Download checkpoints/ablation_preprocessing_no_adaptive_pooling/meta.json from rocky250/Braintotext: direct link, hf CLI and curl.
- Browser
- Download file 13.8 kB
-
https://huggingface.co/rocky250/Braintotext/resolve/main/checkpoints/ablation_preprocessing_no_adaptive_pooling/meta.json
- Command line
-
hf download hf://rocky250/Braintotext/checkpoints/ablation_preprocessing_no_adaptive_pooling/meta.json
-
curl -L -o meta.json https://huggingface.co/rocky250/Braintotext/resolve/main/checkpoints/ablation_preprocessing_no_adaptive_pooling/meta.json
13.8 kB
Invalid JSON:Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
| {"epochs_done": 10, "history": [{"epoch": 0, "total": 2.344102304180463, "articulatory": 0.3250249839363954, "phoneme": 2.3278510528497205, "semantic": 3.8904200627253607, "encoder_grad_norm": 3.0684217474399467, "phoneme_head_grad_norm": Infinity, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 76.32753656172171, "mean_decoded_phoneme_len": 16.92624659291326, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.6279484260158918, "learning_rate": 0.0009935547237482068, "artic_target_positive_rate": 0.12211586766135998, "artic_pred_positive_rate": 0.15161320740977924, "weight_articulatory": 0.05, "weight_phoneme": 1.0, "weight_semantic": 0.0, "conf_warmup_alpha": 0.0, "gate_mean": 0.5111079263786522, "gate_min": 0.5039503574371338, "gate_max": 0.6817492842674255, "reliability_mean": 0.589669914415331, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999998807907104, "drift_severity_mean": 0.04467889739187631, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.3177183350012171, "val_heuristic_wer": 0.9369172326464833, "val_heuristic_cer": 0.5225953762724764, "val_lexicon_wer": 0.5457580292390929, "val_lexicon_cer": 0.39275520882884596, "val_wer": 0.5457580292390929, "val_cer": 0.39275520882884596}, {"epoch": 1, "total": 1.625613810160221, "articulatory": 0.29659276337195667, "phoneme": 1.2444457140488503, "semantic": 3.5150881027564025, "encoder_grad_norm": 2.4900145321320264, "phoneme_head_grad_norm": 1.7476189685173524, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 66.73945641041892, "mean_decoded_phoneme_len": 25.116562449895785, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.3715448947597446, "learning_rate": 0.000942845842533964, "artic_target_positive_rate": 0.12211523922399069, "artic_pred_positive_rate": 0.1228452838670749, "weight_articulatory": 0.1, "weight_phoneme": 1.0, "weight_semantic": 0.09999999999999999, "conf_warmup_alpha": 0.3333333333333333, "gate_mean": 0.5116789211385121, "gate_min": 0.5043685436248779, "gate_max": 0.6823399066925049, "reliability_mean": 0.583455979451358, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999997019767761, "drift_severity_mean": 0.04577792228864175, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.23925567305871853, "val_heuristic_wer": 0.9232186025094969, "val_heuristic_cer": 0.4796641613547712, "val_lexicon_wer": 0.41648440197997005, "val_lexicon_cer": 0.301898011606888, "val_wer": 0.41648440197997005, "val_cer": 0.301898011606888}, {"epoch": 2, "total": 1.6627690929632921, "articulatory": 0.2958210267699682, "phoneme": 1.0147994849162223, "semantic": 3.017982197419191, "encoder_grad_norm": 2.8706238061953813, "phoneme_head_grad_norm": 1.7491770181900417, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 65.96493704599438, "mean_decoded_phoneme_len": 25.53759820426487, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.312483979995112, "learning_rate": 0.0008468568807348963, "artic_target_positive_rate": 0.12211310711617653, "artic_pred_positive_rate": 0.12278797473662939, "weight_articulatory": 0.15, "weight_phoneme": 1.0, "weight_semantic": 0.19999999999999998, "conf_warmup_alpha": 0.6666666666666666, "gate_mean": 0.5123269458946288, "gate_min": 0.5049902200698853, "gate_max": 0.6745696067810059, "reliability_mean": 0.5765223054333873, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999988079071045, "drift_severity_mean": 0.04685466948353711, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.20536607794877343, "val_heuristic_wer": 0.9142396684701278, "val_heuristic_cer": 0.46251545999429167, "val_lexicon_wer": 0.355819039944745, "val_lexicon_cer": 0.25989439634668443, "val_wer": 0.355819039944745, "val_cer": 0.25989439634668443}, {"epoch": 3, "total": 1.7150611171355614, "articulatory": 0.29519748802368456, "phoneme": 0.8576487878576303, "semantic": 2.661242654690376, "encoder_grad_norm": 3.642182378127025, "phoneme_head_grad_norm": 1.7007208753854801, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 65.46150666126985, "mean_decoded_phoneme_len": 25.816257816257817, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.26983947209985754, "learning_rate": 0.0007159897317841356, "artic_target_positive_rate": 0.12211517855907097, "artic_pred_positive_rate": 0.12265995555581191, "weight_articulatory": 0.2, "weight_phoneme": 1.0, "weight_semantic": 0.3, "conf_warmup_alpha": 1.0, "gate_mean": 0.5126522614517445, "gate_min": 0.5046866536140442, "gate_max": 0.6855029463768005, "reliability_mean": 0.5715914438488354, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999995231628418, "drift_severity_mean": 0.047822408902543814, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.18510805182159953, "val_heuristic_wer": 0.9082537124438816, "val_heuristic_cer": 0.45435733992959754, "val_lexicon_wer": 0.32116956371589733, "val_lexicon_cer": 0.23330320616497002, "val_wer": 0.32116956371589733, "val_cer": 0.23330320616497002}, {"epoch": 4, "total": 1.4928609165625695, "articulatory": 0.2940583020830766, "phoneme": 0.7181794987657131, "semantic": 2.386232432952294, "encoder_grad_norm": 3.747703869220538, "phoneme_head_grad_norm": 1.5927215592983441, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 65.02404798361367, "mean_decoded_phoneme_len": 26.04954304954305, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.23191006145720944, "learning_rate": 0.0005644258818835341, "artic_target_positive_rate": 0.12211650454272062, "artic_pred_positive_rate": 0.12272265577354492, "weight_articulatory": 0.2, "weight_phoneme": 1.0, "weight_semantic": 0.3, "conf_warmup_alpha": 1.0, "gate_mean": 0.5130180275094266, "gate_min": 0.5051165223121643, "gate_max": 0.6822546124458313, "reliability_mean": 0.5671468124340718, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999996423721313, "drift_severity_mean": 0.04853122873586857, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.17044870581234955, "val_heuristic_wer": 0.9036491308852308, "val_heuristic_cer": 0.4460327276186852, "val_lexicon_wer": 0.29803154138367677, "val_lexicon_cer": 0.21589287413186187, "val_wer": 0.29803154138367677, "val_cer": 0.21589287413186187}, {"epoch": 5, "total": 1.3070450093501653, "articulatory": 0.2928615675904812, "phoneme": 0.6045687474310398, "semantic": 2.1463464067532465, "encoder_grad_norm": Infinity, "phoneme_head_grad_norm": 1.4743319431176551, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 64.8288696799508, "mean_decoded_phoneme_len": 26.20522687189354, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.20021578574025833, "learning_rate": 0.00040858962592526263, "artic_target_positive_rate": 0.12211424843049966, "artic_pred_positive_rate": 0.12279850767973141, "weight_articulatory": 0.2, "weight_phoneme": 1.0, "weight_semantic": 0.3, "conf_warmup_alpha": 1.0, "gate_mean": 0.5134793071768233, "gate_min": 0.505348265171051, "gate_max": 0.6846207976341248, "reliability_mean": 0.5651425534722819, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999996423721313, "drift_severity_mean": 0.048908260482348025, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.15973818732588646, "val_heuristic_wer": 0.900771267411074, "val_heuristic_cer": 0.44070497573970124, "val_lexicon_wer": 0.27869229883734314, "val_lexicon_cer": 0.20152697174388737, "val_wer": 0.27869229883734314, "val_cer": 0.20152697174388737}, {"epoch": 6, "total": 1.138443461442605, "articulatory": 0.2919508832387435, "phoneme": 0.4979444351333838, "semantic": 1.9403627545405657, "encoder_grad_norm": NaN, "phoneme_head_grad_norm": NaN, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 64.39227526046581, "mean_decoded_phoneme_len": 26.34471701138368, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.16885532221818203, "learning_rate": 0.00026536824029105174, "artic_target_positive_rate": 0.12211624505237127, "artic_pred_positive_rate": 0.12272757867781016, "weight_articulatory": 0.2, "weight_phoneme": 1.0, "weight_semantic": 0.3, "conf_warmup_alpha": 1.0, "gate_mean": 0.5134792563738264, "gate_min": 0.5056519508361816, "gate_max": 0.6955140829086304, "reliability_mean": 0.5633915993997984, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999995231628418, "drift_severity_mean": 0.0492999138779722, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.15221918697427853, "val_heuristic_wer": 0.8984689766317486, "val_heuristic_cer": 0.4376843307011702, "val_lexicon_wer": 0.26568435593415446, "val_lexicon_cer": 0.19277423651412806, "val_wer": 0.26568435593415446, "val_cer": 0.19277423651412806}, {"epoch": 7, "total": 1.007253857759329, "articulatory": 0.29130942737444854, "phoneme": 0.4216892274048848, "semantic": 1.7576757522729727, "encoder_grad_norm": 3.569359525656089, "phoneme_head_grad_norm": 1.1898851667459194, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 63.893022123568755, "mean_decoded_phoneme_len": 26.440756774090108, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.14651970982182774, "learning_rate": 0.00015028198442313846, "artic_target_positive_rate": 0.12211496022840342, "artic_pred_positive_rate": 0.12272620305227927, "weight_articulatory": 0.2, "weight_phoneme": 1.0, "weight_semantic": 0.3, "conf_warmup_alpha": 1.0, "gate_mean": 0.5134877474315277, "gate_min": 0.505486786365509, "gate_max": 0.6910261511802673, "reliability_mean": 0.5627308396817553, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999980330467224, "drift_severity_mean": 0.04941032826661566, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.14505179455278175, "val_heuristic_wer": 0.8960515713134569, "val_heuristic_cer": 0.4333793169060984, "val_lexicon_wer": 0.25359732934269597, "val_lexicon_cer": 0.18497288554847302, "val_wer": 0.25359732934269597, "val_cer": 0.18497288554847302}, {"epoch": 8, "total": 0.9156643587809342, "articulatory": 0.29090367593826394, "phoneme": 0.36651882332486985, "semantic": 1.6365492643454136, "encoder_grad_norm": 3.4787762268995626, "phoneme_head_grad_norm": 1.1110958100129396, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 63.582177267993494, "mean_decoded_phoneme_len": 26.501202501202503, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.12857730434730774, "learning_rate": 7.580223989206306e-05, "artic_target_positive_rate": 0.12211636625803433, "artic_pred_positive_rate": 0.1227161255497963, "weight_articulatory": 0.2, "weight_phoneme": 1.0, "weight_semantic": 0.3, "conf_warmup_alpha": 1.0, "gate_mean": 0.5135817939852491, "gate_min": 0.5055266618728638, "gate_max": 0.7021175622940063, "reliability_mean": 0.562100642878447, "reliability_min": 9.999999747378752e-05, "reliability_max": 1.0, "drift_severity_mean": 0.0495502870540044, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.14177913612636248, "val_heuristic_wer": 0.8953608840796593, "val_heuristic_cer": 0.4324992864618019, "val_lexicon_wer": 0.24738114423851731, "val_lexicon_cer": 0.18021596422795166, "val_wer": 0.24738114423851731, "val_cer": 0.18021596422795166}, {"epoch": 9, "total": 0.8601560653020174, "articulatory": 0.2906091904410949, "phoneme": 0.3341045826769028, "semantic": 1.5597654197460566, "encoder_grad_norm": 3.327585264811149, "phoneme_head_grad_norm": 1.0464987884347254, "nan_inf_batch_pct": 0.0, "blank_prediction_pct": 63.43424507768961, "mean_decoded_phoneme_len": 26.53519320185987, "mean_target_phoneme_len": 26.89738656405323, "train_per": 0.118491407316448, "learning_rate": 5.0000042690076865e-05, "artic_target_positive_rate": 0.1221164602213181, "artic_pred_positive_rate": 0.12258932860806966, "weight_articulatory": 0.2, "weight_phoneme": 1.0, "weight_semantic": 0.3, "conf_warmup_alpha": 1.0, "gate_mean": 0.5136146931879907, "gate_min": 0.5055201649665833, "gate_max": 0.7054228782653809, "reliability_mean": 0.5613468259463081, "reliability_min": 9.999999747378752e-05, "reliability_max": 0.9999986886978149, "drift_severity_mean": 0.04974880816702547, "reference_initialized": true, "ctc_infeasible_pct_this_epoch": 0.0, "ctc_infeasible_pct_cumulative": 0.0, "val_per": 0.13950720796256727, "val_heuristic_wer": 0.8955911131575918, "val_heuristic_cer": 0.43119113309865853, "val_lexicon_wer": 0.24565442615402325, "val_lexicon_cer": 0.17933593378365523, "val_wer": 0.24565442615402325, "val_cer": 0.17933593378365523}], "cfg": {"d_model": 256, "n_heads": 8, "graph_dim": 32, "graph_heads": 4, "n_graph_layers": 2, "trunk_layers": 2, "trunk_kind": "gru", "patch_bins": 8, "semantic_embed_dim": 256, "baseline_d_model": 256, "graph_patch_chunk_size": 16, "dropout": 0.1, "sinkhorn_iters": 50, "batch_size": 8, "accumulation_steps": 1, "epochs": 30, "ablation_epochs": 10, "lr": 0.001, "weight_decay": 0.01, "warmup_frac": 0.05, "min_lr_frac": 0.05, "aug_noise_std": 0.3, "aug_offset_std": 0.1, "aug_channel_drop": 0.08, "aug_time_mask_n": 2, "aug_time_mask_max": 10, "ema_decay": 0.999, "use_session_affine": true, "collapse_patience_epochs": 4, "log_every": 100, "cl_lr": 0.0001, "cl_base_epochs": 3, "cl_task_epochs": 3, "cl_adapt_epochs": 8, "cl_lora_rank": 8, "cl_ot_shrink": 0.05, "cl_replay_ratio": 0.5, "cl_ewc_lambda": 100.0}, "pipeline_version": "chgft_v6_strict_cl_ema_affine_ot"} |