diff options
Diffstat (limited to 'results/crossover_81_audit.json')
| -rw-r--r-- | results/crossover_81_audit.json | 2740 |
1 files changed, 2740 insertions, 0 deletions
diff --git a/results/crossover_81_audit.json b/results/crossover_81_audit.json new file mode 100644 index 0000000..b347e5c --- /dev/null +++ b/results/crossover_81_audit.json @@ -0,0 +1,2740 @@ +{ + "audit_status": "passed", + "complete_grid": true, + "failed_or_incomplete_cells": [ + "vgglike::pepita", + "vgg16::dfa", + "vgg16::pepita", + "resnet56::dualprop" + ], + "failure_retaining": true, + "inputs": { + "plain_cnn": { + "path": "/orion/u/oscarwan/scratch/sdil-work/dualprop-sdil-a6000-v2/runs/plain-p2-audit.json", + "sha256": "1644086fab211ba34890520813ab9e8e64eb31b48e76c4d267b383e228542251" + }, + "resnet": { + "path": "/orion/u/oscarwan/scratch/sdil-work/sdil-a6000-full-rerun/results/resnet_crossover/r2_audit.json", + "sha256": "6d220790b88e342c56ec190970f5f74b2598f4e4356d3612d353ce0bd41de2be" + }, + "transformer": { + "path": "/orion/u/oscarwan/sdil/results/transformer_crossover/t2_audit.json", + "sha256": "69269a464f70c7b6661727fee35a2e1e6104153a42fe308c1c8d66ff00cabe7f" + } + }, + "num_architecture_scale_points": 9, + "num_audited_cells": 81, + "num_expected_cells": 81, + "num_finite_cells": 77, + "num_methods": 9, + "records": [ + { + "architecture": "transformer4", + "cell_id": "transformer4::bp", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5384794776119403, + "final_validation_nll": 1.5438438168075683, + "final_validation_perplexity": 4.682554605158508, + "finite": true, + "method": "bp", + "peak_memory_mib": 104.576171875, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5438438168075683, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 52.463932275772095, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 8905260908544, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer4", + "cell_id": "transformer4::fa", + "family": "decoder_transformer", + "final_validation_accuracy": 0.42518477324913895, + "final_validation_nll": 1.9627436496633888, + "final_validation_perplexity": 7.118831876418274, + "finite": true, + "method": "fa", + "peak_memory_mib": 139.859375, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.9627436496633888, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 74.94509363174438, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 8905260908544, + "feedback_parameter_count": 794752, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer4", + "cell_id": "transformer4::dfa", + "family": "decoder_transformer", + "final_validation_accuracy": 0.27500717566016075, + "final_validation_nll": 2.5794334192637325, + "final_validation_perplexity": 13.189663033234519, + "finite": true, + "method": "dfa", + "peak_memory_mib": 63.1943359375, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 2.5794334192637325, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 69.4969162940979, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 8905260908544, + "feedback_parameter_count": 49920, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 40960000, + "local_head_token_evaluations": 10240000, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer4", + "cell_id": "transformer4::pepita", + "family": "decoder_transformer", + "final_validation_accuracy": 0.146894733065442, + "final_validation_nll": 3.506126876813536, + "final_validation_perplexity": 33.318969078984956, + "finite": true, + "method": "pepita", + "peak_memory_mib": 71.05078125, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 3.506126876813536, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 79.91580104827881, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 17714610028544, + "feedback_parameter_count": 8320, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 20591488, + "local_block_token_evaluations": 40960000, + "local_head_token_evaluations": 10240000, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 20480000 + } + }, + { + "architecture": "transformer4", + "cell_id": "transformer4::ff", + "family": "decoder_transformer", + "final_validation_accuracy": 0.14900258323765786, + "final_validation_nll": 15.273328165783537, + "final_validation_perplexity": 4296572.482773151, + "finite": true, + "method": "ff", + "peak_memory_mib": 83.71142578125, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 15.273328165783537, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 54.61257076263428, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 7246720, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 23852964495360, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 27726720, + "local_block_token_evaluations": 6832128, + "local_head_token_evaluations": 1699840, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 20480000 + } + }, + { + "architecture": "transformer4", + "cell_id": "transformer4::ep", + "family": "decoder_transformer", + "final_validation_accuracy": 0.05020271239954076, + "final_validation_nll": 4.163535973234647, + "final_validation_perplexity": 64.29847880802028, + "finite": true, + "method": "ep", + "peak_memory_mib": 84.38525390625, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 4.163535973234647, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 1218.220811843872, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 17714610028544, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 20591488, + "local_block_token_evaluations": 2065838080, + "local_head_token_evaluations": 516459520, + "local_vjp_token_evaluations": 1228800000, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 247989760, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer4", + "cell_id": "transformer4::dualprop", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5384884471871412, + "final_validation_nll": 1.5438450912931072, + "final_validation_perplexity": 4.68256057301044, + "finite": true, + "method": "dualprop", + "peak_memory_mib": 92.76220703125, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5438450912931072, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 817.4134051799774, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 8905260908544, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 1351680000, + "local_head_token_evaluations": 337920000, + "local_vjp_token_evaluations": 819200000, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 163840000, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer4", + "cell_id": "transformer4::clean_kp", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5444621842709529, + "final_validation_nll": 1.5314903817685956, + "final_validation_perplexity": 4.62506480042341, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 146.45458984375, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5314903817685956, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 86.56656241416931, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 8905260908544, + "feedback_parameter_count": 794752, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer4", + "cell_id": "transformer4::sdil", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5310795780711826, + "final_validation_nll": 1.5739281793412603, + "final_validation_perplexity": 4.8255667032327025, + "finite": true, + "method": "sdil", + "peak_memory_mib": 150.45556640625, + "scale_index": 0, + "size": { + "blocks": 4, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5739281793412603, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 100.5429425239563, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 8905260908544, + "feedback_parameter_count": 794752, + "forward_affine_macs_per_token": 794752, + "forward_attention_macs_per_token": 65536, + "forward_parameter_count": 813568, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::bp", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5412331371986223, + "final_validation_nll": 1.5344256500699627, + "final_validation_perplexity": 4.638660550349236, + "finite": true, + "method": "bp", + "peak_memory_mib": 185.791015625, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5344256500699627, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 92.79100942611694, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 17724397436928, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::fa", + "family": "decoder_transformer", + "final_validation_accuracy": 0.41869977037887485, + "final_validation_nll": 1.9878737879949104, + "final_validation_perplexity": 7.29999591083727, + "finite": true, + "method": "fa", + "peak_memory_mib": 244.07421875, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.9878737879949104, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 138.1208574771881, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 17724397436928, + "feedback_parameter_count": 1581184, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::dfa", + "family": "decoder_transformer", + "final_validation_accuracy": 0.2835820895522388, + "final_validation_nll": 2.5325543089383777, + "final_validation_perplexity": 12.585612657130635, + "finite": true, + "method": "dfa", + "peak_memory_mib": 79.4033203125, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 2.5325543089383777, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 128.47220158576965, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 17724397436928, + "feedback_parameter_count": 83200, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 81920000, + "local_head_token_evaluations": 10240000, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::pepita", + "family": "decoder_transformer", + "final_validation_accuracy": 0.08782111079219289, + "final_validation_nll": 3.4281133570709406, + "final_validation_perplexity": 30.818444466912283, + "finite": true, + "method": "pepita", + "peak_memory_mib": 91.1328125, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 3.4281133570709406, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 146.5909140110016, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 35257898876928, + "feedback_parameter_count": 8320, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 20591488, + "local_block_token_evaluations": 81920000, + "local_head_token_evaluations": 10240000, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 20480000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::ff", + "family": "decoder_transformer", + "final_validation_accuracy": 0.14900258323765786, + "final_validation_nll": 15.473054809219938, + "final_validation_perplexity": 5246411.1441388475, + "finite": true, + "method": "ff", + "peak_memory_mib": 108.78564453125, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 15.473054809219938, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 85.08061099052429, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 7246720, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 47475242680320, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 27726720, + "local_block_token_evaluations": 8192000, + "local_head_token_evaluations": 1024000, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 20480000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::ep", + "family": "decoder_transformer", + "final_validation_accuracy": 0.14694855051664754, + "final_validation_nll": 4.165920734952977, + "final_validation_perplexity": 64.45199833982316, + "finite": true, + "method": "ep", + "peak_memory_mib": 116.46728515625, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 4.165920734952977, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 2358.1659557819366, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 35257898876928, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 20591488, + "local_block_token_evaluations": 4131676160, + "local_head_token_evaluations": 516459520, + "local_vjp_token_evaluations": 2211840000, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 247989760, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::dualprop", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5412241676234214, + "final_validation_nll": 1.534395503669327, + "final_validation_perplexity": 4.638520713537671, + "finite": true, + "method": "dualprop", + "peak_memory_mib": 117.83642578125, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.534395503669327, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 1584.3499374389648, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 17724397436928, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 2703360000, + "local_head_token_evaluations": 337920000, + "local_vjp_token_evaluations": 1474560000, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 163840000, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::clean_kp", + "family": "decoder_transformer", + "final_validation_accuracy": 0.548310132032147, + "final_validation_nll": 1.5202474867845923, + "final_validation_perplexity": 4.5733569004894195, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 256.66943359375, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5202474867845923, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 160.54460978507996, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 17724397436928, + "feedback_parameter_count": 1581184, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer8", + "cell_id": "transformer8::sdil", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5370353760045924, + "final_validation_nll": 1.5471759986658458, + "final_validation_perplexity": 4.698183753785533, + "finite": true, + "method": "sdil", + "peak_memory_mib": 260.67041015625, + "scale_index": 1, + "size": { + "blocks": 8, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5471759986658458, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 194.04587197303772, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 17724397436928, + "feedback_parameter_count": 1581184, + "forward_affine_macs_per_token": 1581184, + "forward_attention_macs_per_token": 131072, + "forward_parameter_count": 1602048, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::bp", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5420493685419059, + "final_validation_nll": 1.5376900980584793, + "final_validation_perplexity": 4.653827959592091, + "finite": true, + "method": "bp", + "peak_memory_mib": 267.005859375, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5376900980584793, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 128.5215814113617, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 26543533965312, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::fa", + "family": "decoder_transformer", + "final_validation_accuracy": 0.4187535878300804, + "final_validation_nll": 1.9916676314361608, + "final_validation_perplexity": 7.327743554270728, + "finite": true, + "method": "fa", + "peak_memory_mib": 348.2890625, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.9916676314361608, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 205.85945415496826, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 26543533965312, + "feedback_parameter_count": 2367616, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::dfa", + "family": "decoder_transformer", + "final_validation_accuracy": 0.29320644374282434, + "final_validation_nll": 2.49942932172703, + "final_validation_perplexity": 12.175543659471092, + "finite": true, + "method": "dfa", + "peak_memory_mib": 95.6123046875, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 2.49942932172703, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 182.98420119285583, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 26543533965312, + "feedback_parameter_count": 116480, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 122880000, + "local_head_token_evaluations": 10240000, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::pepita", + "family": "decoder_transformer", + "final_validation_accuracy": 0.15772997990815155, + "final_validation_nll": 3.567648601860458, + "final_validation_perplexity": 35.433177610577104, + "finite": true, + "method": "pepita", + "peak_memory_mib": 111.21484375, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 3.567648601860458, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 211.51944947242737, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 52801187725312, + "feedback_parameter_count": 8320, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 20591488, + "local_block_token_evaluations": 122880000, + "local_head_token_evaluations": 10240000, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 20480000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::ff", + "family": "decoder_transformer", + "final_validation_accuracy": 0.14900258323765786, + "final_validation_nll": 15.80519527080549, + "final_validation_perplexity": 7313227.6320203915, + "finite": true, + "method": "ff", + "peak_memory_mib": 133.85986328125, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 15.80519527080549, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 112.26771593093872, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 7246720, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 71097520865280, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 27726720, + "local_block_token_evaluations": 8798208, + "local_head_token_evaluations": 708608, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 20480000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::ep", + "family": "decoder_transformer", + "final_validation_accuracy": 0.1193222588978186, + "final_validation_nll": 4.166560651514205, + "final_validation_perplexity": 64.493255440108, + "finite": true, + "method": "ep", + "peak_memory_mib": 148.54931640625, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 4.166560651514205, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 3415.9928121566772, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 52801187725312, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 20591488, + "local_block_token_evaluations": 6197514240, + "local_head_token_evaluations": 516459520, + "local_vjp_token_evaluations": 3194880000, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 247989760, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::dualprop", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5419327640642939, + "final_validation_nll": 1.5377139749537927, + "final_validation_perplexity": 4.653939079881686, + "finite": true, + "method": "dualprop", + "peak_memory_mib": 144.259765625, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5377139749537927, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 2493.0971586704254, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 26543533965312, + "feedback_parameter_count": 0, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 4055040000, + "local_head_token_evaluations": 337920000, + "local_vjp_token_evaluations": 2129920000, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 163840000, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::clean_kp", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5502027123995408, + "final_validation_nll": 1.511195543028594, + "final_validation_perplexity": 4.5321459321278486, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 366.88427734375, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.511195543028594, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 231.48704957962036, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 26543533965312, + "feedback_parameter_count": 2367616, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "transformer12", + "cell_id": "transformer12::sdil", + "family": "decoder_transformer", + "final_validation_accuracy": 0.5351248564867968, + "final_validation_nll": 1.5582328713172886, + "final_validation_perplexity": 4.750419221534573, + "finite": true, + "method": "sdil", + "peak_memory_mib": 370.88525390625, + "scale_index": 2, + "size": { + "blocks": 12, + "context": 64, + "width": 128 + }, + "status": "completed", + "validation_metric": 1.5582328713172886, + "validation_metric_name": "negative_log_likelihood", + "wall_seconds": 274.1436097621918, + "work": { + "block_affine_macs_per_token": 196608, + "block_attention_macs_per_token": 16384, + "candidate_token_presentations": 0, + "completed_optimizer_steps": 5000, + "enumerated_full_forward_macs": 26543533965312, + "feedback_parameter_count": 2367616, + "forward_affine_macs_per_token": 2367616, + "forward_attention_macs_per_token": 196608, + "forward_parameter_count": 2390528, + "full_forward_token_passes": 10351488, + "local_block_token_evaluations": 0, + "local_head_token_evaluations": 0, + "local_vjp_token_evaluations": 0, + "logical_task_loss_queries": 0, + "ordinary_training_tokens": 10240000, + "ordinary_validation_tokens": 111488, + "relaxation_token_passes": 0, + "training_token_presentations": 10240000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.8340000152587891, + "cell_id": "minicnn::bp", + "family": "plain_cnn", + "final_validation_accuracy": 0.8323999786376953, + "finite": true, + "method": "bp", + "peak_memory_mib": 4392.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": 0.4863418638706207, + "validation_metric": 0.8340000152587891, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 175.2685842514038, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 0, + "num_trainable_layers": 3, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.6927999877929687, + "cell_id": "minicnn::fa", + "family": "plain_cnn", + "final_validation_accuracy": 0.6909999847412109, + "finite": true, + "method": "fa", + "peak_memory_mib": 4392.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": 0.9306934475898743, + "validation_metric": 0.6927999877929687, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 224.17574501037598, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 0, + "num_trainable_layers": 3, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.7113999938964843, + "cell_id": "minicnn::dfa", + "family": "plain_cnn", + "final_validation_accuracy": 0.7094000244140625, + "finite": true, + "method": "dfa", + "peak_memory_mib": 4394.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": 0.8873730301856995, + "validation_metric": 0.7113999938964843, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 220.31079077720642, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 0, + "num_trainable_layers": 3, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.305, + "cell_id": "minicnn::pepita", + "family": "plain_cnn", + "final_validation_accuracy": 0.10100000381469726, + "finite": true, + "method": "pepita", + "peak_memory_mib": 4392.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": 2622863360.0, + "validation_metric": 0.305, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 175.6019115447998, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 14000000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 0, + "num_trainable_layers": 3, + "ordinary_training_examples": 4500000, + "ordinary_validation_examples": 500000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 45000, + "training_example_presentations": 9000000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.156, + "cell_id": "minicnn::ff", + "family": "plain_cnn", + "final_validation_accuracy": 0.156, + "finite": true, + "method": "ff", + "peak_memory_mib": 4394.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.156, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 151.01846766471863, + "work": { + "candidate_label_evaluation_presentations": 50000, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 10850000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 0, + "num_trainable_layers": 3, + "ordinary_training_examples": 1800000, + "ordinary_validation_examples": 5000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 0, + "training_example_presentations": 10800000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.10880000114440919, + "cell_id": "minicnn::ep", + "family": "plain_cnn", + "final_validation_accuracy": 0.10880000114440919, + "finite": true, + "method": "ep", + "peak_memory_mib": 4392.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": 0.17824001610279083, + "validation_metric": 0.10880000114440919, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 544.7784323692322, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 5000000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 0, + "num_trainable_layers": 3, + "ordinary_training_examples": 4500000, + "ordinary_validation_examples": 500000, + "relaxation_example_passes": 118000000, + "task_loss_query_batches": 45000, + "training_example_presentations": 4500000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.8305999755859375, + "cell_id": "minicnn::dualprop", + "family": "plain_cnn", + "final_validation_accuracy": 0.8283999633789062, + "finite": true, + "method": "dualprop", + "peak_memory_mib": 4392.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": 0.49406135082244873, + "validation_metric": 0.8305999755859375, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 591.0971002578735, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 6500000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 0, + "num_trainable_layers": 3, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 93600000, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.825, + "cell_id": "minicnn::clean_kp", + "family": "plain_cnn", + "final_validation_accuracy": 0.8237999725341797, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 4394.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": 0.5079880356788635, + "validation_metric": 0.825, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 268.72046995162964, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 0, + "num_trainable_layers": 3, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "minicnn", + "best_validation_accuracy": 0.825, + "cell_id": "minicnn::sdil", + "family": "plain_cnn", + "final_validation_accuracy": 0.8233999633789062, + "finite": true, + "method": "sdil", + "peak_memory_mib": 4396.0, + "scale_index": 0, + "size": { + "trainable_layers": 3 + }, + "status": "completed", + "validation_loss": 0.508787214756012, + "validation_metric": 0.825, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 301.3666183948517, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 11247616, + "forward_parameter_count": 79690, + "neutral_predictor_observations": 64, + "num_trainable_layers": 3, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.9073999786376953, + "cell_id": "vgglike::bp", + "family": "plain_cnn", + "final_validation_accuracy": 0.9045999908447265, + "finite": true, + "method": "bp", + "peak_memory_mib": 4394.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": 0.3427182734012604, + "validation_metric": 0.9073999786376953, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 377.7440617084503, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 0, + "num_trainable_layers": 5, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.5793999862670899, + "cell_id": "vgglike::fa", + "family": "plain_cnn", + "final_validation_accuracy": 0.5706000137329101, + "finite": true, + "method": "fa", + "peak_memory_mib": 4396.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": 684108632031232.0, + "validation_metric": 0.5793999862670899, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 486.18425488471985, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 0, + "num_trainable_layers": 5, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.5622000122070312, + "cell_id": "vgglike::dfa", + "family": "plain_cnn", + "final_validation_accuracy": 0.56, + "finite": true, + "method": "dfa", + "peak_memory_mib": 4396.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": 4.485233227514139e+23, + "validation_metric": 0.5622000122070312, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 411.71502327919006, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 0, + "num_trainable_layers": 5, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.09399999618530273, + "cell_id": "vgglike::pepita", + "family": "plain_cnn", + "final_validation_accuracy": 0.09399999618530273, + "finite": false, + "method": "pepita", + "peak_memory_mib": 4396.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.09399999618530273, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 34.89489960670471, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 14000000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 0, + "num_trainable_layers": 5, + "ordinary_training_examples": 4500000, + "ordinary_validation_examples": 500000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 45000, + "training_example_presentations": 9000000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.24420000000000003, + "cell_id": "vgglike::ff", + "family": "plain_cnn", + "final_validation_accuracy": 0.24420000000000003, + "finite": true, + "method": "ff", + "peak_memory_mib": 4396.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.24420000000000003, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 354.73135900497437, + "work": { + "candidate_label_evaluation_presentations": 50000, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 18050000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 0, + "num_trainable_layers": 5, + "ordinary_training_examples": 1800000, + "ordinary_validation_examples": 5000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 0, + "training_example_presentations": 18000000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.09399999618530273, + "cell_id": "vgglike::ep", + "family": "plain_cnn", + "final_validation_accuracy": 0.09399999618530273, + "finite": true, + "method": "ep", + "peak_memory_mib": 4396.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": 0.10000000149011612, + "validation_metric": 0.09399999618530273, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 3683.5971179008484, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 5000000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 0, + "num_trainable_layers": 5, + "ordinary_training_examples": 4500000, + "ordinary_validation_examples": 500000, + "relaxation_example_passes": 118000000, + "task_loss_query_batches": 45000, + "training_example_presentations": 4500000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.9091999816894532, + "cell_id": "vgglike::dualprop", + "family": "plain_cnn", + "final_validation_accuracy": 0.9081999969482422, + "finite": true, + "method": "dualprop", + "peak_memory_mib": 4396.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": 0.3318080008029938, + "validation_metric": 0.9091999816894532, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 4057.811935901642, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 6500000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 0, + "num_trainable_layers": 5, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 93600000, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.8959999847412109, + "cell_id": "vgglike::clean_kp", + "family": "plain_cnn", + "final_validation_accuracy": 0.8944000244140625, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 4398.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": 0.42668625712394714, + "validation_metric": 0.8959999847412109, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 672.1137459278107, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 0, + "num_trainable_layers": 5, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgglike", + "best_validation_accuracy": 0.8941999816894531, + "cell_id": "vgglike::sdil", + "family": "plain_cnn", + "final_validation_accuracy": 0.8936000061035156, + "finite": true, + "method": "sdil", + "peak_memory_mib": 4400.0, + "scale_index": 1, + "size": { + "trainable_layers": 5 + }, + "status": "completed", + "validation_loss": 0.42905259132385254, + "validation_metric": 0.8941999816894531, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 771.7490882873535, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 192303104, + "forward_parameter_count": 3859210, + "neutral_predictor_observations": 64, + "num_trainable_layers": 5, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.9263999938964844, + "cell_id": "vgg16::bp", + "family": "plain_cnn", + "final_validation_accuracy": 0.9237999725341797, + "finite": true, + "method": "bp", + "peak_memory_mib": 8490.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": 0.33109450340270996, + "validation_metric": 0.9263999938964844, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 942.0694010257721, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 0, + "num_trainable_layers": 16, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.34380001068115235, + "cell_id": "vgg16::fa", + "family": "plain_cnn", + "final_validation_accuracy": 0.33740001678466797, + "finite": true, + "method": "fa", + "peak_memory_mib": 8496.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": 1.825251817703247, + "validation_metric": 0.34380001068115235, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 986.5610544681549, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 0, + "num_trainable_layers": 16, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.09399999618530273, + "cell_id": "vgg16::dfa", + "family": "plain_cnn", + "final_validation_accuracy": 0.09399999618530273, + "finite": false, + "method": "dfa", + "peak_memory_mib": 8496.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.09399999618530273, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 36.65889382362366, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 0, + "num_trainable_layers": 16, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.14739999771118165, + "cell_id": "vgg16::pepita", + "family": "plain_cnn", + "final_validation_accuracy": 0.09399999618530273, + "finite": false, + "method": "pepita", + "peak_memory_mib": 8496.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.14739999771118165, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 65.2109751701355, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 14000000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 0, + "num_trainable_layers": 16, + "ordinary_training_examples": 4500000, + "ordinary_validation_examples": 500000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 45000, + "training_example_presentations": 9000000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.1778, + "cell_id": "vgg16::ff", + "family": "plain_cnn", + "final_validation_accuracy": 0.1778, + "finite": true, + "method": "ff", + "peak_memory_mib": 8500.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.1778, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1920.0846574306488, + "work": { + "candidate_label_evaluation_presentations": 50000, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 57650000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 0, + "num_trainable_layers": 16, + "ordinary_training_examples": 1800000, + "ordinary_validation_examples": 5000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 0, + "training_example_presentations": 57600000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.09399999618530273, + "cell_id": "vgg16::ep", + "family": "plain_cnn", + "final_validation_accuracy": 0.09399999618530273, + "finite": true, + "method": "ep", + "peak_memory_mib": 8496.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": 0.10000000149011612, + "validation_metric": 0.09399999618530273, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 8148.400481939316, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 5000000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 0, + "num_trainable_layers": 16, + "ordinary_training_examples": 4500000, + "ordinary_validation_examples": 500000, + "relaxation_example_passes": 118000000, + "task_loss_query_batches": 45000, + "training_example_presentations": 4500000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.9291999816894532, + "cell_id": "vgg16::dualprop", + "family": "plain_cnn", + "final_validation_accuracy": 0.9290000152587891, + "finite": true, + "method": "dualprop", + "peak_memory_mib": 8496.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": 0.3212488293647766, + "validation_metric": 0.9291999816894532, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 7957.64118885994, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 6500000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 0, + "num_trainable_layers": 16, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 93600000, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.9059999847412109, + "cell_id": "vgg16::clean_kp", + "family": "plain_cnn", + "final_validation_accuracy": 0.9054000091552734, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 8498.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": 0.5295314192771912, + "validation_metric": 0.9059999847412109, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1477.3218109607697, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 0, + "num_trainable_layers": 16, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "vgg16", + "best_validation_accuracy": 0.9095999908447265, + "cell_id": "vgg16::sdil", + "family": "plain_cnn", + "final_validation_accuracy": 0.909000015258789, + "finite": true, + "method": "sdil", + "peak_memory_mib": 8500.0, + "scale_index": 2, + "size": { + "trainable_layers": 16 + }, + "status": "completed", + "validation_loss": 0.5051072239875793, + "validation_metric": 0.9095999908447265, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1655.812320947647, + "work": { + "candidate_label_evaluation_presentations": 0, + "counting_note": "Explicit example/pass counts; affine correlation work is audited separately and is not inferred from wall time.", + "feedforward_example_passes": 12350000, + "forward_affine_macs_per_example": 332111872, + "forward_parameter_count": 33638218, + "neutral_predictor_observations": 64, + "num_trainable_layers": 16, + "ordinary_training_examples": 5850000, + "ordinary_validation_examples": 650000, + "relaxation_example_passes": 0, + "task_loss_query_batches": 58500, + "training_example_presentations": 5850000 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.9196, + "cell_id": "resnet20::bp", + "family": "residual_cnn", + "final_validation_accuracy": 0.9176, + "finite": true, + "method": "bp", + "peak_memory_mib": 1923.67529296875, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": 0.39629163818359375, + "validation_metric": 0.9196, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1228.3254659175873, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 40551040, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 1094878080000000 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.7066, + "cell_id": "resnet20::fa", + "family": "residual_cnn", + "final_validation_accuracy": 0.6854, + "finite": true, + "method": "fa", + "peak_memory_mib": 1916.57421875, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": 0.9467566040039063, + "validation_metric": 0.7066, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1365.4268357753754, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 40551040, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 1090896768000000 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.4502, + "cell_id": "resnet20::dfa", + "family": "residual_cnn", + "final_validation_accuracy": 0.3492, + "finite": true, + "method": "dfa", + "peak_memory_mib": 1924.89404296875, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": 3.6838392578125, + "validation_metric": 0.4502, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1187.1560876369476, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 40551040, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 746876160000000 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.2782, + "cell_id": "resnet20::pepita", + "family": "residual_cnn", + "final_validation_accuracy": 0.2726, + "finite": true, + "method": "pepita", + "peak_memory_mib": 1915.67578125, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": 1.9752171875, + "validation_metric": 0.2782, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 565.1650466918945, + "work": { + "candidate_label_evaluation_presentations": 0, + "feedforward_example_passes": 14005000, + "forward_macs_per_example": 40551040, + "local_vjp_example_evaluations": 0, + "relaxation_example_passes": 0 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.1174, + "cell_id": "resnet20::ff", + "family": "residual_cnn", + "final_validation_accuracy": 0.1174, + "finite": true, + "method": "ff", + "peak_memory_mib": 5220.77099609375, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.1174, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 3187.272476911545, + "work": { + "candidate_label_evaluation_presentations": 50000, + "feedforward_example_passes": 72000000, + "forward_macs_per_example": 40551040, + "local_vjp_example_evaluations": 0, + "relaxation_example_passes": 0 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.1, + "cell_id": "resnet20::ep", + "family": "residual_cnn", + "final_validation_accuracy": 0.1, + "finite": true, + "method": "ep", + "peak_memory_mib": 4621.134765625, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": 0.09437472686767578, + "validation_metric": 0.1, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 15028.455446720123, + "work": { + "candidate_label_evaluation_presentations": 0, + "feedforward_example_passes": 5005000, + "forward_macs_per_example": 40551040, + "local_vjp_example_evaluations": 118100000, + "relaxation_example_passes": 118100000 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.9112, + "cell_id": "resnet20::dualprop", + "family": "residual_cnn", + "final_validation_accuracy": 0.9104, + "finite": true, + "method": "dualprop", + "peak_memory_mib": 1915.55859375, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": 0.43468799438476563, + "validation_metric": 0.9112, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 49289.51700615883, + "work": { + "candidate_label_evaluation_presentations": 0, + "feedforward_example_passes": 10005000, + "forward_macs_per_example": 40551040, + "local_vjp_example_evaluations": 144000000, + "relaxation_example_passes": 144000000 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.9174, + "cell_id": "resnet20::clean_kp", + "family": "residual_cnn", + "final_validation_accuracy": 0.9144, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 1917.59814453125, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": 0.4333634704589844, + "validation_metric": 0.9174, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1784.3792207241058, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 40551040, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 1451874816000000 + } + }, + { + "architecture": "resnet20", + "best_validation_accuracy": 0.9158, + "cell_id": "resnet20::sdil", + "family": "residual_cnn", + "final_validation_accuracy": 0.9132, + "finite": true, + "method": "sdil", + "peak_memory_mib": 1919.763671875, + "scale_index": 0, + "size": { + "base_width": 16, + "depth": 20 + }, + "status": "completed", + "validation_loss": 0.4233872314453125, + "validation_metric": 0.9158, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 3006.6630001068115, + "work": { + "elementwise_operations_estimate": 40698241875968, + "forward_macs_per_example": 40551040, + "total_forward_equivalent_examples": 9000064, + "total_macs_estimate": 1451879978221568 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.928, + "cell_id": "resnet32::bp", + "family": "residual_cnn", + "final_validation_accuracy": 0.928, + "finite": true, + "method": "bp", + "peak_memory_mib": 2222.84375, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": 0.38430252685546873, + "validation_metric": 0.928, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1930.4826476573944, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 68862592, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 1859289984000000 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.6762, + "cell_id": "resnet32::fa", + "family": "residual_cnn", + "final_validation_accuracy": 0.6548, + "finite": true, + "method": "fa", + "peak_memory_mib": 2216.48095703125, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": 0.9980604614257812, + "validation_metric": 0.6762, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 2180.355216741562, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 68862592, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 1855308672000000 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.4546, + "cell_id": "resnet32::dfa", + "family": "residual_cnn", + "final_validation_accuracy": 0.3874, + "finite": true, + "method": "dfa", + "peak_memory_mib": 2229.75, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": 6.95157197265625, + "validation_metric": 0.4546, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1879.2923183441162, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 68862592, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 1266806016000000 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.2682, + "cell_id": "resnet32::pepita", + "family": "residual_cnn", + "final_validation_accuracy": 0.2618, + "finite": true, + "method": "pepita", + "peak_memory_mib": 2214.8359375, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": 2.029934326171875, + "validation_metric": 0.2682, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 909.2571918964386, + "work": { + "candidate_label_evaluation_presentations": 0, + "feedforward_example_passes": 14005000, + "forward_macs_per_example": 68862592, + "local_vjp_example_evaluations": 0, + "relaxation_example_passes": 0 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.1228, + "cell_id": "resnet32::ff", + "family": "residual_cnn", + "final_validation_accuracy": 0.1228, + "finite": true, + "method": "ff", + "peak_memory_mib": 7869.49755859375, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.1228, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 7484.067112207413, + "work": { + "candidate_label_evaluation_presentations": 50000, + "feedforward_example_passes": 115200000, + "forward_macs_per_example": 68862592, + "local_vjp_example_evaluations": 0, + "relaxation_example_passes": 0 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.1, + "cell_id": "resnet32::ep", + "family": "residual_cnn", + "final_validation_accuracy": 0.1, + "finite": true, + "method": "ep", + "peak_memory_mib": 6664.490234375, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": 0.09370457153320312, + "validation_metric": 0.1, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 24671.18036866188, + "work": { + "candidate_label_evaluation_presentations": 0, + "feedforward_example_passes": 5005000, + "forward_macs_per_example": 68862592, + "local_vjp_example_evaluations": 118100000, + "relaxation_example_passes": 118100000 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.871, + "cell_id": "resnet32::dualprop", + "family": "residual_cnn", + "final_validation_accuracy": 0.8654, + "finite": true, + "method": "dualprop", + "peak_memory_mib": 2214.71875, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": 0.6110852416992187, + "validation_metric": 0.871, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 111870.60429954529, + "work": { + "candidate_label_evaluation_presentations": 0, + "feedforward_example_passes": 10005000, + "forward_macs_per_example": 68862592, + "local_vjp_example_evaluations": 144000000, + "relaxation_example_passes": 144000000 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.9262, + "cell_id": "resnet32::clean_kp", + "family": "residual_cnn", + "final_validation_accuracy": 0.9238, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 2218.2431640625, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": 0.39594262084960935, + "validation_metric": 0.9262, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 2880.570287704468, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 68862592, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 2471090688000000 + } + }, + { + "architecture": "resnet32", + "best_validation_accuracy": 0.9254, + "cell_id": "resnet32::sdil", + "family": "residual_cnn", + "final_validation_accuracy": 0.9236, + "finite": true, + "method": "sdil", + "peak_memory_mib": 2221.609375, + "scale_index": 1, + "size": { + "base_width": 16, + "depth": 32 + }, + "status": "completed", + "validation_loss": 0.3822314697265625, + "validation_metric": 0.9254, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 4740.1479415893555, + "work": { + "elementwise_operations_estimate": 65471084756992, + "forward_macs_per_example": 68862592, + "total_forward_equivalent_examples": 9000064, + "total_macs_estimate": 2471099474100224 + } + }, + { + "architecture": "resnet56", + "best_validation_accuracy": 0.9308, + "cell_id": "resnet56::bp", + "family": "residual_cnn", + "final_validation_accuracy": 0.93, + "finite": true, + "method": "bp", + "peak_memory_mib": 3106.6787109375, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "completed", + "validation_loss": 0.3817935302734375, + "validation_metric": 0.9308, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 3326.933441400528, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 125485696, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 3388113792000000 + } + }, + { + "architecture": "resnet56", + "best_validation_accuracy": 0.6674, + "cell_id": "resnet56::fa", + "family": "residual_cnn", + "final_validation_accuracy": 0.6436, + "finite": true, + "method": "fa", + "peak_memory_mib": 3101.94873046875, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "completed", + "validation_loss": 1.0704212158203126, + "validation_metric": 0.6674, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 3838.2244839668274, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 125485696, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 3384132480000000 + } + }, + { + "architecture": "resnet56", + "best_validation_accuracy": 0.4056, + "cell_id": "resnet56::dfa", + "family": "residual_cnn", + "final_validation_accuracy": 0.284, + "finite": true, + "method": "dfa", + "peak_memory_mib": 3125.1162109375, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "completed", + "validation_loss": 56.89802421875, + "validation_metric": 0.4056, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 3246.3810822963715, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 125485696, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 2306665728000000 + } + }, + { + "architecture": "resnet56", + "best_validation_accuracy": 0.2354, + "cell_id": "resnet56::pepita", + "family": "residual_cnn", + "final_validation_accuracy": 0.228, + "finite": true, + "method": "pepita", + "peak_memory_mib": 3098.83544921875, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "completed", + "validation_loss": 2.124036181640625, + "validation_metric": 0.2354, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 1602.2039358615875, + "work": { + "candidate_label_evaluation_presentations": 0, + "feedforward_example_passes": 14005000, + "forward_macs_per_example": 125485696, + "local_vjp_example_evaluations": 0, + "relaxation_example_passes": 0 + } + }, + { + "architecture": "resnet56", + "best_validation_accuracy": 0.1122, + "cell_id": "resnet56::ff", + "family": "residual_cnn", + "final_validation_accuracy": 0.1122, + "finite": true, + "method": "ff", + "peak_memory_mib": 13168.1923828125, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "completed", + "validation_loss": null, + "validation_metric": 0.1122, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 23686.061631202698, + "work": { + "candidate_label_evaluation_presentations": 50000, + "feedforward_example_passes": 201600000, + "forward_macs_per_example": 125485696, + "local_vjp_example_evaluations": 0, + "relaxation_example_passes": 0 + } + }, + { + "architecture": "resnet56", + "best_validation_accuracy": 0.1, + "cell_id": "resnet56::ep", + "family": "residual_cnn", + "final_validation_accuracy": 0.1, + "finite": true, + "method": "ep", + "peak_memory_mib": 11142.107421875, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "completed", + "validation_loss": 0.09450665283203125, + "validation_metric": 0.1, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 50669.00653362274, + "work": { + "candidate_label_evaluation_presentations": 0, + "feedforward_example_passes": 5005000, + "forward_macs_per_example": 125485696, + "local_vjp_example_evaluations": 118100000, + "relaxation_example_passes": 118100000 + } + }, + { + "architecture": "resnet56", + "best_validation_accuracy": null, + "cell_id": "resnet56::dualprop", + "family": "residual_cnn", + "final_validation_accuracy": null, + "finite": false, + "method": "dualprop", + "peak_memory_mib": null, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "timeout", + "validation_loss": null, + "validation_metric": null, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 172800.46599197388, + "work": null + }, + { + "architecture": "resnet56", + "best_validation_accuracy": 0.9308, + "cell_id": "resnet56::clean_kp", + "family": "residual_cnn", + "final_validation_accuracy": 0.9274, + "finite": true, + "method": "clean_kp", + "peak_memory_mib": 3105.1875, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "completed", + "validation_loss": 0.3740764099121094, + "validation_metric": 0.9308, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 5064.757298946381, + "work": { + "elementwise_operations_estimate": 0, + "forward_macs_per_example": 125485696, + "total_forward_equivalent_examples": 9000000, + "total_macs_estimate": 4509522432000000 + } + }, + { + "architecture": "resnet56", + "best_validation_accuracy": 0.935, + "cell_id": "resnet56::sdil", + "family": "residual_cnn", + "final_validation_accuracy": 0.9322, + "finite": true, + "method": "sdil", + "peak_memory_mib": 3111.15185546875, + "scale_index": 2, + "size": { + "base_width": 16, + "depth": 56 + }, + "status": "completed", + "validation_loss": 0.3675199523925781, + "validation_metric": 0.935, + "validation_metric_name": "top1_accuracy", + "wall_seconds": 8350.019015789032, + "work": { + "elementwise_operations_estimate": 115016770519040, + "forward_macs_per_example": 125485696, + "total_forward_equivalent_examples": 9000064, + "total_macs_estimate": 4509538465857536 + } + } + ], + "stage": "crossover_x1_81", + "test_policy": "none" +} |