mean_answer_nll": 5.985038113594055
},
{
"answer_targets": 11,
"loss_sum": 68.15844941139221,
"mean_answer_nll": 6.196222673762929
}
],
"correct": true,
"id": "piqa:validation:1517",
"label": 0,
"predicted": 0,
"source_index": 23
},
{
"candidates": [
{
"answer_targets": 17,
"loss_sum": 64.74342477321625,
"mean_answer_nll": 3.8084367513656616
},
{
"answer_targets": 17,
"loss_sum": 68.32545977830887,
"mean_answer_nll": 4.0191446928416985
}
],
"correct": false,
"id": "piqa:validation:1305",
"label": 1,
"predicted": 0,
"source_index": 24
},
{
"candidates": [
{
"answer_targets": 39,
"loss_sum": 138.43891184031963,
"mean_answer_nll": 3.549715688213324
},
{
"answer_targets": 39,
"loss_sum": 142.69705894589424,
"mean_answer_nll": 3.6588989473306217
}
],
"correct": false,
"id": "piqa:validation:824",
"label": 1,
"predicted": 0,
"source_index": 25
},
{
"candidates": [
{
"answer_targets": 14,
"loss_sum": 59.271815955638885,
"mean_answer_nll": 4.233701139688492
},
{
"answer_targets": 14,
"loss_sum": 58.68761947751045,
"mean_answer_nll": 4.191972819822175
}
],
"correct": false,
"id": "piqa:validation:404",
"label": 0,
"predicted": 1,
"source_index": 26
},
{
"candidates": [
{
"answer_targets": 8,
"loss_sum": 59.27084457874298,
"mean_answer_nll": 7.408855572342873
},
{
"answer_targets": 9,
"loss_sum": 75.27413439750671,
"mean_answer_nll": 8.36379271083408
}
],
"correct": true,
"id": "piqa:validation:376",
"label": 0,
"predicted": 0,
"source_index": 27
},
{
"candidates": [
{
"answer_targets": 145,
"loss_sum": 594.6252560764551,
"mean_answer_nll": 4.100863835010036
},
{
"answer_targets": 144,
"loss_sum": 596.8856801390648,
"mean_answer_nll": 4.145039445410172
}
],
"correct": true,
"id": "piqa:validation:476",
"label": 0,
"predicted": 0,
"source_index": 28
},
{
"candidates": [
{
"answer_targets": 78,
"loss_sum": 504.5745495259762,
"mean_answer_nll": 6.468904481102259
},
{
"answer_targets": 78,
"loss_sum": 503.8303242921829,
"mean_answer_nll": 6.459363131951063
}
],
"correct": true,
"id": "piqa:validation:1353",
"label": 1,
"predicted": 1,
"source_index": 29
},
{
"candidates": [
{
"answer_targets": 38,
"loss_sum": 152.18538855481893,
"mean_answer_nll": 4.004878646179446
},
{
"answer_targets": 27,
"loss_sum": 115.69065392017365,
"mean_answer_nll": 4.284839034080505
}
],
"correct": true,
"id": "piqa:validation:979",
"label": 0,
"predicted": 0,
"source_index": 30
},
{
"candidates": [
{
"answer_targets": 7,
"loss_sum": 41.740835309028625,
"mean_answer_nll": 5.962976472718375
},
{
"answer_targets": 7,
"loss_sum": 45.72141253948212,
"mean_answer_nll": 6.53163036278316
}
],
"correct": false,
"id": "piqa:validation:1209",
"label": 1,
"predicted": 0,
"source_index": 31
},
{
"candidates": [
{
"answer_targets": 3,
"loss_sum": 27.53316593170166,
"mean_answer_nll": 9.177721977233887
},
{
"answer_targets": 3,
"loss_sum": 27.291183471679688,
"mean_answer_nll": 9.097061157226562
}
],
"correct": true,
"id": "piqa:validation:905",
"label": 1,
"predicted": 1,
"source_index": 32
},
{
"candidates": [
{
"answer_targets": 51,
"loss_sum": 203.88678428903222,
"mean_answer_nll": 3.997780084098671
},
{
"answer_targets": 51,
"loss_sum": 179.6838759109378,
"mean_answer_nll": 3.523213253155643
}
],
"correct": true,
"id": "piqa:validation:6",
"label": 1,
"predicted": 1,
"source_index": 33
},
{
"candidates": [
{
"answer_targets": 5,
"loss_sum": 34.197272062301636,
"mean_answer_nll": 6.839454412460327
},
{
"answer_targets": 4,
"loss_sum": 19.360405921936035,
"mean_answer_nll": 4.840101480484009
}
],
"correct": false,
"id": "piqa:validation:1011",
"label": 0,
"predicted": 1,
"source_index": 34
},
{
"candidates": [
{
"answer_targets": 13,
"loss_sum": 49.02774082124233,
"mean_answer_nll": 3.7713646785571027
},
{
"answer_targets": 13,
"loss_sum": 46.08056040108204,
"mean_answer_nll": 3.544658492390926
}
],
"correct": false,
"id": "piqa:validation:1016",
"label": 0,
"predicted": 1,
"source_index": 35
},
{
"candidates": [
{
"answer_targets": 16,
"loss_sum": 79.49679970741272,
"mean_answer_nll": 4.968549981713295
},
{
"answer_targets": 18,
"loss_sum": 78.04774770140648,
"mean_answer_nll": 4.335985983411471
}
],
"correct": true,
"id": "piqa:validation:1796",
"label": 1,
"predicted": 1,
"source_index": 36
},
{
"candidates": [
{
"answer_targets": 7,
"loss_sum": 30.05105686187744,
"mean_answer_nll": 4.2930081231253485
},
{
"answer_targets": 6,
"loss_sum": 30.051896631717682,
"mean_answer_nll": 5.008649438619614
}
],
"correct": false,
"id": "piqa:validation:1606",
"label": 1,
"predicted": 0,
"source_index": 37
},
{
"candidates": [
{
"answer_targets": 31,
"loss_sum": 124.32921782135963,
"mean_answer_nll": 4.010619929721279
},
{
"answer_targets": 30,
"loss_sum": 117.08171704411507,
"mean_answer_nll": 3.902723901470502
}
],
"correct": true,
"id": "piqa:validation:310",
"label": 1,
"predicted": 1,
"source_index": 38
},
{
"candidates": [
{
"answer_targets": 9,
"loss_sum": 32.239086359739304,
"mean_answer_nll": 3.5821207066377005
},
{
"answer_targets": 9,
"loss_sum": 35.75029677152634,
"mean_answer_nll": 3.97225519683626
}
],
"correct": true,
"id": "piqa:validation:1256",
"label": 0,
"predicted": 0,
"source_index": 39
},
{
"candidates": [
{
"answer_targets": 38,
"loss_sum": 152.84790597856045,
"mean_answer_nll": 4.022313315225275
},
{
"answer_targets": 38,
"loss_sum": 151.73591051250696,
"mean_answer_nll": 3.99305027664492
}
],
"correct": true,
"id": "piqa:validation:801",
"label": 1,
"predicted": 1,
"source_index": 40
},
{
"candidates": [
{
"answer_targets": 14,
"loss_sum": 62.53528136014938,
"mean_answer_nll": 4.466805811439242
},
{
"answer_targets": 13,
"loss_sum": 53.655186370015144,
"mean_answer_nll": 4.127322028462704
}
],
"correct": true,
"id": "piqa:validation:608",
"label": 1,
"predicted": 1,
"source_index": 41
},
{
"candidates": [
{
"answer_targets": 8,
"loss_sum": 28.815916180610657,
"mean_answer_nll": 3.601989522576332
},
{
"answer_targets": 9,
"loss_sum": 40.550236999988556,
"mean_answer_nll": 4.505581888887617
}
],
"correct": true,
"id": "piqa:validation:1048",
"label": 0,
"predicted": 0,
"source_index": 42
},
{
"candidates": [
{
"answer_targets": 20,
"loss_sum": 73.49589157104492,
"mean_answer_nll": 3.674794578552246
},
{
"answer_targets": 20,
"loss_sum": 67.57065208256245,
"mean_answer_nll": 3.3785326041281225
}
],
"correct": true,
"id": "piqa:validation:519",
"label": 1,
"predicted": 1,
"source_index": 43
},
{
"candidates": [
{
"answer_targets": 25,
"loss_sum": 92.55776011943817,
"mean_answer_nll": 3.702310404777527
},
{
"answer_targets": 25,
"loss_sum": 90.88231357932091,
"mean_answer_nll": 3.6352925431728362
}
],
"correct": false,
"id": "piqa:validation:1143",
"label": 0,
"predicted": 1,
"source_index": 44
},
{
"candidates": [
{
"answer_targets": 45,
"loss_sum": 157.98248773161322,
"mean_answer_nll": 3.510721949591405
},
{
"answer_targets": 45,
"loss_sum": 181.00169798731804,
"mean_answer_nll": 4.022259955273734
}
],
"correct": true,
"id": "piqa:validation:1594",
"label": 0,
"predicted": 0,
"source_index": 45
},
{
"candidates": [
{
"answer_targets": 39,
"loss_sum": 157.74828431010246,
"mean_answer_nll": 4.04482780282314
},
{
"answer_targets": 37,
"loss_sum": 135.8927779495716,
"mean_answer_nll": 3.6727777824208543
}
],
"correct": false,
"id": "piqa:validation:737",
"label": 0,
"predicted": 1,
"source_index": 46
},
{
"candidates": [
{
"answer_targets": 41,
"loss_sum": 158.3344525396824,
"mean_answer_nll": 3.8618159156020093
},
{
"answer_targets": 42,
"loss_sum": 159.10637986660004,
"mean_answer_nll": 3.788247139680953
}
],
"correct": true,
"id": "piqa:validation:915",
"label": 1,
"predicted": 1,
"source_index": 47
},
{
"candidates": [
{
"answer_targets": 17,
"loss_sum": 60.27305355668068,
"mean_answer_nll": 3.5454737386282753
},
{
"answer_targets": 15,
"loss_sum": 63.66574463248253,
"mean_answer_nll": 4.244382975498835
}
],
"correct": true,
"id": "piqa:validation:1773",
"label": 0,
"predicted": 0,
"source_index": 48
},
{
"candidates": [
{
"answer_targets": 31,
"loss_sum": 244.4357331059873,
"mean_answer_nll": 7.885023648580236
},
{
"answer_targets": 34,
"loss_sum": 262.4472395963967,
"mean_answer_nll": 7.719036458717549
}
],
"correct": true,
"id": "piqa:validation:1223",
"label": 1,
"predicted": 1,
"source_index": 49
},
{
"candidates": [
{
"answer_targets": 6,
"loss_sum": 35.47548735141754,
"mean_answer_nll": 5.912581225236257
},
{
"answer_targets": 7,
"loss_sum": 37.37843841314316,
"mean_answer_nll": 5.339776916163308
}
],
"correct": true,
"id": "piqa:validation:297",
"label": 1,
"predicted": 1,
"source_index": 50
},
{
"candidates": [
{
"answer_targets": 56,
"loss_sum": 206.1907831672579,
"mean_answer_nll": 3.6819782708438913
},
{
"answer_targets": 62,
"loss_sum": 221.96145190298557,
"mean_answer_nll": 3.58002341779009
}
],
"correct": true,
"id": "piqa:validation:1202",
"label": 1,
"predicted": 1,
"source_index": 51
},
{
"candidates": [
{
"answer_targets": 8,
"loss_sum": 48.21444845199585,
"mean_answer_nll": 6.026806056499481
},
{
"answer_targets": 9,
"loss_sum": 54.28672158718109,
"mean_answer_nll": 6.031857954131232
}
],
"correct": false,
"id": "piqa:validation:489",
"label": 1,
"predicted": 0,
"source_index": 52
},
{
"candidates": [
{
"answer_targets": 9,
"loss_sum": 31.84696339070797,
"mean_answer_nll": 3.538551487856441
},
{
"answer_targets": 8,
"loss_sum": 21.101733058691025,
"mean_answer_nll": 2.637716632336378
}
],
"correct": false,
"id": "piqa:validation:335",
"label": 0,
"predicted": 1,
"source_index": 53
},
{
"candidates": [
{
"answer_targets": 15,
"loss_sum": 66.23619198799133,
"mean_answer_nll": 4.415746132532756
},
{
"answer_targets": 9,
"loss_sum": 38.3056583404541,
"mean_answer_nll": 4.256184260050456
}
],
"correct": true,
"id": "piqa:validation:1230",
"label": 1,
"predicted": 1,
"source_index": 54
},
{
"candidates": [
{
"answer_targets": 25,
"loss_sum": 93.76409621536732,
"mean_answer_nll": 3.750563848614693
},
{
"answer_targets": 25,
"loss_sum": 97.005739107728,
"mean_answer_nll": 3.8802295643091202
}
],
"correct": true,
"id": "piqa:validation:40",
"label": 0,
"predicted": 0,
"source_index": 55
},
{
"candidates": [
{
"answer_targets": 4,
"loss_sum": 26.139243125915527,
"mean_answer_nll": 6.534810781478882
},
{
"answer_targets": 4,
"loss_sum": 29.49397325515747,
"mean_answer_nll": 7.373493313789368
}
],
"correct": false,
"id": "piqa:validation:848",
"label": 1,
"predicted": 0,
"source_index": 56
},
{
"candidates": [
{
"answer_targets": 26,
"loss_sum": 87.83841823041439,
"mean_answer_nll": 3.378400701169784
},
{
"answer_targets": 13,
"loss_sum": 58.2717467546463,
"mean_answer_nll": 4.482442058049715
}
],
"correct": false,
"id": "piqa:validation:255",
"label": 1,
"predicted": 0,
"source_index": 57
},
{
"candidates": [
{
"answer_targets": 6,
"loss_sum": 31.52444678544998,
"mean_answer_nll": 5.254074464241664
},
{
"answer_targets": 6,
"loss_sum": 36.8781356215477,
"mean_answer_nll": 6.146355936924617
}
],
"correct": true,
"id": "piqa:validation:433",
"label": 0,
"predicted": 0,
"source_index": 58
},
{
"candidates": [
{
"answer_targets": 4,
"loss_sum": 22.5718355178833,
"mean_answer_nll": 5.642958879470825
},
{
"answer_targets": 4,
"loss_sum": 18.861961603164673,
"mean_answer_nll": 4.715490400791168
}
],
"correct": true,
"id": "piqa:validation:370",
"label": 1,
"predicted": 1,
"source_index": 59
},
{
"candidates": [
{
"answer_targets": 41,
"loss_sum": 195.100503385067,
"mean_answer_nll": 4.7585488630504145
},
{
"answer_targets": 42,
"loss_sum": 198.83656460046768,
"mean_answer_nll": 4.734203919058754
}
],
"correct": true,
"id": "piqa:validation:1403",
"label": 1,
"predicted": 1,
"source_index": 60
},
{
"candidates": [
{
"answer_targets": 13,
"loss_sum": 40.40865534543991,
"mean_answer_nll": 3.1083581034953776
},
{
"answer_targets": 13,
"loss_sum": 31.075763165950775,
"mean_answer_nll": 2.3904433204577518
}
],
"correct": true,
"id": "piqa:validation:861",
"label": 1,
"predicted": 1,
"source_index": 61
},
{
"candidates": [
{
"answer_targets": 6,
"loss_sum": 46.91595816612244,
"mean_answer_nll": 7.819326361020406
},
{
"answer_targets": 8,
"loss_sum": 48.481138467788696,
"mean_answer_nll": 6.060142308473587
}
],
"correct": true,
"id": "piqa:validation:161",
"label": 1,
"predicted": 1,
"source_index": 62
},
{
"candidates": [
{
"answer_targets": 29,
"loss_sum": 99.11024868488312,
"mean_answer_nll": 3.417594782237349
},
{
"answer_targets": 29,
"loss_sum": 98.31966036558151,
"mean_answer_nll": 3.390333116054535
}
],
"correct": true,
"id": "piqa:validation:1785",
"label": 1,
"predicted": 1,
"source_index": 63
}
],
"correct": 42,
"count": 64,
"optimizer_created": false,
"protocol": {
"answer_prefix": " ",
"eos": false,
"maximum_input_tokens": 1023,
"prompt": "Goal: {goal}\nSolution:",
"prompt_supervised": false,
"scoring": "mean answer-token negative log-likelihood",
"selection": "first 64 saved development rows, original order",
"tie_break": 0,
"tokenization": "prompt and complete candidate separately, then concatenate"
},
"protocol_sha256": "52820e53fdd0e9d42fcc8b35fa538bb8379a807ce7510925e1234bafb321f2e7",
"rng_preserved": true,
"routing_stats_clear": true,
"scope": "Fixed 64-case development diagnostic, not a complete PIQA reproduction",
"seconds": 11.764678032021038,
"source_sha256": "15a6e4494731c4e7cff583391674f1ca2a80a3249ab4757ed0c1dee88c4eaa73",
"state_unchanged": true,
"status": "complete",
"task": "piqa-development64-v1"
},
"total_microbatches": 64,
"unit_rows": 1492
}