diff --git a/LTA_openwebtext_dualt/logs/eval_lm1b_latest_non_owt_methods_genppl_20260506_20260506_101041.log b/LTA_openwebtext_dualt/logs/eval_lm1b_latest_non_owt_methods_genppl_20260506_20260506_101041.log
new file mode 100644
index 0000000000000000000000000000000000000000..e7d7121634c53fd66eecc26789fed59826a90238
--- /dev/null
+++ b/LTA_openwebtext_dualt/logs/eval_lm1b_latest_non_owt_methods_genppl_20260506_20260506_101041.log
@@ -0,0 +1,90 @@
+[load] mdlm_latest step=860000 ckpt=runs/lm1b_mdlm_unified_ddit_small_len128_gbs512_8gpu_1m_20260505_repro/latest.pt
+[mdlm] generated 32/256
+[mdlm] generated 64/256
+[mdlm] generated 96/256
+[mdlm] generated 128/256
+[mdlm] generated 160/256
+[mdlm] generated 192/256
+[mdlm] generated 224/256
+[mdlm] generated 256/256
+[summary] {"type": "summary", "name": "mdlm_latest", "kind": "mdlm", "checkpoint": "runs/lm1b_mdlm_unified_ddit_small_len128_gbs512_8gpu_1m_20260505_repro/latest.pt", "step": 860000, "decode": {"kind": "mdlm", "steps": 256, "decode_rule": "stochastic_confidence_unmask", "start": "all_mask", "temp": 1.0, "selection_noise": 0.0, "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 13.254623394640433, "nll_per_token": 2.584346427012978, "tokens": 38421, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 15.945599491226918, "nll_per_token": 2.769182897198147, "tokens": 31758, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.5550877572540878, "unique_tokens": 2672, "token_count": 32768, "distinct_1": 0.08154296875, "distinct_2": 0.27294537401574803, "top_token_mass": 0.182861328125}}
+[load] duo_latest step=698000 ckpt=runs/lm1b_duo_unified_ddit_small_len128_gbs512_8gpu_1m_20260505_repro/latest.pt
+[duo discrete] generated 32/256
+[duo discrete] generated 64/256
+[duo discrete] generated 96/256
+[duo discrete] generated 128/256
+[duo discrete] generated 160/256
+[duo discrete] generated 192/256
+[duo discrete] generated 224/256
+[duo discrete] generated 256/256
+[summary] {"type": "summary", "name": "duo_latest", "kind": "duo", "checkpoint": "runs/lm1b_duo_unified_ddit_small_len128_gbs512_8gpu_1m_20260505_repro/latest.pt", "step": 698000, "decode": {"kind": "duo", "steps": 256, "decode_rule": "discrete_random_replace_confidence_lock", "temp": 1.0, "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 1.10513806766837, "nll_per_token": 0.09997027528052237, "tokens": 65280, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 1.0, "nll_per_token": 0.0, "tokens": 0, "kept_samples": 0, "total_samples": 0, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 0.0, "unique_tokens": 1, "token_count": 32768, "distinct_1": 3.0517578125e-05, "distinct_2": 3.075787401574803e-05, "top_token_mass": 1.0}}
+[load] flm_latest step=369000 ckpt=runs/lm1b_flm_unified_ddit_small_len128_gbs512_8gpu_1m_20260506_repro/latest.pt
+[flm] generated 32/256
+[flm] generated 64/256
+[flm] generated 96/256
+[flm] generated 128/256
+[flm] generated 160/256
+[flm] generated 192/256
+[flm] generated 224/256
+[flm] generated 256/256
+[summary] {"type": "summary", "name": "flm_latest", "kind": "flm", "checkpoint": "runs/lm1b_flm_unified_ddit_small_len128_gbs512_8gpu_1m_20260506_repro/latest.pt", "step": 369000, "decode": {"kind": "flm", "steps": 256, "decode_rule": "gaussian_to_endpoint_flow", "noise_init": "gaussian", "final_from": "blend", "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 149.36539937152122, "nll_per_token": 5.006395648630426, "tokens": 38794, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 279.18684360473856, "nll_per_token": 5.63188124801842, "tokens": 32329, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 4.3123598709846584, "unique_tokens": 5896, "token_count": 32768, "distinct_1": 0.179931640625, "distinct_2": 0.6859005905511811, "top_token_mass": 0.045074462890625}}
+[load] categorical_fullvocab_latest step=609000 ckpt=runs/lta_lm1b_dirichlet_categorical_fullvocab_dualt_flmpack_onehot_hardce_ddit_small_len128_gbs512_8gpu_1m_nw0/latest.pt
+[categorical Cmax=1024] generated 32/256
+[categorical Cmax=1024] generated 64/256
+[categorical Cmax=1024] generated 96/256
+[categorical Cmax=1024] generated 128/256
+[categorical Cmax=1024] generated 160/256
+[categorical Cmax=1024] generated 192/256
+[categorical Cmax=1024] generated 224/256
+[categorical Cmax=1024] generated 256/256
+[summary] {"type": "summary", "name": "categorical_fullvocab_latest", "kind": "categorical_fullvocab", "checkpoint": "runs/lta_lm1b_dirichlet_categorical_fullvocab_dualt_flmpack_onehot_hardce_ddit_small_len128_gbs512_8gpu_1m_nw0/latest.pt", "step": 609000, "decode": {"kind": "categorical_fullvocab", "steps": 1024, "model_t_mode": "const05", "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.5, "anchor_mode": "state", "noise_init": "dirichlet", "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "endpoint_temp": 1.3, "final_from": "blend", "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 26.097635854173753, "nll_per_token": 3.2618447299261484, "tokens": 38904, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 37.64924659414174, "nll_per_token": 3.6283129432659162, "tokens": 31619, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 4.01496464321082, "unique_tokens": 2943, "token_count": 32768, "distinct_1": 0.089813232421875, "distinct_2": 0.421751968503937, "top_token_mass": 0.0771484375}}
+[load] categorical_fullvocab_c64_latest step=301000 ckpt=runs/lta_lm1b_dirichlet_categorical_fullvocab_c64p0_dualt_flmpack_onehot_hardce_ddit_small_len128_gbs512_8gpu_1m_nw0/latest.pt
+[categorical Cmax=64] generated 32/256
+[categorical Cmax=64] generated 64/256
+[categorical Cmax=64] generated 96/256
+[categorical Cmax=64] generated 128/256
+[categorical Cmax=64] generated 160/256
+[categorical Cmax=64] generated 192/256
+[categorical Cmax=64] generated 224/256
+[categorical Cmax=64] generated 256/256
+[summary] {"type": "summary", "name": "categorical_fullvocab_c64_latest", "kind": "categorical_fullvocab", "checkpoint": "runs/lta_lm1b_dirichlet_categorical_fullvocab_c64p0_dualt_flmpack_onehot_hardce_ddit_small_len128_gbs512_8gpu_1m_nw0/latest.pt", "step": 301000, "decode": {"kind": "categorical_fullvocab", "steps": 1024, "model_t_mode": "const05", "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.5, "anchor_mode": "state", "noise_init": "dirichlet", "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 64.0, "endpoint_temp": 1.3, "final_from": "blend", "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 29.50899600665449, "nll_per_token": 3.3846951662362814, "tokens": 38725, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 43.028414020337806, "nll_per_token": 3.761860688637853, "tokens": 31666, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.9699503322445215, "unique_tokens": 2842, "token_count": 32768, "distinct_1": 0.08673095703125, "distinct_2": 0.41627706692913385, "top_token_mass": 0.0784912109375}}
+[load] categorical_fullvocab_c256_latest step=264000 ckpt=runs/lta_lm1b_dirichlet_categorical_fullvocab_c256p0_dualt_flmpack_onehot_hardce_ddit_small_len128_gbs512_8gpu_1m_nw0/latest.pt
+[categorical Cmax=256] generated 32/256
+[categorical Cmax=256] generated 64/256
+[categorical Cmax=256] generated 96/256
+[categorical Cmax=256] generated 128/256
+[categorical Cmax=256] generated 160/256
+[categorical Cmax=256] generated 192/256
+[categorical Cmax=256] generated 224/256
+[categorical Cmax=256] generated 256/256
+[summary] {"type": "summary", "name": "categorical_fullvocab_c256_latest", "kind": "categorical_fullvocab", "checkpoint": "runs/lta_lm1b_dirichlet_categorical_fullvocab_c256p0_dualt_flmpack_onehot_hardce_ddit_small_len128_gbs512_8gpu_1m_nw0/latest.pt", "step": 264000, "decode": {"kind": "categorical_fullvocab", "steps": 1024, "model_t_mode": "const05", "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.5, "anchor_mode": "state", "noise_init": "dirichlet", "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 256.0, "endpoint_temp": 1.3, "final_from": "blend", "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 29.0563917263388, "nll_per_token": 3.369238484099026, "tokens": 39236, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 42.82775902266668, "nll_per_token": 3.757186467649641, "tokens": 31964, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.9620396608009667, "unique_tokens": 3100, "token_count": 32768, "distinct_1": 0.0946044921875, "distinct_2": 0.42165969488188976, "top_token_mass": 0.091522216796875}}
+[load] categorical_fullvocab_c16_4gpu_latest step=146000 ckpt=runs/lta_lm1b_dirichlet_categorical_fullvocab_c16p0_dualt_flmpack_onehot_hardce_ddit_small_len128_gbs512_4gpu_1m_nw0/latest.pt
+[categorical Cmax=16] generated 32/256
+[categorical Cmax=16] generated 64/256
+[categorical Cmax=16] generated 96/256
+[categorical Cmax=16] generated 128/256
+[categorical Cmax=16] generated 160/256
+[categorical Cmax=16] generated 192/256
+[categorical Cmax=16] generated 224/256
+[categorical Cmax=16] generated 256/256
+[summary] {"type": "summary", "name": "categorical_fullvocab_c16_4gpu_latest", "kind": "categorical_fullvocab", "checkpoint": "runs/lta_lm1b_dirichlet_categorical_fullvocab_c16p0_dualt_flmpack_onehot_hardce_ddit_small_len128_gbs512_4gpu_1m_nw0/latest.pt", "step": 146000, "decode": {"kind": "categorical_fullvocab", "steps": 1024, "model_t_mode": "const05", "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.5, "anchor_mode": "state", "noise_init": "dirichlet", "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 16.0, "endpoint_temp": 1.3, "final_from": "blend", "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 34.32179659456556, "nll_per_token": 3.535780621576138, "tokens": 38197, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 50.163586453004264, "nll_per_token": 3.915289394027899, "tokens": 31430, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.954715470318628, "unique_tokens": 2829, "token_count": 32768, "distinct_1": 0.086334228515625, "distinct_2": 0.4382689468503937, "top_token_mass": 0.07720947265625}}
+[load] ar_rowshard_latest step=267000 ckpt=runs/ar_lm1b_flmpack_bert_small_len128_gbs512_4gpu_1m_rowshard_b64_resume4000_20260504_203021/latest.pt
+[ar temp=0.8] generated 32/256
+[ar temp=0.8] generated 64/256
+[ar temp=0.8] generated 96/256
+[ar temp=0.8] generated 128/256
+[ar temp=0.8] generated 160/256
+[ar temp=0.8] generated 192/256
+[ar temp=0.8] generated 224/256
+[ar temp=0.8] generated 256/256
+[ar temp=1] generated 32/256
+[ar temp=1] generated 64/256
+[ar temp=1] generated 96/256
+[ar temp=1] generated 128/256
+[ar temp=1] generated 160/256
+[ar temp=1] generated 192/256
+[ar temp=1] generated 224/256
+[ar temp=1] generated 256/256
+[summary] {"type": "summary", "name": "ar_rowshard_latest_t0p8", "kind": "ar", "checkpoint": "runs/ar_lm1b_flmpack_bert_small_len128_gbs512_4gpu_1m_rowshard_b64_resume4000_20260504_203021/latest.pt", "step": 267000, "decode": {"kind": "ar_sample", "temp": 0.8, "max_new_tokens": 127, "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 32.75331929719446, "nll_per_token": 3.489004309531525, "tokens": 38672, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 46.344570351610095, "nll_per_token": 3.8361041406323277, "tokens": 31984, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 4.1883945709687715, "unique_tokens": 5282, "token_count": 32768, "distinct_1": 0.16119384765625, "distinct_2": 0.590089812992126, "top_token_mass": 0.065338134765625}}
+[summary] {"type": "summary", "name": "ar_rowshard_latest_t1p0", "kind": "ar", "checkpoint": "runs/ar_lm1b_flmpack_bert_small_len128_gbs512_4gpu_1m_rowshard_b64_resume4000_20260504_203021/latest.pt", "step": 267000, "decode": {"kind": "ar_sample", "temp": 1.0, "max_new_tokens": 127, "n_samples": 256, "seed": 20260506}, "raw_genppl": {"ppl": 70.47566606515413, "nll_per_token": 4.255267488039646, "tokens": 38787, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 111.28221270824383, "nll_per_token": 4.7120694315677865, "tokens": 32392, "kept_samples": 256, "total_samples": 256, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 4.344022305264332, "unique_tokens": 6845, "token_count": 32768, "distinct_1": 0.208892822265625, "distinct_2": 0.7198572834645669, "top_token_mass": 0.043731689453125}}
+[done] docs/lta_samples/metrics_20260506/latest_non_owt_all_methods
diff --git a/LTA_openwebtext_dualt/logs/lta_lm1b_dirichlet_categorical_fullvocab_c1024_fullycoupled_flmpack_onehot_hardce_ddit_small_len128_allmask_gbs512_4gpu_1m_nw0.log b/LTA_openwebtext_dualt/logs/lta_lm1b_dirichlet_categorical_fullvocab_c1024_fullycoupled_flmpack_onehot_hardce_ddit_small_len128_allmask_gbs512_4gpu_1m_nw0.log
new file mode 100644
index 0000000000000000000000000000000000000000..721573a386b9fe59bfc0026ae0b69af55e4f4a98
--- /dev/null
+++ b/LTA_openwebtext_dualt/logs/lta_lm1b_dirichlet_categorical_fullvocab_c1024_fullycoupled_flmpack_onehot_hardce_ddit_small_len128_allmask_gbs512_4gpu_1m_nw0.log
@@ -0,0 +1,2083 @@
+
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+NCCL version 2.25.1+cuda12.8
+{
+  "device": "cuda:0",
+  "rank": 0,
+  "world_size": 4,
+  "samples": "wrapped_stream",
+  "vocab_size": 30522,
+  "save_dir": "runs/lta_lm1b_dirichlet_categorical_fullvocab_c1024_fullycoupled_flmpack_onehot_hardce_ddit_small_len128_allmask_gbs512_4gpu_1m_nw0",
+  "batch_size": 64,
+  "grad_accum": 2,
+  "effective_batch_size": 512,
+  "global_batch_size": 512,
+  "lr_schedule": "constant_warmup",
+  "warmup_steps": 2500,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.999,
+  "adam_eps": 1e-08,
+  "model_type": "ddit",
+  "dual_t": true,
+  "corrupt_t_mode": "same",
+  "corrupt_min_t": 0.0,
+  "corrupt_max_t": 1.0,
+  "dirichlet_endpoint_mode": "categorical_dual_t",
+  "dirichlet_semantic_t_mode": "same",
+  "dirichlet_semantic_t_value": 0.0,
+  "categorical_wrong_from_full_vocab": true,
+  "simplex_bridge_sampler": "dirichlet",
+  "logistic_normal_sigma_min": 0.18,
+  "logistic_normal_sigma_max": 2.2,
+  "logistic_normal_tau_min": 0.65,
+  "logistic_normal_tau_max": 1.15,
+  "torch_compile": false,
+  "compile_mode": "max-autotune",
+  "state_format": "prob",
+  "target_loss": "hard_ce",
+  "meanflow_weight": 0.0,
+  "bridge_noise_init": "logistic_normal",
+  "noise_sigma": -1.0,
+  "wrap": true,
+  "wrap_mode": "stream",
+  "wrap_record_buffer_size": 200,
+  "owt_cached_chunks": false,
+  "owt_chunk_cache_dir": "",
+  "owt_chunk_cache_rebuild": false,
+  "owt_chunk_cache_write_batch": 4096,
+  "online_chunk_shuffle": false,
+  "online_chunk_shuffle_buffer": 10000,
+  "openwebtext_split": "all",
+  "detokenizer": "auto",
+  "resolved_detokenizer": "lm1b",
+  "num_workers": 0,
+  "latest_every": 1000,
+  "resume_path": ""
+}
+step=100 micro_steps=200 elapsed=35.3s lr=1.212000e-05 loss_all=10.1766 acc_all=0.3556 loss_corrupt=10.1766 acc_corrupt=0.3556 corrupt_frac=1.0000 loss=10.1766 loss_recon=10.1766 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4978 init_acc_corrupt=0.4684 init_gold_top10=0.4968 init_gold_top100=0.5024
+step=200 micro_steps=400 elapsed=34.2s lr=2.412000e-05 loss_all=8.9348 acc_all=0.0801 loss_corrupt=8.9348 acc_corrupt=0.0801 corrupt_frac=1.0000 loss=8.9348 loss_recon=8.9348 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5004 init_acc_corrupt=0.4649 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=300 micro_steps=600 elapsed=34.0s lr=3.612000e-05 loss_all=7.0787 acc_all=0.1137 loss_corrupt=7.0787 acc_corrupt=0.1137 corrupt_frac=1.0000 loss=7.0787 loss_recon=7.0787 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4999 init_acc_corrupt=0.4652 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=400 micro_steps=800 elapsed=33.6s lr=4.812000e-05 loss_all=4.7591 acc_all=0.3843 loss_corrupt=4.7591 acc_corrupt=0.3843 corrupt_frac=1.0000 loss=4.7591 loss_recon=4.7591 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4985 init_acc_corrupt=0.4667 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=500 micro_steps=1000 elapsed=33.8s lr=6.012000e-05 loss_all=3.9909 acc_all=0.4839 loss_corrupt=3.9909 acc_corrupt=0.4839 corrupt_frac=1.0000 loss=3.9909 loss_recon=3.9909 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5019 init_acc_corrupt=0.4631 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=600 micro_steps=1200 elapsed=33.9s lr=7.212000e-05 loss_all=3.7483 acc_all=0.5108 loss_corrupt=3.7483 acc_corrupt=0.5108 corrupt_frac=1.0000 loss=3.7483 loss_recon=3.7483 loss_meanflow=0.0000 mean_model_t=0.5040 mean_corrupt_t=0.5040 wrong_frac=0.4960 init_acc_corrupt=0.4699 init_gold_top10=0.4985 init_gold_top100=0.5041
+step=700 micro_steps=1400 elapsed=33.9s lr=8.412000e-05 loss_all=3.6968 acc_all=0.5112 loss_corrupt=3.6968 acc_corrupt=0.5112 corrupt_frac=1.0000 loss=3.6968 loss_recon=3.6968 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5031 init_acc_corrupt=0.4622 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=800 micro_steps=1600 elapsed=33.8s lr=9.612000e-05 loss_all=3.6405 acc_all=0.5148 loss_corrupt=3.6405 acc_corrupt=0.5148 corrupt_frac=1.0000 loss=3.6405 loss_recon=3.6405 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5039 init_acc_corrupt=0.4609 init_gold_top10=0.4904 init_gold_top100=0.4962
+step=900 micro_steps=1800 elapsed=33.7s lr=1.081200e-04 loss_all=3.5498 acc_all=0.5235 loss_corrupt=3.5498 acc_corrupt=0.5235 corrupt_frac=1.0000 loss=3.5498 loss_recon=3.5498 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5000 init_acc_corrupt=0.4661 init_gold_top10=0.4944 init_gold_top100=0.5000
+step=1000 micro_steps=2000 elapsed=33.5s lr=1.201200e-04 loss_all=3.5017 acc_all=0.5280 loss_corrupt=3.5017 acc_corrupt=0.5280 corrupt_frac=1.0000 loss=3.5017 loss_recon=3.5017 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4982 init_acc_corrupt=0.4675 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=1100 micro_steps=2200 elapsed=34.9s lr=1.321200e-04 loss_all=3.5129 acc_all=0.5246 loss_corrupt=3.5129 acc_corrupt=0.5246 corrupt_frac=1.0000 loss=3.5129 loss_recon=3.5129 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5036 init_acc_corrupt=0.4620 init_gold_top10=0.4908 init_gold_top100=0.4965
+step=1200 micro_steps=2400 elapsed=33.4s lr=1.441200e-04 loss_all=3.4677 acc_all=0.5290 loss_corrupt=3.4677 acc_corrupt=0.5290 corrupt_frac=1.0000 loss=3.4677 loss_recon=3.4677 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5013 init_acc_corrupt=0.4644 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=1300 micro_steps=2600 elapsed=33.4s lr=1.561200e-04 loss_all=3.4382 acc_all=0.5316 loss_corrupt=3.4382 acc_corrupt=0.5316 corrupt_frac=1.0000 loss=3.4382 loss_recon=3.4382 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5000 init_acc_corrupt=0.4648 init_gold_top10=0.4943 init_gold_top100=0.5002
+step=1400 micro_steps=2800 elapsed=33.6s lr=1.681200e-04 loss_all=3.3909 acc_all=0.5368 loss_corrupt=3.3909 acc_corrupt=0.5368 corrupt_frac=1.0000 loss=3.3909 loss_recon=3.3909 loss_meanflow=0.0000 mean_model_t=0.5036 mean_corrupt_t=0.5036 wrong_frac=0.4967 init_acc_corrupt=0.4685 init_gold_top10=0.4977 init_gold_top100=0.5035
+step=1500 micro_steps=3000 elapsed=33.4s lr=1.801200e-04 loss_all=3.4048 acc_all=0.5340 loss_corrupt=3.4048 acc_corrupt=0.5340 corrupt_frac=1.0000 loss=3.4048 loss_recon=3.4048 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5014 init_acc_corrupt=0.4636 init_gold_top10=0.4929 init_gold_top100=0.4988
+step=1600 micro_steps=3200 elapsed=33.6s lr=1.921200e-04 loss_all=3.3910 acc_all=0.5345 loss_corrupt=3.3910 acc_corrupt=0.5345 corrupt_frac=1.0000 loss=3.3910 loss_recon=3.3910 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5019 init_acc_corrupt=0.4633 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=1700 micro_steps=3400 elapsed=33.7s lr=2.041200e-04 loss_all=3.3620 acc_all=0.5370 loss_corrupt=3.3620 acc_corrupt=0.5370 corrupt_frac=1.0000 loss=3.3620 loss_recon=3.3620 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5017 init_acc_corrupt=0.4637 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=1800 micro_steps=3600 elapsed=33.6s lr=2.161200e-04 loss_all=3.3047 acc_all=0.5442 loss_corrupt=3.3047 acc_corrupt=0.5442 corrupt_frac=1.0000 loss=3.3047 loss_recon=3.3047 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4967 init_acc_corrupt=0.4700 init_gold_top10=0.4979 init_gold_top100=0.5035
+step=1900 micro_steps=3800 elapsed=33.7s lr=2.281200e-04 loss_all=3.3264 acc_all=0.5408 loss_corrupt=3.3264 acc_corrupt=0.5408 corrupt_frac=1.0000 loss=3.3264 loss_recon=3.3264 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5000 init_acc_corrupt=0.4653 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=2000 micro_steps=4000 elapsed=33.6s lr=2.401200e-04 loss_all=3.3062 acc_all=0.5428 loss_corrupt=3.3062 acc_corrupt=0.5428 corrupt_frac=1.0000 loss=3.3062 loss_recon=3.3062 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4992 init_acc_corrupt=0.4662 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=2100 micro_steps=4200 elapsed=35.1s lr=2.521200e-04 loss_all=3.3097 acc_all=0.5416 loss_corrupt=3.3097 acc_corrupt=0.5416 corrupt_frac=1.0000 loss=3.3097 loss_recon=3.3097 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5016 init_acc_corrupt=0.4640 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=2200 micro_steps=4400 elapsed=33.5s lr=2.641200e-04 loss_all=3.3189 acc_all=0.5401 loss_corrupt=3.3189 acc_corrupt=0.5401 corrupt_frac=1.0000 loss=3.3189 loss_recon=3.3189 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5037 init_acc_corrupt=0.4606 init_gold_top10=0.4907 init_gold_top100=0.4965
+step=2300 micro_steps=4600 elapsed=33.5s lr=2.761200e-04 loss_all=3.2750 acc_all=0.5453 loss_corrupt=3.2750 acc_corrupt=0.5453 corrupt_frac=1.0000 loss=3.2750 loss_recon=3.2750 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4994 init_acc_corrupt=0.4658 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=2400 micro_steps=4800 elapsed=33.6s lr=2.881200e-04 loss_all=3.2664 acc_all=0.5461 loss_corrupt=3.2664 acc_corrupt=0.5461 corrupt_frac=1.0000 loss=3.2664 loss_recon=3.2664 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4997 init_acc_corrupt=0.4657 init_gold_top10=0.4947 init_gold_top100=0.5005
+step=2500 micro_steps=5000 elapsed=33.4s lr=3.000000e-04 loss_all=3.2715 acc_all=0.5449 loss_corrupt=3.2715 acc_corrupt=0.5449 corrupt_frac=1.0000 loss=3.2715 loss_recon=3.2715 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5015 init_acc_corrupt=0.4636 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=2600 micro_steps=5200 elapsed=33.6s lr=3.000000e-04 loss_all=3.2670 acc_all=0.5451 loss_corrupt=3.2670 acc_corrupt=0.5451 corrupt_frac=1.0000 loss=3.2670 loss_recon=3.2670 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5022 init_acc_corrupt=0.4628 init_gold_top10=0.4922 init_gold_top100=0.4979
+step=2700 micro_steps=5400 elapsed=33.5s lr=3.000000e-04 loss_all=3.2677 acc_all=0.5442 loss_corrupt=3.2677 acc_corrupt=0.5442 corrupt_frac=1.0000 loss=3.2677 loss_recon=3.2677 loss_meanflow=0.0000 mean_model_t=0.4961 mean_corrupt_t=0.4961 wrong_frac=0.5033 init_acc_corrupt=0.4618 init_gold_top10=0.4909 init_gold_top100=0.4968
+step=2800 micro_steps=5600 elapsed=33.7s lr=3.000000e-04 loss_all=3.2503 acc_all=0.5462 loss_corrupt=3.2503 acc_corrupt=0.5462 corrupt_frac=1.0000 loss=3.2503 loss_recon=3.2503 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5030 init_acc_corrupt=0.4624 init_gold_top10=0.4913 init_gold_top100=0.4972
+step=2900 micro_steps=5800 elapsed=33.6s lr=3.000000e-04 loss_all=3.2390 acc_all=0.5477 loss_corrupt=3.2390 acc_corrupt=0.5477 corrupt_frac=1.0000 loss=3.2390 loss_recon=3.2390 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5022 init_acc_corrupt=0.4633 init_gold_top10=0.4921 init_gold_top100=0.4978
+step=3000 micro_steps=6000 elapsed=33.9s lr=3.000000e-04 loss_all=3.2025 acc_all=0.5525 loss_corrupt=3.2025 acc_corrupt=0.5525 corrupt_frac=1.0000 loss=3.2025 loss_recon=3.2025 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4976 init_acc_corrupt=0.4680 init_gold_top10=0.4968 init_gold_top100=0.5026
+step=3100 micro_steps=6200 elapsed=35.4s lr=3.000000e-04 loss_all=3.2080 acc_all=0.5511 loss_corrupt=3.2080 acc_corrupt=0.5511 corrupt_frac=1.0000 loss=3.2080 loss_recon=3.2080 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4997 init_acc_corrupt=0.4659 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=3200 micro_steps=6400 elapsed=33.9s lr=3.000000e-04 loss_all=3.2223 acc_all=0.5489 loss_corrupt=3.2223 acc_corrupt=0.5489 corrupt_frac=1.0000 loss=3.2223 loss_recon=3.2223 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5024 init_acc_corrupt=0.4633 init_gold_top10=0.4918 init_gold_top100=0.4977
+step=3300 micro_steps=6600 elapsed=33.8s lr=3.000000e-04 loss_all=3.2237 acc_all=0.5485 loss_corrupt=3.2237 acc_corrupt=0.5485 corrupt_frac=1.0000 loss=3.2237 loss_recon=3.2237 loss_meanflow=0.0000 mean_model_t=0.4960 mean_corrupt_t=0.4960 wrong_frac=0.5041 init_acc_corrupt=0.4611 init_gold_top10=0.4901 init_gold_top100=0.4960
+step=3400 micro_steps=6800 elapsed=33.6s lr=3.000000e-04 loss_all=3.2116 acc_all=0.5500 loss_corrupt=3.2116 acc_corrupt=0.5500 corrupt_frac=1.0000 loss=3.2116 loss_recon=3.2116 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5024 init_acc_corrupt=0.4632 init_gold_top10=0.4919 init_gold_top100=0.4977
+step=3500 micro_steps=7000 elapsed=33.5s lr=3.000000e-04 loss_all=3.1708 acc_all=0.5550 loss_corrupt=3.1708 acc_corrupt=0.5550 corrupt_frac=1.0000 loss=3.1708 loss_recon=3.1708 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4981 init_acc_corrupt=0.4674 init_gold_top10=0.4963 init_gold_top100=0.5021
+step=3600 micro_steps=7200 elapsed=33.4s lr=3.000000e-04 loss_all=3.1718 acc_all=0.5546 loss_corrupt=3.1718 acc_corrupt=0.5546 corrupt_frac=1.0000 loss=3.1718 loss_recon=3.1718 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4992 init_acc_corrupt=0.4666 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=3700 micro_steps=7400 elapsed=33.4s lr=3.000000e-04 loss_all=3.1891 acc_all=0.5526 loss_corrupt=3.1891 acc_corrupt=0.5526 corrupt_frac=1.0000 loss=3.1891 loss_recon=3.1891 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5014 init_acc_corrupt=0.4636 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=3800 micro_steps=7600 elapsed=33.7s lr=3.000000e-04 loss_all=3.1515 acc_all=0.5570 loss_corrupt=3.1515 acc_corrupt=0.5570 corrupt_frac=1.0000 loss=3.1515 loss_recon=3.1515 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4986 init_acc_corrupt=0.4671 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=3900 micro_steps=7800 elapsed=33.8s lr=3.000000e-04 loss_all=3.2034 acc_all=0.5498 loss_corrupt=3.2034 acc_corrupt=0.5498 corrupt_frac=1.0000 loss=3.2034 loss_recon=3.2034 loss_meanflow=0.0000 mean_model_t=0.4945 mean_corrupt_t=0.4945 wrong_frac=0.5044 init_acc_corrupt=0.4597 init_gold_top10=0.4898 init_gold_top100=0.4957
+step=4000 micro_steps=8000 elapsed=33.6s lr=3.000000e-04 loss_all=3.1721 acc_all=0.5537 loss_corrupt=3.1721 acc_corrupt=0.5537 corrupt_frac=1.0000 loss=3.1721 loss_recon=3.1721 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5010 init_acc_corrupt=0.4641 init_gold_top10=0.4935 init_gold_top100=0.4991
+step=4100 micro_steps=8200 elapsed=34.9s lr=3.000000e-04 loss_all=3.1584 acc_all=0.5556 loss_corrupt=3.1584 acc_corrupt=0.5556 corrupt_frac=1.0000 loss=3.1584 loss_recon=3.1584 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4998 init_acc_corrupt=0.4653 init_gold_top10=0.4945 init_gold_top100=0.5004
+step=4200 micro_steps=8400 elapsed=33.4s lr=3.000000e-04 loss_all=3.1657 acc_all=0.5548 loss_corrupt=3.1657 acc_corrupt=0.5548 corrupt_frac=1.0000 loss=3.1657 loss_recon=3.1657 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5015 init_acc_corrupt=0.4640 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=4300 micro_steps=8600 elapsed=33.4s lr=3.000000e-04 loss_all=3.1475 acc_all=0.5564 loss_corrupt=3.1475 acc_corrupt=0.5564 corrupt_frac=1.0000 loss=3.1475 loss_recon=3.1475 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5003 init_acc_corrupt=0.4645 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=4400 micro_steps=8800 elapsed=33.4s lr=3.000000e-04 loss_all=3.1449 acc_all=0.5569 loss_corrupt=3.1449 acc_corrupt=0.5569 corrupt_frac=1.0000 loss=3.1449 loss_recon=3.1449 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5005 init_acc_corrupt=0.4650 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=4500 micro_steps=9000 elapsed=33.6s lr=3.000000e-04 loss_all=3.1135 acc_all=0.5607 loss_corrupt=3.1135 acc_corrupt=0.5607 corrupt_frac=1.0000 loss=3.1135 loss_recon=3.1135 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4975 init_acc_corrupt=0.4685 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=4600 micro_steps=9200 elapsed=33.7s lr=3.000000e-04 loss_all=3.1074 acc_all=0.5612 loss_corrupt=3.1074 acc_corrupt=0.5612 corrupt_frac=1.0000 loss=3.1074 loss_recon=3.1074 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4969 init_acc_corrupt=0.4695 init_gold_top10=0.4977 init_gold_top100=0.5032
+step=4700 micro_steps=9400 elapsed=33.7s lr=3.000000e-04 loss_all=3.0904 acc_all=0.5635 loss_corrupt=3.0904 acc_corrupt=0.5635 corrupt_frac=1.0000 loss=3.0904 loss_recon=3.0904 loss_meanflow=0.0000 mean_model_t=0.5055 mean_corrupt_t=0.5055 wrong_frac=0.4946 init_acc_corrupt=0.4713 init_gold_top10=0.4999 init_gold_top100=0.5055
+step=4800 micro_steps=9600 elapsed=33.6s lr=3.000000e-04 loss_all=3.1015 acc_all=0.5620 loss_corrupt=3.1015 acc_corrupt=0.5620 corrupt_frac=1.0000 loss=3.1015 loss_recon=3.1015 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4966 init_acc_corrupt=0.4689 init_gold_top10=0.4979 init_gold_top100=0.5036
+step=4900 micro_steps=9800 elapsed=33.7s lr=3.000000e-04 loss_all=3.0846 acc_all=0.5635 loss_corrupt=3.0846 acc_corrupt=0.5635 corrupt_frac=1.0000 loss=3.0846 loss_recon=3.0846 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5045 wrong_frac=0.4953 init_acc_corrupt=0.4705 init_gold_top10=0.4992 init_gold_top100=0.5049
+step=5000 micro_steps=10000 elapsed=33.7s lr=3.000000e-04 loss_all=3.1064 acc_all=0.5613 loss_corrupt=3.1064 acc_corrupt=0.5613 corrupt_frac=1.0000 loss=3.1064 loss_recon=3.1064 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4981 init_acc_corrupt=0.4676 init_gold_top10=0.4963 init_gold_top100=0.5021
+step=5100 micro_steps=10200 elapsed=35.2s lr=3.000000e-04 loss_all=3.1060 acc_all=0.5615 loss_corrupt=3.1060 acc_corrupt=0.5615 corrupt_frac=1.0000 loss=3.1060 loss_recon=3.1060 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4977 init_acc_corrupt=0.4680 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=5200 micro_steps=10400 elapsed=33.6s lr=3.000000e-04 loss_all=3.1293 acc_all=0.5580 loss_corrupt=3.1293 acc_corrupt=0.5580 corrupt_frac=1.0000 loss=3.1293 loss_recon=3.1293 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5012 init_acc_corrupt=0.4640 init_gold_top10=0.4932 init_gold_top100=0.4989
+step=5300 micro_steps=10600 elapsed=33.6s lr=3.000000e-04 loss_all=3.1231 acc_all=0.5584 loss_corrupt=3.1231 acc_corrupt=0.5584 corrupt_frac=1.0000 loss=3.1231 loss_recon=3.1231 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5013 init_acc_corrupt=0.4640 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=5400 micro_steps=10800 elapsed=33.7s lr=3.000000e-04 loss_all=3.0877 acc_all=0.5634 loss_corrupt=3.0877 acc_corrupt=0.5634 corrupt_frac=1.0000 loss=3.0877 loss_recon=3.0877 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4972 init_acc_corrupt=0.4691 init_gold_top10=0.4974 init_gold_top100=0.5029
+step=5500 micro_steps=11000 elapsed=33.6s lr=3.000000e-04 loss_all=3.0977 acc_all=0.5615 loss_corrupt=3.0977 acc_corrupt=0.5615 corrupt_frac=1.0000 loss=3.0977 loss_recon=3.0977 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4987 init_acc_corrupt=0.4665 init_gold_top10=0.4956 init_gold_top100=0.5015
+step=5600 micro_steps=11200 elapsed=33.6s lr=3.000000e-04 loss_all=3.0901 acc_all=0.5625 loss_corrupt=3.0901 acc_corrupt=0.5625 corrupt_frac=1.0000 loss=3.0901 loss_recon=3.0901 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4986 init_acc_corrupt=0.4672 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=5700 micro_steps=11400 elapsed=49.6s lr=3.000000e-04 loss_all=3.1270 acc_all=0.5577 loss_corrupt=3.1270 acc_corrupt=0.5577 corrupt_frac=1.0000 loss=3.1270 loss_recon=3.1270 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5029 init_acc_corrupt=0.4615 init_gold_top10=0.4914 init_gold_top100=0.4973
+step=5800 micro_steps=11600 elapsed=33.8s lr=3.000000e-04 loss_all=3.0925 acc_all=0.5619 loss_corrupt=3.0925 acc_corrupt=0.5619 corrupt_frac=1.0000 loss=3.0925 loss_recon=3.0925 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4993 init_acc_corrupt=0.4663 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=5900 micro_steps=11800 elapsed=33.7s lr=3.000000e-04 loss_all=3.1141 acc_all=0.5594 loss_corrupt=3.1141 acc_corrupt=0.5594 corrupt_frac=1.0000 loss=3.1141 loss_recon=3.1141 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5029 init_acc_corrupt=0.4628 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=6000 micro_steps=12000 elapsed=33.6s lr=3.000000e-04 loss_all=3.1019 acc_all=0.5609 loss_corrupt=3.1019 acc_corrupt=0.5609 corrupt_frac=1.0000 loss=3.1019 loss_recon=3.1019 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5009 init_acc_corrupt=0.4650 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=6100 micro_steps=12200 elapsed=35.1s lr=3.000000e-04 loss_all=3.1000 acc_all=0.5609 loss_corrupt=3.1000 acc_corrupt=0.5609 corrupt_frac=1.0000 loss=3.1000 loss_recon=3.1000 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4636 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=6200 micro_steps=12400 elapsed=33.7s lr=3.000000e-04 loss_all=3.0635 acc_all=0.5654 loss_corrupt=3.0635 acc_corrupt=0.5654 corrupt_frac=1.0000 loss=3.0635 loss_recon=3.0635 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4968 init_acc_corrupt=0.4690 init_gold_top10=0.4977 init_gold_top100=0.5034
+step=6300 micro_steps=12600 elapsed=33.6s lr=3.000000e-04 loss_all=3.0695 acc_all=0.5644 loss_corrupt=3.0695 acc_corrupt=0.5644 corrupt_frac=1.0000 loss=3.0695 loss_recon=3.0695 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4980 init_acc_corrupt=0.4675 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=6400 micro_steps=12800 elapsed=33.6s lr=3.000000e-04 loss_all=3.0532 acc_all=0.5665 loss_corrupt=3.0532 acc_corrupt=0.5665 corrupt_frac=1.0000 loss=3.0532 loss_recon=3.0532 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4967 init_acc_corrupt=0.4693 init_gold_top10=0.4976 init_gold_top100=0.5034
+step=6500 micro_steps=13000 elapsed=33.6s lr=3.000000e-04 loss_all=3.1072 acc_all=0.5594 loss_corrupt=3.1072 acc_corrupt=0.5594 corrupt_frac=1.0000 loss=3.1072 loss_recon=3.1072 loss_meanflow=0.0000 mean_model_t=0.4962 mean_corrupt_t=0.4962 wrong_frac=0.5040 init_acc_corrupt=0.4610 init_gold_top10=0.4902 init_gold_top100=0.4961
+step=6600 micro_steps=13200 elapsed=33.5s lr=3.000000e-04 loss_all=3.0604 acc_all=0.5653 loss_corrupt=3.0604 acc_corrupt=0.5653 corrupt_frac=1.0000 loss=3.0604 loss_recon=3.0604 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4980 init_acc_corrupt=0.4677 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=6700 micro_steps=13400 elapsed=33.6s lr=3.000000e-04 loss_all=3.0854 acc_all=0.5627 loss_corrupt=3.0854 acc_corrupt=0.5627 corrupt_frac=1.0000 loss=3.0854 loss_recon=3.0854 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5004 init_acc_corrupt=0.4651 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=6800 micro_steps=13600 elapsed=33.6s lr=3.000000e-04 loss_all=3.0952 acc_all=0.5610 loss_corrupt=3.0952 acc_corrupt=0.5610 corrupt_frac=1.0000 loss=3.0952 loss_recon=3.0952 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5025 init_acc_corrupt=0.4636 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=6900 micro_steps=13800 elapsed=33.5s lr=3.000000e-04 loss_all=3.0690 acc_all=0.5646 loss_corrupt=3.0690 acc_corrupt=0.5646 corrupt_frac=1.0000 loss=3.0690 loss_recon=3.0690 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4987 init_acc_corrupt=0.4671 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=7000 micro_steps=14000 elapsed=33.7s lr=3.000000e-04 loss_all=3.0710 acc_all=0.5640 loss_corrupt=3.0710 acc_corrupt=0.5640 corrupt_frac=1.0000 loss=3.0710 loss_recon=3.0710 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4996 init_acc_corrupt=0.4661 init_gold_top10=0.4947 init_gold_top100=0.5005
+step=7100 micro_steps=14200 elapsed=35.2s lr=3.000000e-04 loss_all=3.0591 acc_all=0.5659 loss_corrupt=3.0591 acc_corrupt=0.5659 corrupt_frac=1.0000 loss=3.0591 loss_recon=3.0591 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4991 init_acc_corrupt=0.4666 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=7200 micro_steps=14400 elapsed=33.6s lr=3.000000e-04 loss_all=3.0549 acc_all=0.5663 loss_corrupt=3.0549 acc_corrupt=0.5663 corrupt_frac=1.0000 loss=3.0549 loss_recon=3.0549 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4974 init_acc_corrupt=0.4679 init_gold_top10=0.4971 init_gold_top100=0.5027
+step=7300 micro_steps=14600 elapsed=33.6s lr=3.000000e-04 loss_all=3.1097 acc_all=0.5585 loss_corrupt=3.1097 acc_corrupt=0.5585 corrupt_frac=1.0000 loss=3.1097 loss_recon=3.1097 loss_meanflow=0.0000 mean_model_t=0.4943 mean_corrupt_t=0.4943 wrong_frac=0.5056 init_acc_corrupt=0.4590 init_gold_top10=0.4887 init_gold_top100=0.4945
+step=7400 micro_steps=14800 elapsed=33.8s lr=3.000000e-04 loss_all=3.0806 acc_all=0.5625 loss_corrupt=3.0806 acc_corrupt=0.5625 corrupt_frac=1.0000 loss=3.0806 loss_recon=3.0806 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=7500 micro_steps=15000 elapsed=33.7s lr=3.000000e-04 loss_all=3.0143 acc_all=0.5709 loss_corrupt=3.0143 acc_corrupt=0.5709 corrupt_frac=1.0000 loss=3.0143 loss_recon=3.0143 loss_meanflow=0.0000 mean_model_t=0.5048 mean_corrupt_t=0.5048 wrong_frac=0.4950 init_acc_corrupt=0.4711 init_gold_top10=0.4996 init_gold_top100=0.5051
+step=7600 micro_steps=15200 elapsed=33.8s lr=3.000000e-04 loss_all=3.0564 acc_all=0.5653 loss_corrupt=3.0564 acc_corrupt=0.5653 corrupt_frac=1.0000 loss=3.0564 loss_recon=3.0564 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4996 init_acc_corrupt=0.4652 init_gold_top10=0.4947 init_gold_top100=0.5006
+step=7700 micro_steps=15400 elapsed=33.7s lr=3.000000e-04 loss_all=3.0665 acc_all=0.5642 loss_corrupt=3.0665 acc_corrupt=0.5642 corrupt_frac=1.0000 loss=3.0665 loss_recon=3.0665 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5004 init_acc_corrupt=0.4642 init_gold_top10=0.4938 init_gold_top100=0.4997
+step=7800 micro_steps=15600 elapsed=33.6s lr=3.000000e-04 loss_all=3.0619 acc_all=0.5653 loss_corrupt=3.0619 acc_corrupt=0.5653 corrupt_frac=1.0000 loss=3.0619 loss_recon=3.0619 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4998 init_acc_corrupt=0.4649 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=7900 micro_steps=15800 elapsed=33.6s lr=3.000000e-04 loss_all=3.0462 acc_all=0.5670 loss_corrupt=3.0462 acc_corrupt=0.5670 corrupt_frac=1.0000 loss=3.0462 loss_recon=3.0462 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4979 init_acc_corrupt=0.4672 init_gold_top10=0.4965 init_gold_top100=0.5021
+step=8000 micro_steps=16000 elapsed=33.7s lr=3.000000e-04 loss_all=3.0484 acc_all=0.5664 loss_corrupt=3.0484 acc_corrupt=0.5664 corrupt_frac=1.0000 loss=3.0484 loss_recon=3.0484 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4993 init_acc_corrupt=0.4663 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=8100 micro_steps=16200 elapsed=35.4s lr=3.000000e-04 loss_all=3.0336 acc_all=0.5685 loss_corrupt=3.0336 acc_corrupt=0.5685 corrupt_frac=1.0000 loss=3.0336 loss_recon=3.0336 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4982 init_acc_corrupt=0.4680 init_gold_top10=0.4964 init_gold_top100=0.5019
+step=8200 micro_steps=16400 elapsed=33.8s lr=3.000000e-04 loss_all=3.0412 acc_all=0.5671 loss_corrupt=3.0412 acc_corrupt=0.5671 corrupt_frac=1.0000 loss=3.0412 loss_recon=3.0412 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4994 init_acc_corrupt=0.4661 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=8300 micro_steps=16600 elapsed=33.8s lr=3.000000e-04 loss_all=3.0303 acc_all=0.5685 loss_corrupt=3.0303 acc_corrupt=0.5685 corrupt_frac=1.0000 loss=3.0303 loss_recon=3.0303 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4982 init_acc_corrupt=0.4671 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=8400 micro_steps=16800 elapsed=33.8s lr=3.000000e-04 loss_all=3.0577 acc_all=0.5646 loss_corrupt=3.0577 acc_corrupt=0.5646 corrupt_frac=1.0000 loss=3.0577 loss_recon=3.0577 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5015 init_acc_corrupt=0.4638 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=8500 micro_steps=17000 elapsed=33.7s lr=3.000000e-04 loss_all=3.0671 acc_all=0.5634 loss_corrupt=3.0671 acc_corrupt=0.5634 corrupt_frac=1.0000 loss=3.0671 loss_recon=3.0671 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5036 init_acc_corrupt=0.4616 init_gold_top10=0.4908 init_gold_top100=0.4966
+step=8600 micro_steps=17200 elapsed=33.7s lr=3.000000e-04 loss_all=3.0891 acc_all=0.5607 loss_corrupt=3.0891 acc_corrupt=0.5607 corrupt_frac=1.0000 loss=3.0891 loss_recon=3.0891 loss_meanflow=0.0000 mean_model_t=0.4940 mean_corrupt_t=0.4940 wrong_frac=0.5061 init_acc_corrupt=0.4591 init_gold_top10=0.4882 init_gold_top100=0.4940
+step=8700 micro_steps=17400 elapsed=33.7s lr=3.000000e-04 loss_all=3.0006 acc_all=0.5726 loss_corrupt=3.0006 acc_corrupt=0.5726 corrupt_frac=1.0000 loss=3.0006 loss_recon=3.0006 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4957 init_acc_corrupt=0.4711 init_gold_top10=0.4989 init_gold_top100=0.5044
+step=8800 micro_steps=17600 elapsed=33.7s lr=3.000000e-04 loss_all=3.0292 acc_all=0.5687 loss_corrupt=3.0292 acc_corrupt=0.5687 corrupt_frac=1.0000 loss=3.0292 loss_recon=3.0292 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4989 init_acc_corrupt=0.4666 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=8900 micro_steps=17800 elapsed=33.7s lr=3.000000e-04 loss_all=3.0657 acc_all=0.5639 loss_corrupt=3.0657 acc_corrupt=0.5639 corrupt_frac=1.0000 loss=3.0657 loss_recon=3.0657 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5031 init_acc_corrupt=0.4619 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=9000 micro_steps=18000 elapsed=33.6s lr=3.000000e-04 loss_all=3.0677 acc_all=0.5634 loss_corrupt=3.0677 acc_corrupt=0.5634 corrupt_frac=1.0000 loss=3.0677 loss_recon=3.0677 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5035 init_acc_corrupt=0.4617 init_gold_top10=0.4908 init_gold_top100=0.4965
+step=9100 micro_steps=18200 elapsed=35.0s lr=3.000000e-04 loss_all=3.0778 acc_all=0.5622 loss_corrupt=3.0778 acc_corrupt=0.5622 corrupt_frac=1.0000 loss=3.0778 loss_recon=3.0778 loss_meanflow=0.0000 mean_model_t=0.4950 mean_corrupt_t=0.4950 wrong_frac=0.5045 init_acc_corrupt=0.4604 init_gold_top10=0.4898 init_gold_top100=0.4956
+step=9200 micro_steps=18400 elapsed=33.6s lr=3.000000e-04 loss_all=3.0022 acc_all=0.5714 loss_corrupt=3.0022 acc_corrupt=0.5714 corrupt_frac=1.0000 loss=3.0022 loss_recon=3.0022 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5045 wrong_frac=0.4958 init_acc_corrupt=0.4702 init_gold_top10=0.4987 init_gold_top100=0.5043
+step=9300 micro_steps=18600 elapsed=33.9s lr=3.000000e-04 loss_all=3.0293 acc_all=0.5683 loss_corrupt=3.0293 acc_corrupt=0.5683 corrupt_frac=1.0000 loss=3.0293 loss_recon=3.0293 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4993 init_acc_corrupt=0.4667 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=9400 micro_steps=18800 elapsed=34.3s lr=3.000000e-04 loss_all=3.0244 acc_all=0.5686 loss_corrupt=3.0244 acc_corrupt=0.5686 corrupt_frac=1.0000 loss=3.0244 loss_recon=3.0244 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4986 init_acc_corrupt=0.4664 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=9500 micro_steps=19000 elapsed=34.1s lr=3.000000e-04 loss_all=3.0466 acc_all=0.5660 loss_corrupt=3.0466 acc_corrupt=0.5660 corrupt_frac=1.0000 loss=3.0466 loss_recon=3.0466 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5021 init_acc_corrupt=0.4624 init_gold_top10=0.4922 init_gold_top100=0.4981
+step=9600 micro_steps=19200 elapsed=34.2s lr=3.000000e-04 loss_all=3.0327 acc_all=0.5675 loss_corrupt=3.0327 acc_corrupt=0.5675 corrupt_frac=1.0000 loss=3.0327 loss_recon=3.0327 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5009 init_acc_corrupt=0.4647 init_gold_top10=0.4937 init_gold_top100=0.4993
+step=9700 micro_steps=19400 elapsed=34.4s lr=3.000000e-04 loss_all=3.0283 acc_all=0.5685 loss_corrupt=3.0283 acc_corrupt=0.5685 corrupt_frac=1.0000 loss=3.0283 loss_recon=3.0283 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4983 init_acc_corrupt=0.4668 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=9800 micro_steps=19600 elapsed=34.3s lr=3.000000e-04 loss_all=3.0091 acc_all=0.5703 loss_corrupt=3.0091 acc_corrupt=0.5703 corrupt_frac=1.0000 loss=3.0091 loss_recon=3.0091 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4988 init_acc_corrupt=0.4677 init_gold_top10=0.4957 init_gold_top100=0.5013
+step=9900 micro_steps=19800 elapsed=34.0s lr=3.000000e-04 loss_all=3.0219 acc_all=0.5688 loss_corrupt=3.0219 acc_corrupt=0.5688 corrupt_frac=1.0000 loss=3.0219 loss_recon=3.0219 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4998 init_acc_corrupt=0.4656 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=10000 micro_steps=20000 elapsed=33.9s lr=3.000000e-04 loss_all=3.0320 acc_all=0.5681 loss_corrupt=3.0320 acc_corrupt=0.5681 corrupt_frac=1.0000 loss=3.0320 loss_recon=3.0320 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5003 init_acc_corrupt=0.4648 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=10100 micro_steps=20200 elapsed=35.2s lr=3.000000e-04 loss_all=3.0374 acc_all=0.5669 loss_corrupt=3.0374 acc_corrupt=0.5669 corrupt_frac=1.0000 loss=3.0374 loss_recon=3.0374 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5014 init_acc_corrupt=0.4638 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=10200 micro_steps=20400 elapsed=33.9s lr=3.000000e-04 loss_all=3.0384 acc_all=0.5664 loss_corrupt=3.0384 acc_corrupt=0.5664 corrupt_frac=1.0000 loss=3.0384 loss_recon=3.0384 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5031 init_acc_corrupt=0.4623 init_gold_top10=0.4914 init_gold_top100=0.4971
+step=10300 micro_steps=20600 elapsed=33.8s lr=3.000000e-04 loss_all=3.0320 acc_all=0.5681 loss_corrupt=3.0320 acc_corrupt=0.5681 corrupt_frac=1.0000 loss=3.0320 loss_recon=3.0320 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5005 init_acc_corrupt=0.4648 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=10400 micro_steps=20800 elapsed=34.0s lr=3.000000e-04 loss_all=3.0729 acc_all=0.5626 loss_corrupt=3.0729 acc_corrupt=0.5626 corrupt_frac=1.0000 loss=3.0729 loss_recon=3.0729 loss_meanflow=0.0000 mean_model_t=0.4941 mean_corrupt_t=0.4941 wrong_frac=0.5059 init_acc_corrupt=0.4591 init_gold_top10=0.4883 init_gold_top100=0.4942
+step=10500 micro_steps=21000 elapsed=33.7s lr=3.000000e-04 loss_all=3.0250 acc_all=0.5684 loss_corrupt=3.0250 acc_corrupt=0.5684 corrupt_frac=1.0000 loss=3.0250 loss_recon=3.0250 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5001 init_acc_corrupt=0.4652 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=10600 micro_steps=21200 elapsed=33.6s lr=3.000000e-04 loss_all=3.0181 acc_all=0.5696 loss_corrupt=3.0181 acc_corrupt=0.5696 corrupt_frac=1.0000 loss=3.0181 loss_recon=3.0181 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4991 init_acc_corrupt=0.4668 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=10700 micro_steps=21400 elapsed=33.7s lr=3.000000e-04 loss_all=3.0008 acc_all=0.5710 loss_corrupt=3.0008 acc_corrupt=0.5710 corrupt_frac=1.0000 loss=3.0008 loss_recon=3.0008 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4991 init_acc_corrupt=0.4667 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=10800 micro_steps=21600 elapsed=33.8s lr=3.000000e-04 loss_all=3.0362 acc_all=0.5670 loss_corrupt=3.0362 acc_corrupt=0.5670 corrupt_frac=1.0000 loss=3.0362 loss_recon=3.0362 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5027 init_acc_corrupt=0.4623 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=10900 micro_steps=21800 elapsed=33.7s lr=3.000000e-04 loss_all=3.0032 acc_all=0.5713 loss_corrupt=3.0032 acc_corrupt=0.5713 corrupt_frac=1.0000 loss=3.0032 loss_recon=3.0032 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4982 init_acc_corrupt=0.4675 init_gold_top10=0.4962 init_gold_top100=0.5018
+step=11000 micro_steps=22000 elapsed=33.5s lr=3.000000e-04 loss_all=3.0204 acc_all=0.5685 loss_corrupt=3.0204 acc_corrupt=0.5685 corrupt_frac=1.0000 loss=3.0204 loss_recon=3.0204 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5015 init_acc_corrupt=0.4637 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=11100 micro_steps=22200 elapsed=39.8s lr=3.000000e-04 loss_all=2.9952 acc_all=0.5718 loss_corrupt=2.9952 acc_corrupt=0.5718 corrupt_frac=1.0000 loss=2.9952 loss_recon=2.9952 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4985 init_acc_corrupt=0.4676 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=11200 micro_steps=22400 elapsed=33.6s lr=3.000000e-04 loss_all=3.0408 acc_all=0.5658 loss_corrupt=3.0408 acc_corrupt=0.5658 corrupt_frac=1.0000 loss=3.0408 loss_recon=3.0408 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5038 init_acc_corrupt=0.4613 init_gold_top10=0.4906 init_gold_top100=0.4963
+step=11300 micro_steps=22600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9914 acc_all=0.5724 loss_corrupt=2.9914 acc_corrupt=0.5724 corrupt_frac=1.0000 loss=2.9914 loss_recon=2.9914 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4985 init_acc_corrupt=0.4675 init_gold_top10=0.4962 init_gold_top100=0.5017
+step=11400 micro_steps=22800 elapsed=33.7s lr=3.000000e-04 loss_all=3.0056 acc_all=0.5708 loss_corrupt=3.0056 acc_corrupt=0.5708 corrupt_frac=1.0000 loss=3.0056 loss_recon=3.0056 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4994 init_acc_corrupt=0.4658 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=11500 micro_steps=23000 elapsed=33.7s lr=3.000000e-04 loss_all=3.0264 acc_all=0.5680 loss_corrupt=3.0264 acc_corrupt=0.5680 corrupt_frac=1.0000 loss=3.0264 loss_recon=3.0264 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5016 init_acc_corrupt=0.4630 init_gold_top10=0.4927 init_gold_top100=0.4986
+step=11600 micro_steps=23200 elapsed=33.7s lr=3.000000e-04 loss_all=3.0114 acc_all=0.5701 loss_corrupt=3.0114 acc_corrupt=0.5701 corrupt_frac=1.0000 loss=3.0114 loss_recon=3.0114 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5000 init_acc_corrupt=0.4653 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=11700 micro_steps=23400 elapsed=33.5s lr=3.000000e-04 loss_all=3.0019 acc_all=0.5709 loss_corrupt=3.0019 acc_corrupt=0.5709 corrupt_frac=1.0000 loss=3.0019 loss_recon=3.0019 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4999 init_acc_corrupt=0.4652 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=11800 micro_steps=23600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9707 acc_all=0.5745 loss_corrupt=2.9707 acc_corrupt=0.5745 corrupt_frac=1.0000 loss=2.9707 loss_recon=2.9707 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4969 init_acc_corrupt=0.4685 init_gold_top10=0.4977 init_gold_top100=0.5033
+step=11900 micro_steps=23800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9516 acc_all=0.5774 loss_corrupt=2.9516 acc_corrupt=0.5774 corrupt_frac=1.0000 loss=2.9516 loss_recon=2.9516 loss_meanflow=0.0000 mean_model_t=0.5058 mean_corrupt_t=0.5058 wrong_frac=0.4937 init_acc_corrupt=0.4725 init_gold_top10=0.5008 init_gold_top100=0.5064
+step=12000 micro_steps=24000 elapsed=33.7s lr=3.000000e-04 loss_all=3.0435 acc_all=0.5657 loss_corrupt=3.0435 acc_corrupt=0.5657 corrupt_frac=1.0000 loss=3.0435 loss_recon=3.0435 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5046 init_acc_corrupt=0.4601 init_gold_top10=0.4898 init_gold_top100=0.4956
+step=12100 micro_steps=24200 elapsed=35.2s lr=3.000000e-04 loss_all=3.0127 acc_all=0.5692 loss_corrupt=3.0127 acc_corrupt=0.5692 corrupt_frac=1.0000 loss=3.0127 loss_recon=3.0127 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5020 init_acc_corrupt=0.4632 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=12200 micro_steps=24400 elapsed=33.7s lr=3.000000e-04 loss_all=2.9767 acc_all=0.5742 loss_corrupt=2.9767 acc_corrupt=0.5742 corrupt_frac=1.0000 loss=2.9767 loss_recon=2.9767 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4979 init_acc_corrupt=0.4679 init_gold_top10=0.4966 init_gold_top100=0.5023
+step=12300 micro_steps=24600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9988 acc_all=0.5716 loss_corrupt=2.9988 acc_corrupt=0.5716 corrupt_frac=1.0000 loss=2.9988 loss_recon=2.9988 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4991 init_acc_corrupt=0.4663 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=12400 micro_steps=24800 elapsed=33.7s lr=3.000000e-04 loss_all=3.0170 acc_all=0.5689 loss_corrupt=3.0170 acc_corrupt=0.5689 corrupt_frac=1.0000 loss=3.0170 loss_recon=3.0170 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=12500 micro_steps=25000 elapsed=33.9s lr=3.000000e-04 loss_all=2.9989 acc_all=0.5715 loss_corrupt=2.9989 acc_corrupt=0.5715 corrupt_frac=1.0000 loss=2.9989 loss_recon=2.9989 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5002 init_acc_corrupt=0.4653 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=12600 micro_steps=25200 elapsed=33.8s lr=3.000000e-04 loss_all=2.9697 acc_all=0.5746 loss_corrupt=2.9697 acc_corrupt=0.5746 corrupt_frac=1.0000 loss=2.9697 loss_recon=2.9697 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4977 init_acc_corrupt=0.4683 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=12700 micro_steps=25400 elapsed=33.6s lr=3.000000e-04 loss_all=2.9920 acc_all=0.5722 loss_corrupt=2.9920 acc_corrupt=0.5722 corrupt_frac=1.0000 loss=2.9920 loss_recon=2.9920 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4992 init_acc_corrupt=0.4663 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=12800 micro_steps=25600 elapsed=33.8s lr=3.000000e-04 loss_all=3.0222 acc_all=0.5680 loss_corrupt=3.0222 acc_corrupt=0.5680 corrupt_frac=1.0000 loss=3.0222 loss_recon=3.0222 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5030 init_acc_corrupt=0.4626 init_gold_top10=0.4911 init_gold_top100=0.4970
+step=12900 micro_steps=25800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9882 acc_all=0.5728 loss_corrupt=2.9882 acc_corrupt=0.5728 corrupt_frac=1.0000 loss=2.9882 loss_recon=2.9882 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4984 init_acc_corrupt=0.4673 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=13000 micro_steps=26000 elapsed=33.6s lr=3.000000e-04 loss_all=3.0076 acc_all=0.5703 loss_corrupt=3.0076 acc_corrupt=0.5703 corrupt_frac=1.0000 loss=3.0076 loss_recon=3.0076 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5007 init_acc_corrupt=0.4640 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=13100 micro_steps=26200 elapsed=35.3s lr=3.000000e-04 loss_all=2.9572 acc_all=0.5761 loss_corrupt=2.9572 acc_corrupt=0.5761 corrupt_frac=1.0000 loss=2.9572 loss_recon=2.9572 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4964 init_acc_corrupt=0.4701 init_gold_top10=0.4981 init_gold_top100=0.5037
+step=13200 micro_steps=26400 elapsed=33.9s lr=3.000000e-04 loss_all=2.9912 acc_all=0.5719 loss_corrupt=2.9912 acc_corrupt=0.5719 corrupt_frac=1.0000 loss=2.9912 loss_recon=2.9912 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5003 init_acc_corrupt=0.4659 init_gold_top10=0.4943 init_gold_top100=0.4999
+step=13300 micro_steps=26600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9776 acc_all=0.5740 loss_corrupt=2.9776 acc_corrupt=0.5740 corrupt_frac=1.0000 loss=2.9776 loss_recon=2.9776 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4989 init_acc_corrupt=0.4667 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=13400 micro_steps=26800 elapsed=33.6s lr=3.000000e-04 loss_all=2.9562 acc_all=0.5763 loss_corrupt=2.9562 acc_corrupt=0.5763 corrupt_frac=1.0000 loss=2.9562 loss_recon=2.9562 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4970 init_acc_corrupt=0.4691 init_gold_top10=0.4975 init_gold_top100=0.5032
+step=13500 micro_steps=27000 elapsed=33.5s lr=3.000000e-04 loss_all=3.0134 acc_all=0.5693 loss_corrupt=3.0134 acc_corrupt=0.5693 corrupt_frac=1.0000 loss=3.0134 loss_recon=3.0134 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5020 init_acc_corrupt=0.4632 init_gold_top10=0.4922 init_gold_top100=0.4980
+step=13600 micro_steps=27200 elapsed=33.5s lr=3.000000e-04 loss_all=2.9607 acc_all=0.5766 loss_corrupt=2.9607 acc_corrupt=0.5766 corrupt_frac=1.0000 loss=2.9607 loss_recon=2.9607 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4959 init_acc_corrupt=0.4702 init_gold_top10=0.4986 init_gold_top100=0.5042
+step=13700 micro_steps=27400 elapsed=33.7s lr=3.000000e-04 loss_all=3.0027 acc_all=0.5703 loss_corrupt=3.0027 acc_corrupt=0.5703 corrupt_frac=1.0000 loss=3.0027 loss_recon=3.0027 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5016 init_acc_corrupt=0.4636 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=13800 micro_steps=27600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9688 acc_all=0.5747 loss_corrupt=2.9688 acc_corrupt=0.5747 corrupt_frac=1.0000 loss=2.9688 loss_recon=2.9688 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4985 init_acc_corrupt=0.4674 init_gold_top10=0.4959 init_gold_top100=0.5015
+step=13900 micro_steps=27800 elapsed=36.0s lr=3.000000e-04 loss_all=2.9939 acc_all=0.5717 loss_corrupt=2.9939 acc_corrupt=0.5717 corrupt_frac=1.0000 loss=2.9939 loss_recon=2.9939 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5004 init_acc_corrupt=0.4653 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=14000 micro_steps=28000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9669 acc_all=0.5752 loss_corrupt=2.9669 acc_corrupt=0.5752 corrupt_frac=1.0000 loss=2.9669 loss_recon=2.9669 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4970 init_acc_corrupt=0.4694 init_gold_top10=0.4973 init_gold_top100=0.5031
+step=14100 micro_steps=28200 elapsed=35.3s lr=3.000000e-04 loss_all=3.0162 acc_all=0.5687 loss_corrupt=3.0162 acc_corrupt=0.5687 corrupt_frac=1.0000 loss=3.0162 loss_recon=3.0162 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5025 init_acc_corrupt=0.4624 init_gold_top10=0.4918 init_gold_top100=0.4977
+step=14200 micro_steps=28400 elapsed=33.8s lr=3.000000e-04 loss_all=3.0172 acc_all=0.5689 loss_corrupt=3.0172 acc_corrupt=0.5689 corrupt_frac=1.0000 loss=3.0172 loss_recon=3.0172 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5029 init_acc_corrupt=0.4614 init_gold_top10=0.4913 init_gold_top100=0.4972
+step=14300 micro_steps=28600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9873 acc_all=0.5722 loss_corrupt=2.9873 acc_corrupt=0.5722 corrupt_frac=1.0000 loss=2.9873 loss_recon=2.9873 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5004 init_acc_corrupt=0.4651 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=14400 micro_steps=28800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9683 acc_all=0.5750 loss_corrupt=2.9683 acc_corrupt=0.5750 corrupt_frac=1.0000 loss=2.9683 loss_recon=2.9683 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4976 init_acc_corrupt=0.4678 init_gold_top10=0.4968 init_gold_top100=0.5026
+step=14500 micro_steps=29000 elapsed=33.7s lr=3.000000e-04 loss_all=2.9737 acc_all=0.5737 loss_corrupt=2.9737 acc_corrupt=0.5737 corrupt_frac=1.0000 loss=2.9737 loss_recon=2.9737 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4989 init_acc_corrupt=0.4663 init_gold_top10=0.4955 init_gold_top100=0.5013
+step=14600 micro_steps=29200 elapsed=33.9s lr=3.000000e-04 loss_all=3.0074 acc_all=0.5698 loss_corrupt=3.0074 acc_corrupt=0.5698 corrupt_frac=1.0000 loss=3.0074 loss_recon=3.0074 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5030 init_acc_corrupt=0.4618 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=14700 micro_steps=29400 elapsed=33.9s lr=3.000000e-04 loss_all=2.9472 acc_all=0.5773 loss_corrupt=2.9472 acc_corrupt=0.5773 corrupt_frac=1.0000 loss=2.9472 loss_recon=2.9472 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4964 init_acc_corrupt=0.4695 init_gold_top10=0.4981 init_gold_top100=0.5038
+step=14800 micro_steps=29600 elapsed=33.9s lr=3.000000e-04 loss_all=3.0029 acc_all=0.5708 loss_corrupt=3.0029 acc_corrupt=0.5708 corrupt_frac=1.0000 loss=3.0029 loss_recon=3.0029 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5006 init_acc_corrupt=0.4638 init_gold_top10=0.4936 init_gold_top100=0.4996
+step=14900 micro_steps=29800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9624 acc_all=0.5752 loss_corrupt=2.9624 acc_corrupt=0.5752 corrupt_frac=1.0000 loss=2.9624 loss_recon=2.9624 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4975 init_acc_corrupt=0.4679 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=15000 micro_steps=30000 elapsed=33.9s lr=3.000000e-04 loss_all=3.0051 acc_all=0.5703 loss_corrupt=3.0051 acc_corrupt=0.5703 corrupt_frac=1.0000 loss=3.0051 loss_recon=3.0051 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5026 init_acc_corrupt=0.4619 init_gold_top10=0.4917 init_gold_top100=0.4975
+step=15100 micro_steps=30200 elapsed=35.4s lr=3.000000e-04 loss_all=3.0199 acc_all=0.5679 loss_corrupt=3.0199 acc_corrupt=0.5679 corrupt_frac=1.0000 loss=3.0199 loss_recon=3.0199 loss_meanflow=0.0000 mean_model_t=0.4944 mean_corrupt_t=0.4944 wrong_frac=0.5052 init_acc_corrupt=0.4593 init_gold_top10=0.4891 init_gold_top100=0.4950
+step=15200 micro_steps=30400 elapsed=33.9s lr=3.000000e-04 loss_all=3.0157 acc_all=0.5684 loss_corrupt=3.0157 acc_corrupt=0.5684 corrupt_frac=1.0000 loss=3.0157 loss_recon=3.0157 loss_meanflow=0.0000 mean_model_t=0.4955 mean_corrupt_t=0.4955 wrong_frac=0.5045 init_acc_corrupt=0.4600 init_gold_top10=0.4897 init_gold_top100=0.4956
+step=15300 micro_steps=30600 elapsed=49.5s lr=3.000000e-04 loss_all=2.9528 acc_all=0.5769 loss_corrupt=2.9528 acc_corrupt=0.5769 corrupt_frac=1.0000 loss=2.9528 loss_recon=2.9528 loss_meanflow=0.0000 mean_model_t=0.5036 mean_corrupt_t=0.5036 wrong_frac=0.4964 init_acc_corrupt=0.4697 init_gold_top10=0.4982 init_gold_top100=0.5037
+step=15400 micro_steps=30800 elapsed=38.3s lr=3.000000e-04 loss_all=2.9568 acc_all=0.5763 loss_corrupt=2.9568 acc_corrupt=0.5763 corrupt_frac=1.0000 loss=2.9568 loss_recon=2.9568 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4966 init_acc_corrupt=0.4692 init_gold_top10=0.4977 init_gold_top100=0.5035
+step=15500 micro_steps=31000 elapsed=38.0s lr=3.000000e-04 loss_all=2.9894 acc_all=0.5714 loss_corrupt=2.9894 acc_corrupt=0.5714 corrupt_frac=1.0000 loss=2.9894 loss_recon=2.9894 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5027 init_acc_corrupt=0.4626 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=15600 micro_steps=31200 elapsed=33.8s lr=3.000000e-04 loss_all=2.9679 acc_all=0.5745 loss_corrupt=2.9679 acc_corrupt=0.5745 corrupt_frac=1.0000 loss=2.9679 loss_recon=2.9679 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5000 init_acc_corrupt=0.4658 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=15700 micro_steps=31400 elapsed=34.2s lr=3.000000e-04 loss_all=2.9922 acc_all=0.5711 loss_corrupt=2.9922 acc_corrupt=0.5711 corrupt_frac=1.0000 loss=2.9922 loss_recon=2.9922 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5025 init_acc_corrupt=0.4627 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=15800 micro_steps=31600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9578 acc_all=0.5759 loss_corrupt=2.9578 acc_corrupt=0.5759 corrupt_frac=1.0000 loss=2.9578 loss_recon=2.9578 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4980 init_acc_corrupt=0.4676 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=15900 micro_steps=31800 elapsed=34.3s lr=3.000000e-04 loss_all=2.9495 acc_all=0.5769 loss_corrupt=2.9495 acc_corrupt=0.5769 corrupt_frac=1.0000 loss=2.9495 loss_recon=2.9495 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4976 init_acc_corrupt=0.4682 init_gold_top10=0.4967 init_gold_top100=0.5025
+step=16000 micro_steps=32000 elapsed=34.2s lr=3.000000e-04 loss_all=2.9613 acc_all=0.5752 loss_corrupt=2.9613 acc_corrupt=0.5752 corrupt_frac=1.0000 loss=2.9613 loss_recon=2.9613 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4991 init_acc_corrupt=0.4665 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=16100 micro_steps=32200 elapsed=35.9s lr=3.000000e-04 loss_all=2.9362 acc_all=0.5786 loss_corrupt=2.9362 acc_corrupt=0.5786 corrupt_frac=1.0000 loss=2.9362 loss_recon=2.9362 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4961 init_acc_corrupt=0.4695 init_gold_top10=0.4984 init_gold_top100=0.5040
+step=16200 micro_steps=32400 elapsed=34.0s lr=3.000000e-04 loss_all=2.9651 acc_all=0.5749 loss_corrupt=2.9651 acc_corrupt=0.5749 corrupt_frac=1.0000 loss=2.9651 loss_recon=2.9651 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4997 init_acc_corrupt=0.4665 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=16300 micro_steps=32600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9775 acc_all=0.5732 loss_corrupt=2.9775 acc_corrupt=0.5732 corrupt_frac=1.0000 loss=2.9775 loss_recon=2.9775 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5011 init_acc_corrupt=0.4641 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=16400 micro_steps=32800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9751 acc_all=0.5733 loss_corrupt=2.9751 acc_corrupt=0.5733 corrupt_frac=1.0000 loss=2.9751 loss_recon=2.9751 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5016 init_acc_corrupt=0.4636 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=16500 micro_steps=33000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9616 acc_all=0.5750 loss_corrupt=2.9616 acc_corrupt=0.5750 corrupt_frac=1.0000 loss=2.9616 loss_recon=2.9616 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4997 init_acc_corrupt=0.4658 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=16600 micro_steps=33200 elapsed=33.9s lr=3.000000e-04 loss_all=3.0071 acc_all=0.5690 loss_corrupt=3.0071 acc_corrupt=0.5690 corrupt_frac=1.0000 loss=3.0071 loss_recon=3.0071 loss_meanflow=0.0000 mean_model_t=0.4943 mean_corrupt_t=0.4943 wrong_frac=0.5052 init_acc_corrupt=0.4593 init_gold_top10=0.4891 init_gold_top100=0.4950
+step=16700 micro_steps=33400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9544 acc_all=0.5766 loss_corrupt=2.9544 acc_corrupt=0.5766 corrupt_frac=1.0000 loss=2.9544 loss_recon=2.9544 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4981 init_acc_corrupt=0.4673 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=16800 micro_steps=33600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9310 acc_all=0.5791 loss_corrupt=2.9310 acc_corrupt=0.5791 corrupt_frac=1.0000 loss=2.9310 loss_recon=2.9310 loss_meanflow=0.0000 mean_model_t=0.5046 mean_corrupt_t=0.5046 wrong_frac=0.4954 init_acc_corrupt=0.4705 init_gold_top10=0.4992 init_gold_top100=0.5048
+step=16900 micro_steps=33800 elapsed=33.7s lr=3.000000e-04 loss_all=3.0106 acc_all=0.5694 loss_corrupt=3.0106 acc_corrupt=0.5694 corrupt_frac=1.0000 loss=3.0106 loss_recon=3.0106 loss_meanflow=0.0000 mean_model_t=0.4948 mean_corrupt_t=0.4948 wrong_frac=0.5047 init_acc_corrupt=0.4598 init_gold_top10=0.4895 init_gold_top100=0.4954
+step=17000 micro_steps=34000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9750 acc_all=0.5733 loss_corrupt=2.9750 acc_corrupt=0.5733 corrupt_frac=1.0000 loss=2.9750 loss_recon=2.9750 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5009 init_acc_corrupt=0.4644 init_gold_top10=0.4933 init_gold_top100=0.4991
+step=17100 micro_steps=34200 elapsed=35.3s lr=3.000000e-04 loss_all=2.9636 acc_all=0.5753 loss_corrupt=2.9636 acc_corrupt=0.5753 corrupt_frac=1.0000 loss=2.9636 loss_recon=2.9636 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4989 init_acc_corrupt=0.4667 init_gold_top10=0.4954 init_gold_top100=0.5012
+step=17200 micro_steps=34400 elapsed=33.7s lr=3.000000e-04 loss_all=2.9716 acc_all=0.5737 loss_corrupt=2.9716 acc_corrupt=0.5737 corrupt_frac=1.0000 loss=2.9716 loss_recon=2.9716 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5021 init_acc_corrupt=0.4632 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=17300 micro_steps=34600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9885 acc_all=0.5717 loss_corrupt=2.9885 acc_corrupt=0.5717 corrupt_frac=1.0000 loss=2.9885 loss_recon=2.9885 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5027 init_acc_corrupt=0.4623 init_gold_top10=0.4915 init_gold_top100=0.4974
+step=17400 micro_steps=34800 elapsed=34.3s lr=3.000000e-04 loss_all=2.9674 acc_all=0.5746 loss_corrupt=2.9674 acc_corrupt=0.5746 corrupt_frac=1.0000 loss=2.9674 loss_recon=2.9674 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5008 init_acc_corrupt=0.4647 init_gold_top10=0.4938 init_gold_top100=0.4994
+step=17500 micro_steps=35000 elapsed=34.2s lr=3.000000e-04 loss_all=2.9338 acc_all=0.5789 loss_corrupt=2.9338 acc_corrupt=0.5789 corrupt_frac=1.0000 loss=2.9338 loss_recon=2.9338 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4971 init_acc_corrupt=0.4693 init_gold_top10=0.4973 init_gold_top100=0.5030
+step=17600 micro_steps=35200 elapsed=34.0s lr=3.000000e-04 loss_all=2.9780 acc_all=0.5727 loss_corrupt=2.9780 acc_corrupt=0.5727 corrupt_frac=1.0000 loss=2.9780 loss_recon=2.9780 loss_meanflow=0.0000 mean_model_t=0.4966 mean_corrupt_t=0.4966 wrong_frac=0.5032 init_acc_corrupt=0.4625 init_gold_top10=0.4911 init_gold_top100=0.4969
+step=17700 micro_steps=35400 elapsed=34.1s lr=3.000000e-04 loss_all=2.9328 acc_all=0.5788 loss_corrupt=2.9328 acc_corrupt=0.5788 corrupt_frac=1.0000 loss=2.9328 loss_recon=2.9328 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4965 init_acc_corrupt=0.4699 init_gold_top10=0.4979 init_gold_top100=0.5037
+step=17800 micro_steps=35600 elapsed=34.2s lr=3.000000e-04 loss_all=2.9386 acc_all=0.5778 loss_corrupt=2.9386 acc_corrupt=0.5778 corrupt_frac=1.0000 loss=2.9386 loss_recon=2.9386 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4983 init_acc_corrupt=0.4671 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=17900 micro_steps=35800 elapsed=34.2s lr=3.000000e-04 loss_all=2.9848 acc_all=0.5720 loss_corrupt=2.9848 acc_corrupt=0.5720 corrupt_frac=1.0000 loss=2.9848 loss_recon=2.9848 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5039 init_acc_corrupt=0.4617 init_gold_top10=0.4904 init_gold_top100=0.4962
+step=18000 micro_steps=36000 elapsed=34.1s lr=3.000000e-04 loss_all=2.9617 acc_all=0.5755 loss_corrupt=2.9617 acc_corrupt=0.5755 corrupt_frac=1.0000 loss=2.9617 loss_recon=2.9617 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4992 init_acc_corrupt=0.4668 init_gold_top10=0.4953 init_gold_top100=0.5009
+step=18100 micro_steps=36200 elapsed=35.4s lr=3.000000e-04 loss_all=2.9609 acc_all=0.5756 loss_corrupt=2.9609 acc_corrupt=0.5756 corrupt_frac=1.0000 loss=2.9609 loss_recon=2.9609 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4986 init_acc_corrupt=0.4663 init_gold_top10=0.4956 init_gold_top100=0.5015
+step=18200 micro_steps=36400 elapsed=33.9s lr=3.000000e-04 loss_all=2.9644 acc_all=0.5746 loss_corrupt=2.9644 acc_corrupt=0.5746 corrupt_frac=1.0000 loss=2.9644 loss_recon=2.9644 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5012 init_acc_corrupt=0.4640 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=18300 micro_steps=36600 elapsed=34.0s lr=3.000000e-04 loss_all=2.9838 acc_all=0.5715 loss_corrupt=2.9838 acc_corrupt=0.5715 corrupt_frac=1.0000 loss=2.9838 loss_recon=2.9838 loss_meanflow=0.0000 mean_model_t=0.4947 mean_corrupt_t=0.4947 wrong_frac=0.5049 init_acc_corrupt=0.4603 init_gold_top10=0.4893 init_gold_top100=0.4952
+step=18400 micro_steps=36800 elapsed=34.0s lr=3.000000e-04 loss_all=2.9961 acc_all=0.5706 loss_corrupt=2.9961 acc_corrupt=0.5706 corrupt_frac=1.0000 loss=2.9961 loss_recon=2.9961 loss_meanflow=0.0000 mean_model_t=0.4946 mean_corrupt_t=0.4946 wrong_frac=0.5053 init_acc_corrupt=0.4602 init_gold_top10=0.4889 init_gold_top100=0.4948
+step=18500 micro_steps=37000 elapsed=34.1s lr=3.000000e-04 loss_all=2.9755 acc_all=0.5731 loss_corrupt=2.9755 acc_corrupt=0.5731 corrupt_frac=1.0000 loss=2.9755 loss_recon=2.9755 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5020 init_acc_corrupt=0.4629 init_gold_top10=0.4924 init_gold_top100=0.4981
+step=18600 micro_steps=37200 elapsed=34.1s lr=3.000000e-04 loss_all=2.9374 acc_all=0.5781 loss_corrupt=2.9374 acc_corrupt=0.5781 corrupt_frac=1.0000 loss=2.9374 loss_recon=2.9374 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4976 init_acc_corrupt=0.4683 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=18700 micro_steps=37400 elapsed=33.9s lr=3.000000e-04 loss_all=2.9682 acc_all=0.5737 loss_corrupt=2.9682 acc_corrupt=0.5737 corrupt_frac=1.0000 loss=2.9682 loss_recon=2.9682 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5026 init_acc_corrupt=0.4623 init_gold_top10=0.4917 init_gold_top100=0.4975
+step=18800 micro_steps=37600 elapsed=34.1s lr=3.000000e-04 loss_all=2.9467 acc_all=0.5769 loss_corrupt=2.9467 acc_corrupt=0.5769 corrupt_frac=1.0000 loss=2.9467 loss_recon=2.9467 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4986 init_acc_corrupt=0.4669 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=18900 micro_steps=37800 elapsed=33.8s lr=3.000000e-04 loss_all=3.0092 acc_all=0.5691 loss_corrupt=3.0092 acc_corrupt=0.5691 corrupt_frac=1.0000 loss=3.0092 loss_recon=3.0092 loss_meanflow=0.0000 mean_model_t=0.4942 mean_corrupt_t=0.4942 wrong_frac=0.5056 init_acc_corrupt=0.4587 init_gold_top10=0.4885 init_gold_top100=0.4943
+step=19000 micro_steps=38000 elapsed=33.7s lr=3.000000e-04 loss_all=2.9352 acc_all=0.5785 loss_corrupt=2.9352 acc_corrupt=0.5785 corrupt_frac=1.0000 loss=2.9352 loss_recon=2.9352 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4978 init_acc_corrupt=0.4679 init_gold_top10=0.4965 init_gold_top100=0.5024
+step=19100 micro_steps=38200 elapsed=35.2s lr=3.000000e-04 loss_all=2.9986 acc_all=0.5698 loss_corrupt=2.9986 acc_corrupt=0.5698 corrupt_frac=1.0000 loss=2.9986 loss_recon=2.9986 loss_meanflow=0.0000 mean_model_t=0.4947 mean_corrupt_t=0.4947 wrong_frac=0.5059 init_acc_corrupt=0.4590 init_gold_top10=0.4885 init_gold_top100=0.4943
+step=19200 micro_steps=38400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9581 acc_all=0.5756 loss_corrupt=2.9581 acc_corrupt=0.5756 corrupt_frac=1.0000 loss=2.9581 loss_recon=2.9581 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4993 init_acc_corrupt=0.4658 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=19300 micro_steps=38600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9463 acc_all=0.5771 loss_corrupt=2.9463 acc_corrupt=0.5771 corrupt_frac=1.0000 loss=2.9463 loss_recon=2.9463 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4983 init_acc_corrupt=0.4671 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=19400 micro_steps=38800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9695 acc_all=0.5735 loss_corrupt=2.9695 acc_corrupt=0.5735 corrupt_frac=1.0000 loss=2.9695 loss_recon=2.9695 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5025 init_acc_corrupt=0.4624 init_gold_top10=0.4918 init_gold_top100=0.4977
+step=19500 micro_steps=39000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9396 acc_all=0.5776 loss_corrupt=2.9396 acc_corrupt=0.5776 corrupt_frac=1.0000 loss=2.9396 loss_recon=2.9396 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4988 init_acc_corrupt=0.4668 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=19600 micro_steps=39200 elapsed=33.9s lr=3.000000e-04 loss_all=2.9481 acc_all=0.5762 loss_corrupt=2.9481 acc_corrupt=0.5762 corrupt_frac=1.0000 loss=2.9481 loss_recon=2.9481 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5012 init_acc_corrupt=0.4643 init_gold_top10=0.4934 init_gold_top100=0.4989
+step=19700 micro_steps=39400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9305 acc_all=0.5787 loss_corrupt=2.9305 acc_corrupt=0.5787 corrupt_frac=1.0000 loss=2.9305 loss_recon=2.9305 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4973 init_acc_corrupt=0.4678 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=19800 micro_steps=39600 elapsed=33.9s lr=3.000000e-04 loss_all=2.9185 acc_all=0.5805 loss_corrupt=2.9185 acc_corrupt=0.5805 corrupt_frac=1.0000 loss=2.9185 loss_recon=2.9185 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4971 init_acc_corrupt=0.4694 init_gold_top10=0.4975 init_gold_top100=0.5030
+step=19900 micro_steps=39800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9826 acc_all=0.5725 loss_corrupt=2.9826 acc_corrupt=0.5725 corrupt_frac=1.0000 loss=2.9826 loss_recon=2.9826 loss_meanflow=0.0000 mean_model_t=0.4966 mean_corrupt_t=0.4966 wrong_frac=0.5034 init_acc_corrupt=0.4618 init_gold_top10=0.4910 init_gold_top100=0.4967
+step=20000 micro_steps=40000 elapsed=33.7s lr=3.000000e-04 loss_all=2.9339 acc_all=0.5779 loss_corrupt=2.9339 acc_corrupt=0.5779 corrupt_frac=1.0000 loss=2.9339 loss_recon=2.9339 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4994 init_acc_corrupt=0.4667 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=20100 micro_steps=40200 elapsed=36.8s lr=3.000000e-04 loss_all=2.9992 acc_all=0.5700 loss_corrupt=2.9992 acc_corrupt=0.5700 corrupt_frac=1.0000 loss=2.9992 loss_recon=2.9992 loss_meanflow=0.0000 mean_model_t=0.4947 mean_corrupt_t=0.4947 wrong_frac=0.5056 init_acc_corrupt=0.4592 init_gold_top10=0.4885 init_gold_top100=0.4946
+step=20200 micro_steps=40400 elapsed=34.1s lr=3.000000e-04 loss_all=2.9598 acc_all=0.5754 loss_corrupt=2.9598 acc_corrupt=0.5754 corrupt_frac=1.0000 loss=2.9598 loss_recon=2.9598 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5009 init_acc_corrupt=0.4641 init_gold_top10=0.4936 init_gold_top100=0.4992
+step=20300 micro_steps=40600 elapsed=33.9s lr=3.000000e-04 loss_all=2.9439 acc_all=0.5768 loss_corrupt=2.9439 acc_corrupt=0.5768 corrupt_frac=1.0000 loss=2.9439 loss_recon=2.9439 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4998 init_acc_corrupt=0.4656 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=20400 micro_steps=40800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9407 acc_all=0.5773 loss_corrupt=2.9407 acc_corrupt=0.5773 corrupt_frac=1.0000 loss=2.9407 loss_recon=2.9407 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4999 init_acc_corrupt=0.4651 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=20500 micro_steps=41000 elapsed=33.9s lr=3.000000e-04 loss_all=2.9866 acc_all=0.5712 loss_corrupt=2.9866 acc_corrupt=0.5712 corrupt_frac=1.0000 loss=2.9866 loss_recon=2.9866 loss_meanflow=0.0000 mean_model_t=0.4942 mean_corrupt_t=0.4942 wrong_frac=0.5050 init_acc_corrupt=0.4591 init_gold_top10=0.4892 init_gold_top100=0.4951
+step=20600 micro_steps=41200 elapsed=33.6s lr=3.000000e-04 loss_all=2.9361 acc_all=0.5781 loss_corrupt=2.9361 acc_corrupt=0.5781 corrupt_frac=1.0000 loss=2.9361 loss_recon=2.9361 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4984 init_acc_corrupt=0.4670 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=20700 micro_steps=41400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9742 acc_all=0.5733 loss_corrupt=2.9742 acc_corrupt=0.5733 corrupt_frac=1.0000 loss=2.9742 loss_recon=2.9742 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5028 init_acc_corrupt=0.4623 init_gold_top10=0.4913 init_gold_top100=0.4972
+step=20800 micro_steps=41600 elapsed=34.3s lr=3.000000e-04 loss_all=2.9418 acc_all=0.5772 loss_corrupt=2.9418 acc_corrupt=0.5772 corrupt_frac=1.0000 loss=2.9418 loss_recon=2.9418 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4994 init_acc_corrupt=0.4662 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=20900 micro_steps=41800 elapsed=34.0s lr=3.000000e-04 loss_all=2.9117 acc_all=0.5812 loss_corrupt=2.9117 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9117 loss_recon=2.9117 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4966 init_acc_corrupt=0.4689 init_gold_top10=0.4980 init_gold_top100=0.5036
+step=21000 micro_steps=42000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9353 acc_all=0.5778 loss_corrupt=2.9353 acc_corrupt=0.5778 corrupt_frac=1.0000 loss=2.9353 loss_recon=2.9353 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4996 init_acc_corrupt=0.4658 init_gold_top10=0.4947 init_gold_top100=0.5006
+step=21100 micro_steps=42200 elapsed=35.2s lr=3.000000e-04 loss_all=2.9360 acc_all=0.5782 loss_corrupt=2.9360 acc_corrupt=0.5782 corrupt_frac=1.0000 loss=2.9360 loss_recon=2.9360 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4984 init_acc_corrupt=0.4671 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=21200 micro_steps=42400 elapsed=33.7s lr=3.000000e-04 loss_all=2.9436 acc_all=0.5768 loss_corrupt=2.9436 acc_corrupt=0.5768 corrupt_frac=1.0000 loss=2.9436 loss_recon=2.9436 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5005 init_acc_corrupt=0.4650 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=21300 micro_steps=42600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9308 acc_all=0.5790 loss_corrupt=2.9308 acc_corrupt=0.5790 corrupt_frac=1.0000 loss=2.9308 loss_recon=2.9308 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4980 init_acc_corrupt=0.4678 init_gold_top10=0.4963 init_gold_top100=0.5021
+step=21400 micro_steps=42800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9136 acc_all=0.5810 loss_corrupt=2.9136 acc_corrupt=0.5810 corrupt_frac=1.0000 loss=2.9136 loss_recon=2.9136 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4962 init_acc_corrupt=0.4695 init_gold_top10=0.4982 init_gold_top100=0.5039
+step=21500 micro_steps=43000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9507 acc_all=0.5759 loss_corrupt=2.9507 acc_corrupt=0.5759 corrupt_frac=1.0000 loss=2.9507 loss_recon=2.9507 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5018 init_acc_corrupt=0.4642 init_gold_top10=0.4927 init_gold_top100=0.4983
+step=21600 micro_steps=43200 elapsed=33.7s lr=3.000000e-04 loss_all=2.9663 acc_all=0.5743 loss_corrupt=2.9663 acc_corrupt=0.5743 corrupt_frac=1.0000 loss=2.9663 loss_recon=2.9663 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5022 init_acc_corrupt=0.4630 init_gold_top10=0.4920 init_gold_top100=0.4979
+step=21700 micro_steps=43400 elapsed=33.7s lr=3.000000e-04 loss_all=2.9341 acc_all=0.5781 loss_corrupt=2.9341 acc_corrupt=0.5781 corrupt_frac=1.0000 loss=2.9341 loss_recon=2.9341 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4988 init_acc_corrupt=0.4671 init_gold_top10=0.4955 init_gold_top100=0.5013
+step=21800 micro_steps=43600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9402 acc_all=0.5774 loss_corrupt=2.9402 acc_corrupt=0.5774 corrupt_frac=1.0000 loss=2.9402 loss_recon=2.9402 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5006 init_acc_corrupt=0.4649 init_gold_top10=0.4938 init_gold_top100=0.4994
+step=21900 micro_steps=43800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9125 acc_all=0.5809 loss_corrupt=2.9125 acc_corrupt=0.5809 corrupt_frac=1.0000 loss=2.9125 loss_recon=2.9125 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4969 init_acc_corrupt=0.4694 init_gold_top10=0.4977 init_gold_top100=0.5032
+step=22000 micro_steps=44000 elapsed=33.6s lr=3.000000e-04 loss_all=2.9242 acc_all=0.5793 loss_corrupt=2.9242 acc_corrupt=0.5793 corrupt_frac=1.0000 loss=2.9242 loss_recon=2.9242 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4989 init_acc_corrupt=0.4667 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=22100 micro_steps=44200 elapsed=35.3s lr=3.000000e-04 loss_all=2.9309 acc_all=0.5790 loss_corrupt=2.9309 acc_corrupt=0.5790 corrupt_frac=1.0000 loss=2.9309 loss_recon=2.9309 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4983 init_acc_corrupt=0.4671 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=22200 micro_steps=44400 elapsed=33.6s lr=3.000000e-04 loss_all=2.9409 acc_all=0.5775 loss_corrupt=2.9409 acc_corrupt=0.5775 corrupt_frac=1.0000 loss=2.9409 loss_recon=2.9409 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5000 init_acc_corrupt=0.4652 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=22300 micro_steps=44600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9192 acc_all=0.5791 loss_corrupt=2.9192 acc_corrupt=0.5791 corrupt_frac=1.0000 loss=2.9192 loss_recon=2.9192 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4987 init_acc_corrupt=0.4673 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=22400 micro_steps=44800 elapsed=33.6s lr=3.000000e-04 loss_all=2.9237 acc_all=0.5794 loss_corrupt=2.9237 acc_corrupt=0.5794 corrupt_frac=1.0000 loss=2.9237 loss_recon=2.9237 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4986 init_acc_corrupt=0.4672 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=22500 micro_steps=45000 elapsed=33.9s lr=3.000000e-04 loss_all=2.9426 acc_all=0.5772 loss_corrupt=2.9426 acc_corrupt=0.5772 corrupt_frac=1.0000 loss=2.9426 loss_recon=2.9426 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.4999 init_acc_corrupt=0.4653 init_gold_top10=0.4943 init_gold_top100=0.5002
+step=22600 micro_steps=45200 elapsed=33.7s lr=3.000000e-04 loss_all=2.9696 acc_all=0.5736 loss_corrupt=2.9696 acc_corrupt=0.5736 corrupt_frac=1.0000 loss=2.9696 loss_recon=2.9696 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5032 init_acc_corrupt=0.4617 init_gold_top10=0.4910 init_gold_top100=0.4969
+step=22700 micro_steps=45400 elapsed=33.6s lr=3.000000e-04 loss_all=2.9661 acc_all=0.5738 loss_corrupt=2.9661 acc_corrupt=0.5738 corrupt_frac=1.0000 loss=2.9661 loss_recon=2.9661 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5046 init_acc_corrupt=0.4601 init_gold_top10=0.4898 init_gold_top100=0.4956
+step=22800 micro_steps=45600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9280 acc_all=0.5789 loss_corrupt=2.9280 acc_corrupt=0.5789 corrupt_frac=1.0000 loss=2.9280 loss_recon=2.9280 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4991 init_acc_corrupt=0.4666 init_gold_top10=0.4953 init_gold_top100=0.5009
+step=22900 micro_steps=45800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9394 acc_all=0.5775 loss_corrupt=2.9394 acc_corrupt=0.5775 corrupt_frac=1.0000 loss=2.9394 loss_recon=2.9394 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4995 init_acc_corrupt=0.4657 init_gold_top10=0.4949 init_gold_top100=0.5007
+step=23000 micro_steps=46000 elapsed=33.7s lr=3.000000e-04 loss_all=2.9326 acc_all=0.5782 loss_corrupt=2.9326 acc_corrupt=0.5782 corrupt_frac=1.0000 loss=2.9326 loss_recon=2.9326 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4991 init_acc_corrupt=0.4662 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=23100 micro_steps=46200 elapsed=35.4s lr=3.000000e-04 loss_all=2.9485 acc_all=0.5764 loss_corrupt=2.9485 acc_corrupt=0.5764 corrupt_frac=1.0000 loss=2.9485 loss_recon=2.9485 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5016 init_acc_corrupt=0.4638 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=23200 micro_steps=46400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9214 acc_all=0.5796 loss_corrupt=2.9214 acc_corrupt=0.5796 corrupt_frac=1.0000 loss=2.9214 loss_recon=2.9214 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4982 init_acc_corrupt=0.4677 init_gold_top10=0.4961 init_gold_top100=0.5019
+step=23300 micro_steps=46600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9345 acc_all=0.5781 loss_corrupt=2.9345 acc_corrupt=0.5781 corrupt_frac=1.0000 loss=2.9345 loss_recon=2.9345 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.5001 init_acc_corrupt=0.4652 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=23400 micro_steps=46800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9436 acc_all=0.5769 loss_corrupt=2.9436 acc_corrupt=0.5769 corrupt_frac=1.0000 loss=2.9436 loss_recon=2.9436 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5012 init_acc_corrupt=0.4637 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=23500 micro_steps=47000 elapsed=33.9s lr=3.000000e-04 loss_all=2.9339 acc_all=0.5783 loss_corrupt=2.9339 acc_corrupt=0.5783 corrupt_frac=1.0000 loss=2.9339 loss_recon=2.9339 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5001 init_acc_corrupt=0.4657 init_gold_top10=0.4943 init_gold_top100=0.5000
+step=23600 micro_steps=47200 elapsed=34.0s lr=3.000000e-04 loss_all=2.9311 acc_all=0.5786 loss_corrupt=2.9311 acc_corrupt=0.5786 corrupt_frac=1.0000 loss=2.9311 loss_recon=2.9311 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4989 init_acc_corrupt=0.4668 init_gold_top10=0.4955 init_gold_top100=0.5011
+step=23700 micro_steps=47400 elapsed=33.9s lr=3.000000e-04 loss_all=2.9170 acc_all=0.5803 loss_corrupt=2.9170 acc_corrupt=0.5803 corrupt_frac=1.0000 loss=2.9170 loss_recon=2.9170 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4984 init_acc_corrupt=0.4675 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=23800 micro_steps=47600 elapsed=33.9s lr=3.000000e-04 loss_all=2.9540 acc_all=0.5753 loss_corrupt=2.9540 acc_corrupt=0.5753 corrupt_frac=1.0000 loss=2.9540 loss_recon=2.9540 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5021 init_acc_corrupt=0.4630 init_gold_top10=0.4921 init_gold_top100=0.4980
+step=23900 micro_steps=47800 elapsed=34.1s lr=3.000000e-04 loss_all=2.9160 acc_all=0.5807 loss_corrupt=2.9160 acc_corrupt=0.5807 corrupt_frac=1.0000 loss=2.9160 loss_recon=2.9160 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4962 init_acc_corrupt=0.4694 init_gold_top10=0.4982 init_gold_top100=0.5039
+step=24000 micro_steps=48000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9671 acc_all=0.5732 loss_corrupt=2.9671 acc_corrupt=0.5732 corrupt_frac=1.0000 loss=2.9671 loss_recon=2.9671 loss_meanflow=0.0000 mean_model_t=0.4948 mean_corrupt_t=0.4948 wrong_frac=0.5048 init_acc_corrupt=0.4603 init_gold_top10=0.4894 init_gold_top100=0.4953
+step=24100 micro_steps=48200 elapsed=35.3s lr=3.000000e-04 loss_all=2.9427 acc_all=0.5769 loss_corrupt=2.9427 acc_corrupt=0.5769 corrupt_frac=1.0000 loss=2.9427 loss_recon=2.9427 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5015 init_acc_corrupt=0.4636 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=24200 micro_steps=48400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9362 acc_all=0.5775 loss_corrupt=2.9362 acc_corrupt=0.5775 corrupt_frac=1.0000 loss=2.9362 loss_recon=2.9362 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5015 init_acc_corrupt=0.4640 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=24300 micro_steps=48600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9471 acc_all=0.5764 loss_corrupt=2.9471 acc_corrupt=0.5764 corrupt_frac=1.0000 loss=2.9471 loss_recon=2.9471 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5016 init_acc_corrupt=0.4637 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=24400 micro_steps=48800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9121 acc_all=0.5812 loss_corrupt=2.9121 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9121 loss_recon=2.9121 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4971 init_acc_corrupt=0.4691 init_gold_top10=0.4974 init_gold_top100=0.5031
+step=24500 micro_steps=49000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9135 acc_all=0.5805 loss_corrupt=2.9135 acc_corrupt=0.5805 corrupt_frac=1.0000 loss=2.9135 loss_recon=2.9135 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4984 init_acc_corrupt=0.4673 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=24600 micro_steps=49200 elapsed=33.9s lr=3.000000e-04 loss_all=2.9250 acc_all=0.5785 loss_corrupt=2.9250 acc_corrupt=0.5785 corrupt_frac=1.0000 loss=2.9250 loss_recon=2.9250 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5004 init_acc_corrupt=0.4643 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=24700 micro_steps=49400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9241 acc_all=0.5793 loss_corrupt=2.9241 acc_corrupt=0.5793 corrupt_frac=1.0000 loss=2.9241 loss_recon=2.9241 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5001 init_acc_corrupt=0.4656 init_gold_top10=0.4944 init_gold_top100=0.5000
+step=24800 micro_steps=49600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8822 acc_all=0.5845 loss_corrupt=2.8822 acc_corrupt=0.5845 corrupt_frac=1.0000 loss=2.8822 loss_recon=2.8822 loss_meanflow=0.0000 mean_model_t=0.5052 mean_corrupt_t=0.5052 wrong_frac=0.4950 init_acc_corrupt=0.4714 init_gold_top10=0.4996 init_gold_top100=0.5052
+step=24900 micro_steps=49800 elapsed=34.0s lr=3.000000e-04 loss_all=2.9407 acc_all=0.5778 loss_corrupt=2.9407 acc_corrupt=0.5778 corrupt_frac=1.0000 loss=2.9407 loss_recon=2.9407 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5008 init_acc_corrupt=0.4649 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=25000 micro_steps=50000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8991 acc_all=0.5824 loss_corrupt=2.8991 acc_corrupt=0.5824 corrupt_frac=1.0000 loss=2.8991 loss_recon=2.8991 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4965 init_acc_corrupt=0.4694 init_gold_top10=0.4980 init_gold_top100=0.5036
+step=25100 micro_steps=50200 elapsed=35.4s lr=3.000000e-04 loss_all=2.9248 acc_all=0.5790 loss_corrupt=2.9248 acc_corrupt=0.5790 corrupt_frac=1.0000 loss=2.9248 loss_recon=2.9248 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.4995 init_acc_corrupt=0.4658 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=25200 micro_steps=50400 elapsed=33.9s lr=3.000000e-04 loss_all=2.9324 acc_all=0.5773 loss_corrupt=2.9324 acc_corrupt=0.5773 corrupt_frac=1.0000 loss=2.9324 loss_recon=2.9324 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5025 init_acc_corrupt=0.4626 init_gold_top10=0.4921 init_gold_top100=0.4977
+step=25300 micro_steps=50600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8992 acc_all=0.5832 loss_corrupt=2.8992 acc_corrupt=0.5832 corrupt_frac=1.0000 loss=2.8992 loss_recon=2.8992 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4959 init_acc_corrupt=0.4702 init_gold_top10=0.4985 init_gold_top100=0.5042
+step=25400 micro_steps=50800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9612 acc_all=0.5745 loss_corrupt=2.9612 acc_corrupt=0.5745 corrupt_frac=1.0000 loss=2.9612 loss_recon=2.9612 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5042 init_acc_corrupt=0.4610 init_gold_top10=0.4902 init_gold_top100=0.4960
+step=25500 micro_steps=51000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9444 acc_all=0.5762 loss_corrupt=2.9444 acc_corrupt=0.5762 corrupt_frac=1.0000 loss=2.9444 loss_recon=2.9444 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5032 init_acc_corrupt=0.4624 init_gold_top10=0.4912 init_gold_top100=0.4969
+step=25600 micro_steps=51200 elapsed=34.0s lr=3.000000e-04 loss_all=2.9434 acc_all=0.5770 loss_corrupt=2.9434 acc_corrupt=0.5770 corrupt_frac=1.0000 loss=2.9434 loss_recon=2.9434 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5013 init_acc_corrupt=0.4633 init_gold_top10=0.4930 init_gold_top100=0.4989
+step=25700 micro_steps=51400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9300 acc_all=0.5784 loss_corrupt=2.9300 acc_corrupt=0.5784 corrupt_frac=1.0000 loss=2.9300 loss_recon=2.9300 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5004 init_acc_corrupt=0.4649 init_gold_top10=0.4938 init_gold_top100=0.4998
+step=25800 micro_steps=51600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9059 acc_all=0.5813 loss_corrupt=2.9059 acc_corrupt=0.5813 corrupt_frac=1.0000 loss=2.9059 loss_recon=2.9059 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4986 init_acc_corrupt=0.4674 init_gold_top10=0.4960 init_gold_top100=0.5016
+step=25900 micro_steps=51800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9296 acc_all=0.5786 loss_corrupt=2.9296 acc_corrupt=0.5786 corrupt_frac=1.0000 loss=2.9296 loss_recon=2.9296 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4998 init_acc_corrupt=0.4651 init_gold_top10=0.4945 init_gold_top100=0.5004
+step=26000 micro_steps=52000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8928 acc_all=0.5829 loss_corrupt=2.8928 acc_corrupt=0.5829 corrupt_frac=1.0000 loss=2.8928 loss_recon=2.8928 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4968 init_acc_corrupt=0.4691 init_gold_top10=0.4976 init_gold_top100=0.5034
+step=26100 micro_steps=52200 elapsed=35.2s lr=3.000000e-04 loss_all=2.8987 acc_all=0.5826 loss_corrupt=2.8987 acc_corrupt=0.5826 corrupt_frac=1.0000 loss=2.8987 loss_recon=2.8987 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4963 init_acc_corrupt=0.4692 init_gold_top10=0.4981 init_gold_top100=0.5038
+step=26200 micro_steps=52400 elapsed=33.7s lr=3.000000e-04 loss_all=2.9481 acc_all=0.5763 loss_corrupt=2.9481 acc_corrupt=0.5763 corrupt_frac=1.0000 loss=2.9481 loss_recon=2.9481 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5026 init_acc_corrupt=0.4627 init_gold_top10=0.4917 init_gold_top100=0.4975
+step=26300 micro_steps=52600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9263 acc_all=0.5791 loss_corrupt=2.9263 acc_corrupt=0.5791 corrupt_frac=1.0000 loss=2.9263 loss_recon=2.9263 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4993 init_acc_corrupt=0.4663 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=26400 micro_steps=52800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9040 acc_all=0.5819 loss_corrupt=2.9040 acc_corrupt=0.5819 corrupt_frac=1.0000 loss=2.9040 loss_recon=2.9040 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4975 init_acc_corrupt=0.4680 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=26500 micro_steps=53000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9421 acc_all=0.5765 loss_corrupt=2.9421 acc_corrupt=0.5765 corrupt_frac=1.0000 loss=2.9421 loss_recon=2.9421 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5027 init_acc_corrupt=0.4627 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=26600 micro_steps=53200 elapsed=33.7s lr=3.000000e-04 loss_all=2.9145 acc_all=0.5805 loss_corrupt=2.9145 acc_corrupt=0.5805 corrupt_frac=1.0000 loss=2.9145 loss_recon=2.9145 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4992 init_acc_corrupt=0.4662 init_gold_top10=0.4954 init_gold_top100=0.5010
+step=26700 micro_steps=53400 elapsed=33.9s lr=3.000000e-04 loss_all=2.9081 acc_all=0.5809 loss_corrupt=2.9081 acc_corrupt=0.5809 corrupt_frac=1.0000 loss=2.9081 loss_recon=2.9081 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4995 init_acc_corrupt=0.4665 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=26800 micro_steps=53600 elapsed=33.9s lr=3.000000e-04 loss_all=2.9456 acc_all=0.5761 loss_corrupt=2.9456 acc_corrupt=0.5761 corrupt_frac=1.0000 loss=2.9456 loss_recon=2.9456 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5034 init_acc_corrupt=0.4618 init_gold_top10=0.4908 init_gold_top100=0.4967
+step=26900 micro_steps=53800 elapsed=33.6s lr=3.000000e-04 loss_all=2.9030 acc_all=0.5821 loss_corrupt=2.9030 acc_corrupt=0.5821 corrupt_frac=1.0000 loss=2.9030 loss_recon=2.9030 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4975 init_acc_corrupt=0.4683 init_gold_top10=0.4970 init_gold_top100=0.5026
+step=27000 micro_steps=54000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9338 acc_all=0.5777 loss_corrupt=2.9338 acc_corrupt=0.5777 corrupt_frac=1.0000 loss=2.9338 loss_recon=2.9338 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5018 init_acc_corrupt=0.4641 init_gold_top10=0.4926 init_gold_top100=0.4983
+step=27100 micro_steps=54200 elapsed=35.1s lr=3.000000e-04 loss_all=2.9211 acc_all=0.5792 loss_corrupt=2.9211 acc_corrupt=0.5792 corrupt_frac=1.0000 loss=2.9211 loss_recon=2.9211 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5003 init_acc_corrupt=0.4646 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=27200 micro_steps=54400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9118 acc_all=0.5797 loss_corrupt=2.9118 acc_corrupt=0.5797 corrupt_frac=1.0000 loss=2.9118 loss_recon=2.9118 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5011 init_acc_corrupt=0.4646 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=27300 micro_steps=54600 elapsed=33.9s lr=3.000000e-04 loss_all=2.9227 acc_all=0.5792 loss_corrupt=2.9227 acc_corrupt=0.5792 corrupt_frac=1.0000 loss=2.9227 loss_recon=2.9227 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5003 init_acc_corrupt=0.4655 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=27400 micro_steps=54800 elapsed=33.6s lr=3.000000e-04 loss_all=2.8941 acc_all=0.5829 loss_corrupt=2.8941 acc_corrupt=0.5829 corrupt_frac=1.0000 loss=2.8941 loss_recon=2.8941 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4978 init_acc_corrupt=0.4680 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=27500 micro_steps=55000 elapsed=33.6s lr=3.000000e-04 loss_all=2.9722 acc_all=0.5724 loss_corrupt=2.9722 acc_corrupt=0.5724 corrupt_frac=1.0000 loss=2.9722 loss_recon=2.9722 loss_meanflow=0.0000 mean_model_t=0.4924 mean_corrupt_t=0.4924 wrong_frac=0.5074 init_acc_corrupt=0.4575 init_gold_top10=0.4868 init_gold_top100=0.4927
+step=27600 micro_steps=55200 elapsed=33.6s lr=3.000000e-04 loss_all=2.9094 acc_all=0.5808 loss_corrupt=2.9094 acc_corrupt=0.5808 corrupt_frac=1.0000 loss=2.9094 loss_recon=2.9094 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4995 init_acc_corrupt=0.4661 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=27700 micro_steps=55400 elapsed=33.6s lr=3.000000e-04 loss_all=2.9434 acc_all=0.5762 loss_corrupt=2.9434 acc_corrupt=0.5762 corrupt_frac=1.0000 loss=2.9434 loss_recon=2.9434 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5039 init_acc_corrupt=0.4613 init_gold_top10=0.4906 init_gold_top100=0.4962
+step=27800 micro_steps=55600 elapsed=35.9s lr=3.000000e-04 loss_all=2.9339 acc_all=0.5781 loss_corrupt=2.9339 acc_corrupt=0.5781 corrupt_frac=1.0000 loss=2.9339 loss_recon=2.9339 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5019 init_acc_corrupt=0.4630 init_gold_top10=0.4925 init_gold_top100=0.4983
+step=27900 micro_steps=55800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8993 acc_all=0.5822 loss_corrupt=2.8993 acc_corrupt=0.5822 corrupt_frac=1.0000 loss=2.8993 loss_recon=2.8993 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4977 init_acc_corrupt=0.4684 init_gold_top10=0.4967 init_gold_top100=0.5025
+step=28000 micro_steps=56000 elapsed=33.7s lr=3.000000e-04 loss_all=2.9378 acc_all=0.5767 loss_corrupt=2.9378 acc_corrupt=0.5767 corrupt_frac=1.0000 loss=2.9378 loss_recon=2.9378 loss_meanflow=0.0000 mean_model_t=0.4962 mean_corrupt_t=0.4962 wrong_frac=0.5039 init_acc_corrupt=0.4615 init_gold_top10=0.4905 init_gold_top100=0.4962
+step=28100 micro_steps=56200 elapsed=35.2s lr=3.000000e-04 loss_all=2.9245 acc_all=0.5787 loss_corrupt=2.9245 acc_corrupt=0.5787 corrupt_frac=1.0000 loss=2.9245 loss_recon=2.9245 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.5000 init_acc_corrupt=0.4656 init_gold_top10=0.4943 init_gold_top100=0.5000
+step=28200 micro_steps=56400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9038 acc_all=0.5812 loss_corrupt=2.9038 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9038 loss_recon=2.9038 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4993 init_acc_corrupt=0.4665 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=28300 micro_steps=56600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9257 acc_all=0.5788 loss_corrupt=2.9257 acc_corrupt=0.5788 corrupt_frac=1.0000 loss=2.9257 loss_recon=2.9257 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5016 init_acc_corrupt=0.4639 init_gold_top10=0.4929 init_gold_top100=0.4985
+step=28400 micro_steps=56800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9072 acc_all=0.5811 loss_corrupt=2.9072 acc_corrupt=0.5811 corrupt_frac=1.0000 loss=2.9072 loss_recon=2.9072 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4988 init_acc_corrupt=0.4667 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=28500 micro_steps=57000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8920 acc_all=0.5828 loss_corrupt=2.8920 acc_corrupt=0.5828 corrupt_frac=1.0000 loss=2.8920 loss_recon=2.8920 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4977 init_acc_corrupt=0.4682 init_gold_top10=0.4968 init_gold_top100=0.5024
+step=28600 micro_steps=57200 elapsed=33.8s lr=3.000000e-04 loss_all=2.9155 acc_all=0.5797 loss_corrupt=2.9155 acc_corrupt=0.5797 corrupt_frac=1.0000 loss=2.9155 loss_recon=2.9155 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4997 init_acc_corrupt=0.4661 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=28700 micro_steps=57400 elapsed=33.7s lr=3.000000e-04 loss_all=2.9216 acc_all=0.5794 loss_corrupt=2.9216 acc_corrupt=0.5794 corrupt_frac=1.0000 loss=2.9216 loss_recon=2.9216 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5006 init_acc_corrupt=0.4647 init_gold_top10=0.4936 init_gold_top100=0.4994
+step=28800 micro_steps=57600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9294 acc_all=0.5777 loss_corrupt=2.9294 acc_corrupt=0.5777 corrupt_frac=1.0000 loss=2.9294 loss_recon=2.9294 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5016 init_acc_corrupt=0.4632 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=28900 micro_steps=57800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9608 acc_all=0.5740 loss_corrupt=2.9608 acc_corrupt=0.5740 corrupt_frac=1.0000 loss=2.9608 loss_recon=2.9608 loss_meanflow=0.0000 mean_model_t=0.4939 mean_corrupt_t=0.4939 wrong_frac=0.5063 init_acc_corrupt=0.4587 init_gold_top10=0.4880 init_gold_top100=0.4939
+step=29000 micro_steps=58000 elapsed=34.0s lr=3.000000e-04 loss_all=2.9064 acc_all=0.5812 loss_corrupt=2.9064 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9064 loss_recon=2.9064 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4995 init_acc_corrupt=0.4671 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=29100 micro_steps=58200 elapsed=35.5s lr=3.000000e-04 loss_all=2.8779 acc_all=0.5850 loss_corrupt=2.8779 acc_corrupt=0.5850 corrupt_frac=1.0000 loss=2.8779 loss_recon=2.8779 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4965 init_acc_corrupt=0.4696 init_gold_top10=0.4981 init_gold_top100=0.5037
+step=29200 micro_steps=58400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9091 acc_all=0.5807 loss_corrupt=2.9091 acc_corrupt=0.5807 corrupt_frac=1.0000 loss=2.9091 loss_recon=2.9091 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4990 init_acc_corrupt=0.4666 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=29300 micro_steps=58600 elapsed=33.9s lr=3.000000e-04 loss_all=2.9364 acc_all=0.5774 loss_corrupt=2.9364 acc_corrupt=0.5774 corrupt_frac=1.0000 loss=2.9364 loss_recon=2.9364 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5026 init_acc_corrupt=0.4623 init_gold_top10=0.4917 init_gold_top100=0.4975
+step=29400 micro_steps=58800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8942 acc_all=0.5833 loss_corrupt=2.8942 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8942 loss_recon=2.8942 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4968 init_acc_corrupt=0.4689 init_gold_top10=0.4976 init_gold_top100=0.5033
+step=29500 micro_steps=59000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8786 acc_all=0.5849 loss_corrupt=2.8786 acc_corrupt=0.5849 corrupt_frac=1.0000 loss=2.8786 loss_recon=2.8786 loss_meanflow=0.0000 mean_model_t=0.5042 mean_corrupt_t=0.5042 wrong_frac=0.4961 init_acc_corrupt=0.4700 init_gold_top10=0.4984 init_gold_top100=0.5041
+step=29600 micro_steps=59200 elapsed=33.9s lr=3.000000e-04 loss_all=2.9352 acc_all=0.5769 loss_corrupt=2.9352 acc_corrupt=0.5769 corrupt_frac=1.0000 loss=2.9352 loss_recon=2.9352 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5037 init_acc_corrupt=0.4610 init_gold_top10=0.4907 init_gold_top100=0.4964
+step=29700 micro_steps=59400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8752 acc_all=0.5852 loss_corrupt=2.8752 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8752 loss_recon=2.8752 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4957 init_acc_corrupt=0.4700 init_gold_top10=0.4989 init_gold_top100=0.5044
+step=29800 micro_steps=59600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8725 acc_all=0.5852 loss_corrupt=2.8725 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8725 loss_recon=2.8725 loss_meanflow=0.0000 mean_model_t=0.5039 mean_corrupt_t=0.5039 wrong_frac=0.4964 init_acc_corrupt=0.4698 init_gold_top10=0.4981 init_gold_top100=0.5038
+step=29900 micro_steps=59800 elapsed=33.9s lr=3.000000e-04 loss_all=2.8916 acc_all=0.5828 loss_corrupt=2.8916 acc_corrupt=0.5828 corrupt_frac=1.0000 loss=2.8916 loss_recon=2.8916 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4977 init_acc_corrupt=0.4680 init_gold_top10=0.4968 init_gold_top100=0.5024
+step=30000 micro_steps=60000 elapsed=33.9s lr=3.000000e-04 loss_all=2.9401 acc_all=0.5761 loss_corrupt=2.9401 acc_corrupt=0.5761 corrupt_frac=1.0000 loss=2.9401 loss_recon=2.9401 loss_meanflow=0.0000 mean_model_t=0.4950 mean_corrupt_t=0.4950 wrong_frac=0.5051 init_acc_corrupt=0.4606 init_gold_top10=0.4892 init_gold_top100=0.4950
+step=30100 micro_steps=60200 elapsed=35.4s lr=3.000000e-04 loss_all=2.9287 acc_all=0.5782 loss_corrupt=2.9287 acc_corrupt=0.5782 corrupt_frac=1.0000 loss=2.9287 loss_recon=2.9287 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5022 init_acc_corrupt=0.4626 init_gold_top10=0.4921 init_gold_top100=0.4979
+step=30200 micro_steps=60400 elapsed=33.7s lr=3.000000e-04 loss_all=2.9404 acc_all=0.5766 loss_corrupt=2.9404 acc_corrupt=0.5766 corrupt_frac=1.0000 loss=2.9404 loss_recon=2.9404 loss_meanflow=0.0000 mean_model_t=0.4959 mean_corrupt_t=0.4959 wrong_frac=0.5041 init_acc_corrupt=0.4610 init_gold_top10=0.4903 init_gold_top100=0.4962
+step=30300 micro_steps=60600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9028 acc_all=0.5815 loss_corrupt=2.9028 acc_corrupt=0.5815 corrupt_frac=1.0000 loss=2.9028 loss_recon=2.9028 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4998 init_acc_corrupt=0.4658 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=30400 micro_steps=60800 elapsed=33.9s lr=3.000000e-04 loss_all=2.8985 acc_all=0.5821 loss_corrupt=2.8985 acc_corrupt=0.5821 corrupt_frac=1.0000 loss=2.8985 loss_recon=2.8985 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4989 init_acc_corrupt=0.4668 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=30500 micro_steps=61000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8914 acc_all=0.5828 loss_corrupt=2.8914 acc_corrupt=0.5828 corrupt_frac=1.0000 loss=2.8914 loss_recon=2.8914 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4978 init_acc_corrupt=0.4674 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=30600 micro_steps=61200 elapsed=33.7s lr=3.000000e-04 loss_all=2.9096 acc_all=0.5807 loss_corrupt=2.9096 acc_corrupt=0.5807 corrupt_frac=1.0000 loss=2.9096 loss_recon=2.9096 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5012 init_acc_corrupt=0.4642 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=30700 micro_steps=61400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8928 acc_all=0.5830 loss_corrupt=2.8928 acc_corrupt=0.5830 corrupt_frac=1.0000 loss=2.8928 loss_recon=2.8928 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4981 init_acc_corrupt=0.4667 init_gold_top10=0.4962 init_gold_top100=0.5021
+step=30800 micro_steps=61600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9272 acc_all=0.5775 loss_corrupt=2.9272 acc_corrupt=0.5775 corrupt_frac=1.0000 loss=2.9272 loss_recon=2.9272 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5036 init_acc_corrupt=0.4617 init_gold_top10=0.4908 init_gold_top100=0.4965
+step=30900 micro_steps=61800 elapsed=33.9s lr=3.000000e-04 loss_all=2.9146 acc_all=0.5801 loss_corrupt=2.9146 acc_corrupt=0.5801 corrupt_frac=1.0000 loss=2.9146 loss_recon=2.9146 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5007 init_acc_corrupt=0.4649 init_gold_top10=0.4936 init_gold_top100=0.4995
+step=31000 micro_steps=62000 elapsed=33.7s lr=3.000000e-04 loss_all=2.9139 acc_all=0.5799 loss_corrupt=2.9139 acc_corrupt=0.5799 corrupt_frac=1.0000 loss=2.9139 loss_recon=2.9139 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5007 init_acc_corrupt=0.4650 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=31100 micro_steps=62200 elapsed=35.4s lr=3.000000e-04 loss_all=2.9268 acc_all=0.5785 loss_corrupt=2.9268 acc_corrupt=0.5785 corrupt_frac=1.0000 loss=2.9268 loss_recon=2.9268 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5023 init_acc_corrupt=0.4630 init_gold_top10=0.4922 init_gold_top100=0.4978
+step=31200 micro_steps=62400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9040 acc_all=0.5817 loss_corrupt=2.9040 acc_corrupt=0.5817 corrupt_frac=1.0000 loss=2.9040 loss_recon=2.9040 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4989 init_acc_corrupt=0.4663 init_gold_top10=0.4955 init_gold_top100=0.5013
+step=31300 micro_steps=62600 elapsed=33.9s lr=3.000000e-04 loss_all=2.9235 acc_all=0.5791 loss_corrupt=2.9235 acc_corrupt=0.5791 corrupt_frac=1.0000 loss=2.9235 loss_recon=2.9235 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5014 init_acc_corrupt=0.4643 init_gold_top10=0.4929 init_gold_top100=0.4988
+step=31400 micro_steps=62800 elapsed=33.9s lr=3.000000e-04 loss_all=2.9277 acc_all=0.5785 loss_corrupt=2.9277 acc_corrupt=0.5785 corrupt_frac=1.0000 loss=2.9277 loss_recon=2.9277 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5017 init_acc_corrupt=0.4637 init_gold_top10=0.4926 init_gold_top100=0.4984
+step=31500 micro_steps=63000 elapsed=33.9s lr=3.000000e-04 loss_all=2.9140 acc_all=0.5801 loss_corrupt=2.9140 acc_corrupt=0.5801 corrupt_frac=1.0000 loss=2.9140 loss_recon=2.9140 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5010 init_acc_corrupt=0.4648 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=31600 micro_steps=63200 elapsed=33.7s lr=3.000000e-04 loss_all=2.9050 acc_all=0.5807 loss_corrupt=2.9050 acc_corrupt=0.5807 corrupt_frac=1.0000 loss=2.9050 loss_recon=2.9050 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5005 init_acc_corrupt=0.4651 init_gold_top10=0.4940 init_gold_top100=0.4996
+step=31700 micro_steps=63400 elapsed=34.1s lr=3.000000e-04 loss_all=2.8881 acc_all=0.5832 loss_corrupt=2.8881 acc_corrupt=0.5832 corrupt_frac=1.0000 loss=2.8881 loss_recon=2.8881 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4990 init_acc_corrupt=0.4669 init_gold_top10=0.4955 init_gold_top100=0.5011
+step=31800 micro_steps=63600 elapsed=34.1s lr=3.000000e-04 loss_all=2.9046 acc_all=0.5807 loss_corrupt=2.9046 acc_corrupt=0.5807 corrupt_frac=1.0000 loss=2.9046 loss_recon=2.9046 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5008 init_acc_corrupt=0.4649 init_gold_top10=0.4936 init_gold_top100=0.4994
+step=31900 micro_steps=63800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8858 acc_all=0.5840 loss_corrupt=2.8858 acc_corrupt=0.5840 corrupt_frac=1.0000 loss=2.8858 loss_recon=2.8858 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4969 init_acc_corrupt=0.4689 init_gold_top10=0.4976 init_gold_top100=0.5033
+step=32000 micro_steps=64000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9229 acc_all=0.5792 loss_corrupt=2.9229 acc_corrupt=0.5792 corrupt_frac=1.0000 loss=2.9229 loss_recon=2.9229 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5028 init_acc_corrupt=0.4629 init_gold_top10=0.4917 init_gold_top100=0.4974
+step=32100 micro_steps=64200 elapsed=35.5s lr=3.000000e-04 loss_all=2.9026 acc_all=0.5814 loss_corrupt=2.9026 acc_corrupt=0.5814 corrupt_frac=1.0000 loss=2.9026 loss_recon=2.9026 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4999 init_acc_corrupt=0.4657 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=32200 micro_steps=64400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8725 acc_all=0.5853 loss_corrupt=2.8725 acc_corrupt=0.5853 corrupt_frac=1.0000 loss=2.8725 loss_recon=2.8725 loss_meanflow=0.0000 mean_model_t=0.5049 mean_corrupt_t=0.5049 wrong_frac=0.4950 init_acc_corrupt=0.4704 init_gold_top10=0.4992 init_gold_top100=0.5050
+step=32300 micro_steps=64600 elapsed=33.6s lr=3.000000e-04 loss_all=2.9182 acc_all=0.5797 loss_corrupt=2.9182 acc_corrupt=0.5797 corrupt_frac=1.0000 loss=2.9182 loss_recon=2.9182 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5018 init_acc_corrupt=0.4640 init_gold_top10=0.4925 init_gold_top100=0.4983
+step=32400 micro_steps=64800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8866 acc_all=0.5835 loss_corrupt=2.8866 acc_corrupt=0.5835 corrupt_frac=1.0000 loss=2.8866 loss_recon=2.8866 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4976 init_acc_corrupt=0.4680 init_gold_top10=0.4968 init_gold_top100=0.5025
+step=32500 micro_steps=65000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9137 acc_all=0.5797 loss_corrupt=2.9137 acc_corrupt=0.5797 corrupt_frac=1.0000 loss=2.9137 loss_recon=2.9137 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5013 init_acc_corrupt=0.4640 init_gold_top10=0.4931 init_gold_top100=0.4988
+step=32600 micro_steps=65200 elapsed=65.6s lr=3.000000e-04 loss_all=2.9294 acc_all=0.5777 loss_corrupt=2.9294 acc_corrupt=0.5777 corrupt_frac=1.0000 loss=2.9294 loss_recon=2.9294 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5029 init_acc_corrupt=0.4618 init_gold_top10=0.4912 init_gold_top100=0.4972
+step=32700 micro_steps=65400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8620 acc_all=0.5869 loss_corrupt=2.8620 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8620 loss_recon=2.8620 loss_meanflow=0.0000 mean_model_t=0.5057 mean_corrupt_t=0.5057 wrong_frac=0.4940 init_acc_corrupt=0.4720 init_gold_top10=0.5005 init_gold_top100=0.5062
+step=32800 micro_steps=65600 elapsed=33.6s lr=3.000000e-04 loss_all=2.9194 acc_all=0.5788 loss_corrupt=2.9194 acc_corrupt=0.5788 corrupt_frac=1.0000 loss=2.9194 loss_recon=2.9194 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5025 init_acc_corrupt=0.4628 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=32900 micro_steps=65800 elapsed=38.4s lr=3.000000e-04 loss_all=2.8956 acc_all=0.5823 loss_corrupt=2.8956 acc_corrupt=0.5823 corrupt_frac=1.0000 loss=2.8956 loss_recon=2.8956 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4993 init_acc_corrupt=0.4665 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=33000 micro_steps=66000 elapsed=38.3s lr=3.000000e-04 loss_all=2.9242 acc_all=0.5785 loss_corrupt=2.9242 acc_corrupt=0.5785 corrupt_frac=1.0000 loss=2.9242 loss_recon=2.9242 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5029 init_acc_corrupt=0.4618 init_gold_top10=0.4915 init_gold_top100=0.4972
+step=33100 micro_steps=66200 elapsed=35.4s lr=3.000000e-04 loss_all=2.9296 acc_all=0.5784 loss_corrupt=2.9296 acc_corrupt=0.5784 corrupt_frac=1.0000 loss=2.9296 loss_recon=2.9296 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5024 init_acc_corrupt=0.4618 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=33200 micro_steps=66400 elapsed=33.6s lr=3.000000e-04 loss_all=2.9342 acc_all=0.5774 loss_corrupt=2.9342 acc_corrupt=0.5774 corrupt_frac=1.0000 loss=2.9342 loss_recon=2.9342 loss_meanflow=0.0000 mean_model_t=0.4961 mean_corrupt_t=0.4961 wrong_frac=0.5044 init_acc_corrupt=0.4608 init_gold_top10=0.4899 init_gold_top100=0.4958
+step=33300 micro_steps=66600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9092 acc_all=0.5804 loss_corrupt=2.9092 acc_corrupt=0.5804 corrupt_frac=1.0000 loss=2.9092 loss_recon=2.9092 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4644 init_gold_top10=0.4936 init_gold_top100=0.4995
+step=33400 micro_steps=66800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9074 acc_all=0.5808 loss_corrupt=2.9074 acc_corrupt=0.5808 corrupt_frac=1.0000 loss=2.9074 loss_recon=2.9074 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4649 init_gold_top10=0.4935 init_gold_top100=0.4994
+step=33500 micro_steps=67000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9012 acc_all=0.5817 loss_corrupt=2.9012 acc_corrupt=0.5817 corrupt_frac=1.0000 loss=2.9012 loss_recon=2.9012 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.5002 init_acc_corrupt=0.4660 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=33600 micro_steps=67200 elapsed=33.5s lr=3.000000e-04 loss_all=2.9013 acc_all=0.5819 loss_corrupt=2.9013 acc_corrupt=0.5819 corrupt_frac=1.0000 loss=2.9013 loss_recon=2.9013 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5002 init_acc_corrupt=0.4652 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=33700 micro_steps=67400 elapsed=33.6s lr=3.000000e-04 loss_all=2.9479 acc_all=0.5754 loss_corrupt=2.9479 acc_corrupt=0.5754 corrupt_frac=1.0000 loss=2.9479 loss_recon=2.9479 loss_meanflow=0.0000 mean_model_t=0.4947 mean_corrupt_t=0.4947 wrong_frac=0.5053 init_acc_corrupt=0.4593 init_gold_top10=0.4889 init_gold_top100=0.4948
+step=33800 micro_steps=67600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8592 acc_all=0.5872 loss_corrupt=2.8592 acc_corrupt=0.5872 corrupt_frac=1.0000 loss=2.8592 loss_recon=2.8592 loss_meanflow=0.0000 mean_model_t=0.5051 mean_corrupt_t=0.5051 wrong_frac=0.4951 init_acc_corrupt=0.4710 init_gold_top10=0.4996 init_gold_top100=0.5051
+step=33900 micro_steps=67800 elapsed=33.6s lr=3.000000e-04 loss_all=2.8666 acc_all=0.5859 loss_corrupt=2.8666 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8666 loss_recon=2.8666 loss_meanflow=0.0000 mean_model_t=0.5042 mean_corrupt_t=0.5042 wrong_frac=0.4957 init_acc_corrupt=0.4706 init_gold_top10=0.4987 init_gold_top100=0.5044
+step=34000 micro_steps=68000 elapsed=33.6s lr=3.000000e-04 loss_all=2.9419 acc_all=0.5756 loss_corrupt=2.9419 acc_corrupt=0.5756 corrupt_frac=1.0000 loss=2.9419 loss_recon=2.9419 loss_meanflow=0.0000 mean_model_t=0.4947 mean_corrupt_t=0.4947 wrong_frac=0.5051 init_acc_corrupt=0.4594 init_gold_top10=0.4891 init_gold_top100=0.4950
+step=34100 micro_steps=68200 elapsed=35.0s lr=3.000000e-04 loss_all=2.8787 acc_all=0.5843 loss_corrupt=2.8787 acc_corrupt=0.5843 corrupt_frac=1.0000 loss=2.8787 loss_recon=2.8787 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4979 init_acc_corrupt=0.4677 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=34200 micro_steps=68400 elapsed=33.5s lr=3.000000e-04 loss_all=2.9145 acc_all=0.5799 loss_corrupt=2.9145 acc_corrupt=0.5799 corrupt_frac=1.0000 loss=2.9145 loss_recon=2.9145 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5020 init_acc_corrupt=0.4630 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=34300 micro_steps=68600 elapsed=33.6s lr=3.000000e-04 loss_all=2.8946 acc_all=0.5826 loss_corrupt=2.8946 acc_corrupt=0.5826 corrupt_frac=1.0000 loss=2.8946 loss_recon=2.8946 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4990 init_acc_corrupt=0.4661 init_gold_top10=0.4954 init_gold_top100=0.5010
+step=34400 micro_steps=68800 elapsed=33.6s lr=3.000000e-04 loss_all=2.8865 acc_all=0.5831 loss_corrupt=2.8865 acc_corrupt=0.5831 corrupt_frac=1.0000 loss=2.8865 loss_recon=2.8865 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4996 init_acc_corrupt=0.4661 init_gold_top10=0.4949 init_gold_top100=0.5005
+step=34500 micro_steps=69000 elapsed=33.4s lr=3.000000e-04 loss_all=2.9444 acc_all=0.5760 loss_corrupt=2.9444 acc_corrupt=0.5760 corrupt_frac=1.0000 loss=2.9444 loss_recon=2.9444 loss_meanflow=0.0000 mean_model_t=0.4952 mean_corrupt_t=0.4952 wrong_frac=0.5054 init_acc_corrupt=0.4591 init_gold_top10=0.4889 init_gold_top100=0.4947
+step=34600 micro_steps=69200 elapsed=33.5s lr=3.000000e-04 loss_all=2.9048 acc_all=0.5815 loss_corrupt=2.9048 acc_corrupt=0.5815 corrupt_frac=1.0000 loss=2.9048 loss_recon=2.9048 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4998 init_acc_corrupt=0.4658 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=34700 micro_steps=69400 elapsed=33.6s lr=3.000000e-04 loss_all=2.9326 acc_all=0.5771 loss_corrupt=2.9326 acc_corrupt=0.5771 corrupt_frac=1.0000 loss=2.9326 loss_recon=2.9326 loss_meanflow=0.0000 mean_model_t=0.4950 mean_corrupt_t=0.4950 wrong_frac=0.5049 init_acc_corrupt=0.4603 init_gold_top10=0.4895 init_gold_top100=0.4952
+step=34800 micro_steps=69600 elapsed=33.6s lr=3.000000e-04 loss_all=2.8908 acc_all=0.5827 loss_corrupt=2.8908 acc_corrupt=0.5827 corrupt_frac=1.0000 loss=2.8908 loss_recon=2.8908 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4991 init_acc_corrupt=0.4661 init_gold_top10=0.4955 init_gold_top100=0.5011
+step=34900 micro_steps=69800 elapsed=33.6s lr=3.000000e-04 loss_all=2.9138 acc_all=0.5801 loss_corrupt=2.9138 acc_corrupt=0.5801 corrupt_frac=1.0000 loss=2.9138 loss_recon=2.9138 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5016 init_acc_corrupt=0.4637 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=35000 micro_steps=70000 elapsed=33.5s lr=3.000000e-04 loss_all=2.8697 acc_all=0.5853 loss_corrupt=2.8697 acc_corrupt=0.5853 corrupt_frac=1.0000 loss=2.8697 loss_recon=2.8697 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4972 init_acc_corrupt=0.4684 init_gold_top10=0.4972 init_gold_top100=0.5030
+step=35100 micro_steps=70200 elapsed=35.0s lr=3.000000e-04 loss_all=2.8959 acc_all=0.5818 loss_corrupt=2.8959 acc_corrupt=0.5818 corrupt_frac=1.0000 loss=2.8959 loss_recon=2.8959 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4652 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=35200 micro_steps=70400 elapsed=33.5s lr=3.000000e-04 loss_all=2.9290 acc_all=0.5775 loss_corrupt=2.9290 acc_corrupt=0.5775 corrupt_frac=1.0000 loss=2.9290 loss_recon=2.9290 loss_meanflow=0.0000 mean_model_t=0.4951 mean_corrupt_t=0.4951 wrong_frac=0.5052 init_acc_corrupt=0.4601 init_gold_top10=0.4892 init_gold_top100=0.4949
+step=35300 micro_steps=70600 elapsed=33.6s lr=3.000000e-04 loss_all=2.8679 acc_all=0.5859 loss_corrupt=2.8679 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8679 loss_recon=2.8679 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4966 init_acc_corrupt=0.4691 init_gold_top10=0.4977 init_gold_top100=0.5035
+step=35400 micro_steps=70800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8889 acc_all=0.5830 loss_corrupt=2.8889 acc_corrupt=0.5830 corrupt_frac=1.0000 loss=2.8889 loss_recon=2.8889 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4988 init_acc_corrupt=0.4661 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=35500 micro_steps=71000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8907 acc_all=0.5828 loss_corrupt=2.8907 acc_corrupt=0.5828 corrupt_frac=1.0000 loss=2.8907 loss_recon=2.8907 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4993 init_acc_corrupt=0.4664 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=35600 micro_steps=71200 elapsed=33.9s lr=3.000000e-04 loss_all=2.9415 acc_all=0.5758 loss_corrupt=2.9415 acc_corrupt=0.5758 corrupt_frac=1.0000 loss=2.9415 loss_recon=2.9415 loss_meanflow=0.0000 mean_model_t=0.4949 mean_corrupt_t=0.4949 wrong_frac=0.5054 init_acc_corrupt=0.4591 init_gold_top10=0.4889 init_gold_top100=0.4947
+step=35700 micro_steps=71400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8779 acc_all=0.5847 loss_corrupt=2.8779 acc_corrupt=0.5847 corrupt_frac=1.0000 loss=2.8779 loss_recon=2.8779 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4976 init_acc_corrupt=0.4685 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=35800 micro_steps=71600 elapsed=33.6s lr=3.000000e-04 loss_all=2.9035 acc_all=0.5812 loss_corrupt=2.9035 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9035 loss_recon=2.9035 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5009 init_acc_corrupt=0.4647 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=35900 micro_steps=71800 elapsed=33.6s lr=3.000000e-04 loss_all=2.8951 acc_all=0.5821 loss_corrupt=2.8951 acc_corrupt=0.5821 corrupt_frac=1.0000 loss=2.8951 loss_recon=2.8951 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5008 init_acc_corrupt=0.4645 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=36000 micro_steps=72000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8818 acc_all=0.5835 loss_corrupt=2.8818 acc_corrupt=0.5835 corrupt_frac=1.0000 loss=2.8818 loss_recon=2.8818 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4999 init_acc_corrupt=0.4659 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=36100 micro_steps=72200 elapsed=35.3s lr=3.000000e-04 loss_all=2.9141 acc_all=0.5795 loss_corrupt=2.9141 acc_corrupt=0.5795 corrupt_frac=1.0000 loss=2.9141 loss_recon=2.9141 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5031 init_acc_corrupt=0.4619 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=36200 micro_steps=72400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8897 acc_all=0.5828 loss_corrupt=2.8897 acc_corrupt=0.5828 corrupt_frac=1.0000 loss=2.8897 loss_recon=2.8897 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4991 init_acc_corrupt=0.4662 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=36300 micro_steps=72600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8802 acc_all=0.5839 loss_corrupt=2.8802 acc_corrupt=0.5839 corrupt_frac=1.0000 loss=2.8802 loss_recon=2.8802 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4990 init_acc_corrupt=0.4664 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=36400 micro_steps=72800 elapsed=33.7s lr=3.000000e-04 loss_all=2.9170 acc_all=0.5794 loss_corrupt=2.9170 acc_corrupt=0.5794 corrupt_frac=1.0000 loss=2.9170 loss_recon=2.9170 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5025 init_acc_corrupt=0.4622 init_gold_top10=0.4918 init_gold_top100=0.4977
+step=36500 micro_steps=73000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9285 acc_all=0.5783 loss_corrupt=2.9285 acc_corrupt=0.5783 corrupt_frac=1.0000 loss=2.9285 loss_recon=2.9285 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5035 init_acc_corrupt=0.4614 init_gold_top10=0.4908 init_gold_top100=0.4966
+step=36600 micro_steps=73200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8842 acc_all=0.5834 loss_corrupt=2.8842 acc_corrupt=0.5834 corrupt_frac=1.0000 loss=2.8842 loss_recon=2.8842 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4988 init_acc_corrupt=0.4669 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=36700 micro_steps=73400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8934 acc_all=0.5829 loss_corrupt=2.8934 acc_corrupt=0.5829 corrupt_frac=1.0000 loss=2.8934 loss_recon=2.8934 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4995 init_acc_corrupt=0.4664 init_gold_top10=0.4947 init_gold_top100=0.5006
+step=36800 micro_steps=73600 elapsed=34.0s lr=3.000000e-04 loss_all=2.8399 acc_all=0.5896 loss_corrupt=2.8399 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8399 loss_recon=2.8399 loss_meanflow=0.0000 mean_model_t=0.5064 mean_corrupt_t=0.5064 wrong_frac=0.4932 init_acc_corrupt=0.4725 init_gold_top10=0.5015 init_gold_top100=0.5069
+step=36900 micro_steps=73800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8493 acc_all=0.5881 loss_corrupt=2.8493 acc_corrupt=0.5881 corrupt_frac=1.0000 loss=2.8493 loss_recon=2.8493 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5045 wrong_frac=0.4951 init_acc_corrupt=0.4707 init_gold_top10=0.4995 init_gold_top100=0.5052
+step=37000 micro_steps=74000 elapsed=33.5s lr=3.000000e-04 loss_all=2.9193 acc_all=0.5786 loss_corrupt=2.9193 acc_corrupt=0.5786 corrupt_frac=1.0000 loss=2.9193 loss_recon=2.9193 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5038 init_acc_corrupt=0.4616 init_gold_top10=0.4905 init_gold_top100=0.4963
+step=37100 micro_steps=74200 elapsed=35.1s lr=3.000000e-04 loss_all=2.9130 acc_all=0.5797 loss_corrupt=2.9130 acc_corrupt=0.5797 corrupt_frac=1.0000 loss=2.9130 loss_recon=2.9130 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5018 init_acc_corrupt=0.4631 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=37200 micro_steps=74400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8666 acc_all=0.5862 loss_corrupt=2.8666 acc_corrupt=0.5862 corrupt_frac=1.0000 loss=2.8666 loss_recon=2.8666 loss_meanflow=0.0000 mean_model_t=0.5043 mean_corrupt_t=0.5043 wrong_frac=0.4957 init_acc_corrupt=0.4698 init_gold_top10=0.4987 init_gold_top100=0.5044
+step=37300 micro_steps=74600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8593 acc_all=0.5870 loss_corrupt=2.8593 acc_corrupt=0.5870 corrupt_frac=1.0000 loss=2.8593 loss_recon=2.8593 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4961 init_acc_corrupt=0.4693 init_gold_top10=0.4985 init_gold_top100=0.5042
+step=37400 micro_steps=74800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8734 acc_all=0.5851 loss_corrupt=2.8734 acc_corrupt=0.5851 corrupt_frac=1.0000 loss=2.8734 loss_recon=2.8734 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4980 init_acc_corrupt=0.4676 init_gold_top10=0.4965 init_gold_top100=0.5021
+step=37500 micro_steps=75000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8512 acc_all=0.5876 loss_corrupt=2.8512 acc_corrupt=0.5876 corrupt_frac=1.0000 loss=2.8512 loss_recon=2.8512 loss_meanflow=0.0000 mean_model_t=0.5043 mean_corrupt_t=0.5043 wrong_frac=0.4956 init_acc_corrupt=0.4704 init_gold_top10=0.4988 init_gold_top100=0.5045
+step=37600 micro_steps=75200 elapsed=33.7s lr=3.000000e-04 loss_all=2.9198 acc_all=0.5791 loss_corrupt=2.9198 acc_corrupt=0.5791 corrupt_frac=1.0000 loss=2.9198 loss_recon=2.9198 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5027 init_acc_corrupt=0.4620 init_gold_top10=0.4916 init_gold_top100=0.4974
+step=37700 micro_steps=75400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8969 acc_all=0.5822 loss_corrupt=2.8969 acc_corrupt=0.5822 corrupt_frac=1.0000 loss=2.8969 loss_recon=2.8969 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5002 init_acc_corrupt=0.4648 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=37800 micro_steps=75600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8692 acc_all=0.5856 loss_corrupt=2.8692 acc_corrupt=0.5856 corrupt_frac=1.0000 loss=2.8692 loss_recon=2.8692 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4973 init_acc_corrupt=0.4687 init_gold_top10=0.4971 init_gold_top100=0.5028
+step=37900 micro_steps=75800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9026 acc_all=0.5812 loss_corrupt=2.9026 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9026 loss_recon=2.9026 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5015 init_acc_corrupt=0.4640 init_gold_top10=0.4930 init_gold_top100=0.4986
+step=38000 micro_steps=76000 elapsed=33.6s lr=3.000000e-04 loss_all=2.8958 acc_all=0.5820 loss_corrupt=2.8958 acc_corrupt=0.5820 corrupt_frac=1.0000 loss=2.8958 loss_recon=2.8958 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.4999 init_acc_corrupt=0.4648 init_gold_top10=0.4943 init_gold_top100=0.5002
+step=38100 micro_steps=76200 elapsed=35.2s lr=3.000000e-04 loss_all=2.8767 acc_all=0.5844 loss_corrupt=2.8767 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8767 loss_recon=2.8767 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4986 init_acc_corrupt=0.4670 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=38200 micro_steps=76400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8953 acc_all=0.5820 loss_corrupt=2.8953 acc_corrupt=0.5820 corrupt_frac=1.0000 loss=2.8953 loss_recon=2.8953 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5000 init_acc_corrupt=0.4652 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=38300 micro_steps=76600 elapsed=33.6s lr=3.000000e-04 loss_all=2.9062 acc_all=0.5809 loss_corrupt=2.9062 acc_corrupt=0.5809 corrupt_frac=1.0000 loss=2.9062 loss_recon=2.9062 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5014 init_acc_corrupt=0.4637 init_gold_top10=0.4928 init_gold_top100=0.4987
+step=38400 micro_steps=76800 elapsed=33.5s lr=3.000000e-04 loss_all=2.9170 acc_all=0.5793 loss_corrupt=2.9170 acc_corrupt=0.5793 corrupt_frac=1.0000 loss=2.9170 loss_recon=2.9170 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5031 init_acc_corrupt=0.4616 init_gold_top10=0.4912 init_gold_top100=0.4971
+step=38500 micro_steps=77000 elapsed=33.4s lr=3.000000e-04 loss_all=2.8936 acc_all=0.5820 loss_corrupt=2.8936 acc_corrupt=0.5820 corrupt_frac=1.0000 loss=2.8936 loss_recon=2.8936 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5006 init_acc_corrupt=0.4652 init_gold_top10=0.4939 init_gold_top100=0.4995
+step=38600 micro_steps=77200 elapsed=33.6s lr=3.000000e-04 loss_all=2.8979 acc_all=0.5821 loss_corrupt=2.8979 acc_corrupt=0.5821 corrupt_frac=1.0000 loss=2.8979 loss_recon=2.8979 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4651 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=38700 micro_steps=77400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8937 acc_all=0.5817 loss_corrupt=2.8937 acc_corrupt=0.5817 corrupt_frac=1.0000 loss=2.8937 loss_recon=2.8937 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5008 init_acc_corrupt=0.4646 init_gold_top10=0.4936 init_gold_top100=0.4994
+step=38800 micro_steps=77600 elapsed=33.5s lr=3.000000e-04 loss_all=2.8694 acc_all=0.5857 loss_corrupt=2.8694 acc_corrupt=0.5857 corrupt_frac=1.0000 loss=2.8694 loss_recon=2.8694 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4969 init_acc_corrupt=0.4686 init_gold_top10=0.4975 init_gold_top100=0.5032
+step=38900 micro_steps=77800 elapsed=33.6s lr=3.000000e-04 loss_all=2.9054 acc_all=0.5805 loss_corrupt=2.9054 acc_corrupt=0.5805 corrupt_frac=1.0000 loss=2.9054 loss_recon=2.9054 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5023 init_acc_corrupt=0.4625 init_gold_top10=0.4920 init_gold_top100=0.4980
+step=39000 micro_steps=78000 elapsed=33.6s lr=3.000000e-04 loss_all=2.8787 acc_all=0.5845 loss_corrupt=2.8787 acc_corrupt=0.5845 corrupt_frac=1.0000 loss=2.8787 loss_recon=2.8787 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4975 init_acc_corrupt=0.4678 init_gold_top10=0.4968 init_gold_top100=0.5027
+step=39100 micro_steps=78200 elapsed=35.3s lr=3.000000e-04 loss_all=2.9023 acc_all=0.5812 loss_corrupt=2.9023 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9023 loss_recon=2.9023 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5011 init_acc_corrupt=0.4639 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=39200 micro_steps=78400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8367 acc_all=0.5899 loss_corrupt=2.8367 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8367 loss_recon=2.8367 loss_meanflow=0.0000 mean_model_t=0.5070 mean_corrupt_t=0.5070 wrong_frac=0.4938 init_acc_corrupt=0.4725 init_gold_top10=0.5008 init_gold_top100=0.5064
+step=39300 micro_steps=78600 elapsed=33.9s lr=3.000000e-04 loss_all=2.9062 acc_all=0.5809 loss_corrupt=2.9062 acc_corrupt=0.5809 corrupt_frac=1.0000 loss=2.9062 loss_recon=2.9062 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5020 init_acc_corrupt=0.4632 init_gold_top10=0.4924 init_gold_top100=0.4981
+step=39400 micro_steps=78800 elapsed=33.9s lr=3.000000e-04 loss_all=2.8867 acc_all=0.5831 loss_corrupt=2.8867 acc_corrupt=0.5831 corrupt_frac=1.0000 loss=2.8867 loss_recon=2.8867 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4999 init_acc_corrupt=0.4655 init_gold_top10=0.4945 init_gold_top100=0.5003
+step=39500 micro_steps=79000 elapsed=33.5s lr=3.000000e-04 loss_all=2.8709 acc_all=0.5851 loss_corrupt=2.8709 acc_corrupt=0.5851 corrupt_frac=1.0000 loss=2.8709 loss_recon=2.8709 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4981 init_acc_corrupt=0.4672 init_gold_top10=0.4962 init_gold_top100=0.5021
+step=39600 micro_steps=79200 elapsed=33.6s lr=3.000000e-04 loss_all=2.8781 acc_all=0.5843 loss_corrupt=2.8781 acc_corrupt=0.5843 corrupt_frac=1.0000 loss=2.8781 loss_recon=2.8781 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4983 init_acc_corrupt=0.4670 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=39700 micro_steps=79400 elapsed=33.5s lr=3.000000e-04 loss_all=2.9054 acc_all=0.5806 loss_corrupt=2.9054 acc_corrupt=0.5806 corrupt_frac=1.0000 loss=2.9054 loss_recon=2.9054 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5025 init_acc_corrupt=0.4625 init_gold_top10=0.4919 init_gold_top100=0.4976
+step=39800 micro_steps=79600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8675 acc_all=0.5857 loss_corrupt=2.8675 acc_corrupt=0.5857 corrupt_frac=1.0000 loss=2.8675 loss_recon=2.8675 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4981 init_acc_corrupt=0.4674 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=39900 micro_steps=79800 elapsed=33.8s lr=3.000000e-04 loss_all=2.9182 acc_all=0.5786 loss_corrupt=2.9182 acc_corrupt=0.5786 corrupt_frac=1.0000 loss=2.9182 loss_recon=2.9182 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5032 init_acc_corrupt=0.4615 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=40000 micro_steps=80000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8457 acc_all=0.5885 loss_corrupt=2.8457 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8457 loss_recon=2.8457 loss_meanflow=0.0000 mean_model_t=0.5050 mean_corrupt_t=0.5050 wrong_frac=0.4953 init_acc_corrupt=0.4709 init_gold_top10=0.4992 init_gold_top100=0.5049
+step=40100 micro_steps=80200 elapsed=36.8s lr=3.000000e-04 loss_all=2.8562 acc_all=0.5871 loss_corrupt=2.8562 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8562 loss_recon=2.8562 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4957 init_acc_corrupt=0.4707 init_gold_top10=0.4988 init_gold_top100=0.5044
+step=40200 micro_steps=80400 elapsed=34.0s lr=3.000000e-04 loss_all=2.9034 acc_all=0.5806 loss_corrupt=2.9034 acc_corrupt=0.5806 corrupt_frac=1.0000 loss=2.9034 loss_recon=2.9034 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5027 init_acc_corrupt=0.4623 init_gold_top10=0.4917 init_gold_top100=0.4974
+step=40300 micro_steps=80600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8873 acc_all=0.5833 loss_corrupt=2.8873 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8873 loss_recon=2.8873 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5004 init_acc_corrupt=0.4650 init_gold_top10=0.4942 init_gold_top100=0.4998
+step=40400 micro_steps=80800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8766 acc_all=0.5846 loss_corrupt=2.8766 acc_corrupt=0.5846 corrupt_frac=1.0000 loss=2.8766 loss_recon=2.8766 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4986 init_acc_corrupt=0.4674 init_gold_top10=0.4959 init_gold_top100=0.5015
+step=40500 micro_steps=81000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8647 acc_all=0.5858 loss_corrupt=2.8647 acc_corrupt=0.5858 corrupt_frac=1.0000 loss=2.8647 loss_recon=2.8647 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4981 init_acc_corrupt=0.4676 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=40600 micro_steps=81200 elapsed=33.9s lr=3.000000e-04 loss_all=2.9115 acc_all=0.5795 loss_corrupt=2.9115 acc_corrupt=0.5795 corrupt_frac=1.0000 loss=2.9115 loss_recon=2.9115 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5033 init_acc_corrupt=0.4616 init_gold_top10=0.4909 init_gold_top100=0.4968
+step=40700 micro_steps=81400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8691 acc_all=0.5856 loss_corrupt=2.8691 acc_corrupt=0.5856 corrupt_frac=1.0000 loss=2.8691 loss_recon=2.8691 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4975 init_acc_corrupt=0.4678 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=40800 micro_steps=81600 elapsed=34.1s lr=3.000000e-04 loss_all=2.8323 acc_all=0.5906 loss_corrupt=2.8323 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8323 loss_recon=2.8323 loss_meanflow=0.0000 mean_model_t=0.5069 mean_corrupt_t=0.5069 wrong_frac=0.4932 init_acc_corrupt=0.4727 init_gold_top10=0.5014 init_gold_top100=0.5071
+step=40900 micro_steps=81800 elapsed=33.6s lr=3.000000e-04 loss_all=2.9153 acc_all=0.5790 loss_corrupt=2.9153 acc_corrupt=0.5790 corrupt_frac=1.0000 loss=2.9153 loss_recon=2.9153 loss_meanflow=0.0000 mean_model_t=0.4954 mean_corrupt_t=0.4954 wrong_frac=0.5044 init_acc_corrupt=0.4602 init_gold_top10=0.4899 init_gold_top100=0.4957
+step=41000 micro_steps=82000 elapsed=33.6s lr=3.000000e-04 loss_all=2.8564 acc_all=0.5871 loss_corrupt=2.8564 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8564 loss_recon=2.8564 loss_meanflow=0.0000 mean_model_t=0.5039 mean_corrupt_t=0.5039 wrong_frac=0.4960 init_acc_corrupt=0.4694 init_gold_top10=0.4984 init_gold_top100=0.5041
+step=41100 micro_steps=82200 elapsed=35.1s lr=3.000000e-04 loss_all=2.8823 acc_all=0.5836 loss_corrupt=2.8823 acc_corrupt=0.5836 corrupt_frac=1.0000 loss=2.8823 loss_recon=2.8823 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5004 init_acc_corrupt=0.4656 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=41200 micro_steps=82400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8626 acc_all=0.5865 loss_corrupt=2.8626 acc_corrupt=0.5865 corrupt_frac=1.0000 loss=2.8626 loss_recon=2.8626 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4971 init_acc_corrupt=0.4686 init_gold_top10=0.4973 init_gold_top100=0.5030
+step=41300 micro_steps=82600 elapsed=33.8s lr=3.000000e-04 loss_all=2.9150 acc_all=0.5793 loss_corrupt=2.9150 acc_corrupt=0.5793 corrupt_frac=1.0000 loss=2.9150 loss_recon=2.9150 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5036 init_acc_corrupt=0.4616 init_gold_top10=0.4906 init_gold_top100=0.4964
+step=41400 micro_steps=82800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8522 acc_all=0.5878 loss_corrupt=2.8522 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8522 loss_recon=2.8522 loss_meanflow=0.0000 mean_model_t=0.5042 mean_corrupt_t=0.5042 wrong_frac=0.4958 init_acc_corrupt=0.4702 init_gold_top10=0.4987 init_gold_top100=0.5044
+step=41500 micro_steps=83000 elapsed=33.6s lr=3.000000e-04 loss_all=2.8645 acc_all=0.5855 loss_corrupt=2.8645 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8645 loss_recon=2.8645 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4985 init_acc_corrupt=0.4668 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=41600 micro_steps=83200 elapsed=33.8s lr=3.000000e-04 loss_all=2.9224 acc_all=0.5788 loss_corrupt=2.9224 acc_corrupt=0.5788 corrupt_frac=1.0000 loss=2.9224 loss_recon=2.9224 loss_meanflow=0.0000 mean_model_t=0.4957 mean_corrupt_t=0.4957 wrong_frac=0.5042 init_acc_corrupt=0.4603 init_gold_top10=0.4900 init_gold_top100=0.4959
+step=41700 micro_steps=83400 elapsed=36.0s lr=3.000000e-04 loss_all=2.9114 acc_all=0.5801 loss_corrupt=2.9114 acc_corrupt=0.5801 corrupt_frac=1.0000 loss=2.9114 loss_recon=2.9114 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5025 init_acc_corrupt=0.4621 init_gold_top10=0.4918 init_gold_top100=0.4977
+step=41800 micro_steps=83600 elapsed=34.0s lr=3.000000e-04 loss_all=2.8769 acc_all=0.5846 loss_corrupt=2.8769 acc_corrupt=0.5846 corrupt_frac=1.0000 loss=2.8769 loss_recon=2.8769 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4987 init_acc_corrupt=0.4666 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=41900 micro_steps=83800 elapsed=34.0s lr=3.000000e-04 loss_all=2.9075 acc_all=0.5806 loss_corrupt=2.9075 acc_corrupt=0.5806 corrupt_frac=1.0000 loss=2.9075 loss_recon=2.9075 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5026 init_acc_corrupt=0.4627 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=42000 micro_steps=84000 elapsed=33.8s lr=3.000000e-04 loss_all=2.9189 acc_all=0.5790 loss_corrupt=2.9189 acc_corrupt=0.5790 corrupt_frac=1.0000 loss=2.9189 loss_recon=2.9189 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5038 init_acc_corrupt=0.4613 init_gold_top10=0.4903 init_gold_top100=0.4963
+step=42100 micro_steps=84200 elapsed=35.6s lr=3.000000e-04 loss_all=2.8863 acc_all=0.5833 loss_corrupt=2.8863 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8863 loss_recon=2.8863 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5002 init_acc_corrupt=0.4655 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=42200 micro_steps=84400 elapsed=33.6s lr=3.000000e-04 loss_all=2.8943 acc_all=0.5825 loss_corrupt=2.8943 acc_corrupt=0.5825 corrupt_frac=1.0000 loss=2.8943 loss_recon=2.8943 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5003 init_acc_corrupt=0.4651 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=42300 micro_steps=84600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8868 acc_all=0.5830 loss_corrupt=2.8868 acc_corrupt=0.5830 corrupt_frac=1.0000 loss=2.8868 loss_recon=2.8868 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5001 init_acc_corrupt=0.4652 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=42400 micro_steps=84800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8970 acc_all=0.5816 loss_corrupt=2.8970 acc_corrupt=0.5816 corrupt_frac=1.0000 loss=2.8970 loss_recon=2.8970 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5025 init_acc_corrupt=0.4625 init_gold_top10=0.4918 init_gold_top100=0.4977
+step=42500 micro_steps=85000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8653 acc_all=0.5856 loss_corrupt=2.8653 acc_corrupt=0.5856 corrupt_frac=1.0000 loss=2.8653 loss_recon=2.8653 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4979 init_acc_corrupt=0.4678 init_gold_top10=0.4964 init_gold_top100=0.5022
+step=42600 micro_steps=85200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8709 acc_all=0.5851 loss_corrupt=2.8709 acc_corrupt=0.5851 corrupt_frac=1.0000 loss=2.8709 loss_recon=2.8709 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4995 init_acc_corrupt=0.4663 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=42700 micro_steps=85400 elapsed=33.8s lr=3.000000e-04 loss_all=2.9032 acc_all=0.5810 loss_corrupt=2.9032 acc_corrupt=0.5810 corrupt_frac=1.0000 loss=2.9032 loss_recon=2.9032 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5014 init_acc_corrupt=0.4634 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=42800 micro_steps=85600 elapsed=34.0s lr=3.000000e-04 loss_all=2.8689 acc_all=0.5849 loss_corrupt=2.8689 acc_corrupt=0.5849 corrupt_frac=1.0000 loss=2.8689 loss_recon=2.8689 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4991 init_acc_corrupt=0.4671 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=42900 micro_steps=85800 elapsed=34.0s lr=3.000000e-04 loss_all=2.8400 acc_all=0.5888 loss_corrupt=2.8400 acc_corrupt=0.5888 corrupt_frac=1.0000 loss=2.8400 loss_recon=2.8400 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4954 init_acc_corrupt=0.4707 init_gold_top10=0.4993 init_gold_top100=0.5047
+step=43000 micro_steps=86000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8418 acc_all=0.5892 loss_corrupt=2.8418 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8418 loss_recon=2.8418 loss_meanflow=0.0000 mean_model_t=0.5048 mean_corrupt_t=0.5048 wrong_frac=0.4953 init_acc_corrupt=0.4707 init_gold_top10=0.4993 init_gold_top100=0.5049
+step=43100 micro_steps=86200 elapsed=35.0s lr=3.000000e-04 loss_all=2.8987 acc_all=0.5812 loss_corrupt=2.8987 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.8987 loss_recon=2.8987 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5031 init_acc_corrupt=0.4629 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=43200 micro_steps=86400 elapsed=33.5s lr=3.000000e-04 loss_all=2.8970 acc_all=0.5815 loss_corrupt=2.8970 acc_corrupt=0.5815 corrupt_frac=1.0000 loss=2.8970 loss_recon=2.8970 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5021 init_acc_corrupt=0.4636 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=43300 micro_steps=86600 elapsed=33.5s lr=3.000000e-04 loss_all=2.8785 acc_all=0.5840 loss_corrupt=2.8785 acc_corrupt=0.5840 corrupt_frac=1.0000 loss=2.8785 loss_recon=2.8785 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4994 init_acc_corrupt=0.4658 init_gold_top10=0.4948 init_gold_top100=0.5008
+step=43400 micro_steps=86800 elapsed=33.6s lr=3.000000e-04 loss_all=2.8631 acc_all=0.5864 loss_corrupt=2.8631 acc_corrupt=0.5864 corrupt_frac=1.0000 loss=2.8631 loss_recon=2.8631 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4975 init_acc_corrupt=0.4679 init_gold_top10=0.4969 init_gold_top100=0.5027
+step=43500 micro_steps=87000 elapsed=33.6s lr=3.000000e-04 loss_all=2.8740 acc_all=0.5844 loss_corrupt=2.8740 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8740 loss_recon=2.8740 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4997 init_acc_corrupt=0.4659 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=43600 micro_steps=87200 elapsed=33.8s lr=3.000000e-04 loss_all=2.8816 acc_all=0.5833 loss_corrupt=2.8816 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8816 loss_recon=2.8816 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5006 init_acc_corrupt=0.4647 init_gold_top10=0.4939 init_gold_top100=0.4994
+step=43700 micro_steps=87400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8613 acc_all=0.5862 loss_corrupt=2.8613 acc_corrupt=0.5862 corrupt_frac=1.0000 loss=2.8613 loss_recon=2.8613 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4975 init_acc_corrupt=0.4685 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=43800 micro_steps=87600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8731 acc_all=0.5849 loss_corrupt=2.8731 acc_corrupt=0.5849 corrupt_frac=1.0000 loss=2.8731 loss_recon=2.8731 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4990 init_acc_corrupt=0.4665 init_gold_top10=0.4954 init_gold_top100=0.5012
+step=43900 micro_steps=87800 elapsed=33.9s lr=3.000000e-04 loss_all=2.8817 acc_all=0.5839 loss_corrupt=2.8817 acc_corrupt=0.5839 corrupt_frac=1.0000 loss=2.8817 loss_recon=2.8817 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5000 init_acc_corrupt=0.4651 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=44000 micro_steps=88000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8419 acc_all=0.5895 loss_corrupt=2.8419 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8419 loss_recon=2.8419 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4957 init_acc_corrupt=0.4707 init_gold_top10=0.4989 init_gold_top100=0.5045
+step=44100 micro_steps=88200 elapsed=35.1s lr=3.000000e-04 loss_all=2.8642 acc_all=0.5859 loss_corrupt=2.8642 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8642 loss_recon=2.8642 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4980 init_acc_corrupt=0.4679 init_gold_top10=0.4963 init_gold_top100=0.5021
+step=44200 micro_steps=88400 elapsed=33.5s lr=3.000000e-04 loss_all=2.8778 acc_all=0.5843 loss_corrupt=2.8778 acc_corrupt=0.5843 corrupt_frac=1.0000 loss=2.8778 loss_recon=2.8778 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5006 init_acc_corrupt=0.4650 init_gold_top10=0.4939 init_gold_top100=0.4995
+step=44300 micro_steps=88600 elapsed=34.0s lr=3.000000e-04 loss_all=2.9026 acc_all=0.5810 loss_corrupt=2.9026 acc_corrupt=0.5810 corrupt_frac=1.0000 loss=2.9026 loss_recon=2.9026 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5028 init_acc_corrupt=0.4624 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=44400 micro_steps=88800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8862 acc_all=0.5829 loss_corrupt=2.8862 acc_corrupt=0.5829 corrupt_frac=1.0000 loss=2.8862 loss_recon=2.8862 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5009 init_acc_corrupt=0.4644 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=44500 micro_steps=89000 elapsed=33.6s lr=3.000000e-04 loss_all=2.9184 acc_all=0.5792 loss_corrupt=2.9184 acc_corrupt=0.5792 corrupt_frac=1.0000 loss=2.9184 loss_recon=2.9184 loss_meanflow=0.0000 mean_model_t=0.4948 mean_corrupt_t=0.4948 wrong_frac=0.5052 init_acc_corrupt=0.4598 init_gold_top10=0.4890 init_gold_top100=0.4949
+step=44600 micro_steps=89200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8790 acc_all=0.5841 loss_corrupt=2.8790 acc_corrupt=0.5841 corrupt_frac=1.0000 loss=2.8790 loss_recon=2.8790 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4998 init_acc_corrupt=0.4655 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=44700 micro_steps=89400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8741 acc_all=0.5845 loss_corrupt=2.8741 acc_corrupt=0.5845 corrupt_frac=1.0000 loss=2.8741 loss_recon=2.8741 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4992 init_acc_corrupt=0.4663 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=44800 micro_steps=89600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8675 acc_all=0.5853 loss_corrupt=2.8675 acc_corrupt=0.5853 corrupt_frac=1.0000 loss=2.8675 loss_recon=2.8675 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4989 init_acc_corrupt=0.4669 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=44900 micro_steps=89800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8912 acc_all=0.5824 loss_corrupt=2.8912 acc_corrupt=0.5824 corrupt_frac=1.0000 loss=2.8912 loss_recon=2.8912 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5012 init_acc_corrupt=0.4639 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=45000 micro_steps=90000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8493 acc_all=0.5879 loss_corrupt=2.8493 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8493 loss_recon=2.8493 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4972 init_acc_corrupt=0.4687 init_gold_top10=0.4973 init_gold_top100=0.5029
+step=45100 micro_steps=90200 elapsed=35.3s lr=3.000000e-04 loss_all=2.8707 acc_all=0.5853 loss_corrupt=2.8707 acc_corrupt=0.5853 corrupt_frac=1.0000 loss=2.8707 loss_recon=2.8707 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4992 init_acc_corrupt=0.4662 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=45200 micro_steps=90400 elapsed=33.6s lr=3.000000e-04 loss_all=2.8659 acc_all=0.5858 loss_corrupt=2.8659 acc_corrupt=0.5858 corrupt_frac=1.0000 loss=2.8659 loss_recon=2.8659 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4982 init_acc_corrupt=0.4673 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=45300 micro_steps=90600 elapsed=33.6s lr=3.000000e-04 loss_all=2.9113 acc_all=0.5798 loss_corrupt=2.9113 acc_corrupt=0.5798 corrupt_frac=1.0000 loss=2.9113 loss_recon=2.9113 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5036 init_acc_corrupt=0.4605 init_gold_top10=0.4907 init_gold_top100=0.4965
+step=45400 micro_steps=90800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8918 acc_all=0.5824 loss_corrupt=2.8918 acc_corrupt=0.5824 corrupt_frac=1.0000 loss=2.8918 loss_recon=2.8918 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5019 init_acc_corrupt=0.4633 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=45500 micro_steps=91000 elapsed=33.7s lr=3.000000e-04 loss_all=2.9001 acc_all=0.5808 loss_corrupt=2.9001 acc_corrupt=0.5808 corrupt_frac=1.0000 loss=2.9001 loss_recon=2.9001 loss_meanflow=0.0000 mean_model_t=0.4960 mean_corrupt_t=0.4960 wrong_frac=0.5037 init_acc_corrupt=0.4615 init_gold_top10=0.4905 init_gold_top100=0.4964
+step=45600 micro_steps=91200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8748 acc_all=0.5844 loss_corrupt=2.8748 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8748 loss_recon=2.8748 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5001 init_acc_corrupt=0.4660 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=45700 micro_steps=91400 elapsed=33.5s lr=3.000000e-04 loss_all=2.8772 acc_all=0.5850 loss_corrupt=2.8772 acc_corrupt=0.5850 corrupt_frac=1.0000 loss=2.8772 loss_recon=2.8772 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4996 init_acc_corrupt=0.4657 init_gold_top10=0.4949 init_gold_top100=0.5006
+step=45800 micro_steps=91600 elapsed=33.4s lr=3.000000e-04 loss_all=2.8524 acc_all=0.5874 loss_corrupt=2.8524 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8524 loss_recon=2.8524 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4969 init_acc_corrupt=0.4686 init_gold_top10=0.4976 init_gold_top100=0.5033
+step=45900 micro_steps=91800 elapsed=33.5s lr=3.000000e-04 loss_all=2.8855 acc_all=0.5828 loss_corrupt=2.8855 acc_corrupt=0.5828 corrupt_frac=1.0000 loss=2.8855 loss_recon=2.8855 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5011 init_acc_corrupt=0.4640 init_gold_top10=0.4932 init_gold_top100=0.4991
+step=46000 micro_steps=92000 elapsed=33.5s lr=3.000000e-04 loss_all=2.8671 acc_all=0.5852 loss_corrupt=2.8671 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8671 loss_recon=2.8671 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4995 init_acc_corrupt=0.4659 init_gold_top10=0.4949 init_gold_top100=0.5007
+step=46100 micro_steps=92200 elapsed=34.9s lr=3.000000e-04 loss_all=2.9005 acc_all=0.5809 loss_corrupt=2.9005 acc_corrupt=0.5809 corrupt_frac=1.0000 loss=2.9005 loss_recon=2.9005 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5036 init_acc_corrupt=0.4615 init_gold_top10=0.4908 init_gold_top100=0.4966
+step=46200 micro_steps=92400 elapsed=33.5s lr=3.000000e-04 loss_all=2.8461 acc_all=0.5883 loss_corrupt=2.8461 acc_corrupt=0.5883 corrupt_frac=1.0000 loss=2.8461 loss_recon=2.8461 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4984 init_acc_corrupt=0.4675 init_gold_top10=0.4963 init_gold_top100=0.5019
+step=46300 micro_steps=92600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8599 acc_all=0.5866 loss_corrupt=2.8599 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8599 loss_recon=2.8599 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4975 init_acc_corrupt=0.4678 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=46400 micro_steps=92800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8668 acc_all=0.5855 loss_corrupt=2.8668 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8668 loss_recon=2.8668 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4983 init_acc_corrupt=0.4672 init_gold_top10=0.4960 init_gold_top100=0.5018
+step=46500 micro_steps=93000 elapsed=33.4s lr=3.000000e-04 loss_all=2.8329 acc_all=0.5898 loss_corrupt=2.8329 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8329 loss_recon=2.8329 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4962 init_acc_corrupt=0.4703 init_gold_top10=0.4984 init_gold_top100=0.5040
+step=46600 micro_steps=93200 elapsed=33.4s lr=3.000000e-04 loss_all=2.8662 acc_all=0.5855 loss_corrupt=2.8662 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8662 loss_recon=2.8662 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4986 init_acc_corrupt=0.4669 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=46700 micro_steps=93400 elapsed=33.5s lr=3.000000e-04 loss_all=2.8882 acc_all=0.5827 loss_corrupt=2.8882 acc_corrupt=0.5827 corrupt_frac=1.0000 loss=2.8882 loss_recon=2.8882 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5017 init_acc_corrupt=0.4637 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=46800 micro_steps=93600 elapsed=33.5s lr=3.000000e-04 loss_all=2.8845 acc_all=0.5832 loss_corrupt=2.8845 acc_corrupt=0.5832 corrupt_frac=1.0000 loss=2.8845 loss_recon=2.8845 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5016 init_acc_corrupt=0.4633 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=46900 micro_steps=93800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8797 acc_all=0.5840 loss_corrupt=2.8797 acc_corrupt=0.5840 corrupt_frac=1.0000 loss=2.8797 loss_recon=2.8797 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5003 init_acc_corrupt=0.4647 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=47000 micro_steps=94000 elapsed=33.5s lr=3.000000e-04 loss_all=2.8905 acc_all=0.5826 loss_corrupt=2.8905 acc_corrupt=0.5826 corrupt_frac=1.0000 loss=2.8905 loss_recon=2.8905 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5014 init_acc_corrupt=0.4639 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=47100 micro_steps=94200 elapsed=35.0s lr=3.000000e-04 loss_all=2.9005 acc_all=0.5812 loss_corrupt=2.9005 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9005 loss_recon=2.9005 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5028 init_acc_corrupt=0.4622 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=47200 micro_steps=94400 elapsed=33.6s lr=3.000000e-04 loss_all=2.8952 acc_all=0.5818 loss_corrupt=2.8952 acc_corrupt=0.5818 corrupt_frac=1.0000 loss=2.8952 loss_recon=2.8952 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5034 init_acc_corrupt=0.4618 init_gold_top10=0.4909 init_gold_top100=0.4966
+step=47300 micro_steps=94600 elapsed=33.7s lr=3.000000e-04 loss_all=2.9213 acc_all=0.5780 loss_corrupt=2.9213 acc_corrupt=0.5780 corrupt_frac=1.0000 loss=2.9213 loss_recon=2.9213 loss_meanflow=0.0000 mean_model_t=0.4942 mean_corrupt_t=0.4942 wrong_frac=0.5057 init_acc_corrupt=0.4588 init_gold_top10=0.4884 init_gold_top100=0.4943
+step=47400 micro_steps=94800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8907 acc_all=0.5824 loss_corrupt=2.8907 acc_corrupt=0.5824 corrupt_frac=1.0000 loss=2.8907 loss_recon=2.8907 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5022 init_acc_corrupt=0.4630 init_gold_top10=0.4921 init_gold_top100=0.4979
+step=47500 micro_steps=95000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8835 acc_all=0.5833 loss_corrupt=2.8835 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8835 loss_recon=2.8835 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5010 init_acc_corrupt=0.4642 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=47600 micro_steps=95200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8912 acc_all=0.5826 loss_corrupt=2.8912 acc_corrupt=0.5826 corrupt_frac=1.0000 loss=2.8912 loss_recon=2.8912 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5026 init_acc_corrupt=0.4629 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=47700 micro_steps=95400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8394 acc_all=0.5886 loss_corrupt=2.8394 acc_corrupt=0.5886 corrupt_frac=1.0000 loss=2.8394 loss_recon=2.8394 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4978 init_acc_corrupt=0.4687 init_gold_top10=0.4968 init_gold_top100=0.5023
+step=47800 micro_steps=95600 elapsed=33.6s lr=3.000000e-04 loss_all=2.8904 acc_all=0.5826 loss_corrupt=2.8904 acc_corrupt=0.5826 corrupt_frac=1.0000 loss=2.8904 loss_recon=2.8904 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5009 init_acc_corrupt=0.4641 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=47900 micro_steps=95800 elapsed=33.6s lr=3.000000e-04 loss_all=2.9508 acc_all=0.5745 loss_corrupt=2.9508 acc_corrupt=0.5745 corrupt_frac=1.0000 loss=2.9508 loss_recon=2.9508 loss_meanflow=0.0000 mean_model_t=0.4904 mean_corrupt_t=0.4904 wrong_frac=0.5099 init_acc_corrupt=0.4549 init_gold_top10=0.4843 init_gold_top100=0.4902
+step=48000 micro_steps=96000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8377 acc_all=0.5894 loss_corrupt=2.8377 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8377 loss_recon=2.8377 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4963 init_acc_corrupt=0.4701 init_gold_top10=0.4982 init_gold_top100=0.5038
+step=48100 micro_steps=96200 elapsed=35.2s lr=3.000000e-04 loss_all=2.8875 acc_all=0.5828 loss_corrupt=2.8875 acc_corrupt=0.5828 corrupt_frac=1.0000 loss=2.8875 loss_recon=2.8875 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5025 init_acc_corrupt=0.4624 init_gold_top10=0.4918 init_gold_top100=0.4976
+step=48200 micro_steps=96400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8809 acc_all=0.5832 loss_corrupt=2.8809 acc_corrupt=0.5832 corrupt_frac=1.0000 loss=2.8809 loss_recon=2.8809 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4631 init_gold_top10=0.4927 init_gold_top100=0.4987
+step=48300 micro_steps=96600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8669 acc_all=0.5857 loss_corrupt=2.8669 acc_corrupt=0.5857 corrupt_frac=1.0000 loss=2.8669 loss_recon=2.8669 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5001 init_acc_corrupt=0.4657 init_gold_top10=0.4943 init_gold_top100=0.5000
+step=48400 micro_steps=96800 elapsed=33.9s lr=3.000000e-04 loss_all=2.9018 acc_all=0.5812 loss_corrupt=2.9018 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.9018 loss_recon=2.9018 loss_meanflow=0.0000 mean_model_t=0.4961 mean_corrupt_t=0.4961 wrong_frac=0.5035 init_acc_corrupt=0.4613 init_gold_top10=0.4909 init_gold_top100=0.4967
+step=48500 micro_steps=97000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8656 acc_all=0.5861 loss_corrupt=2.8656 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8656 loss_recon=2.8656 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4986 init_acc_corrupt=0.4671 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=48600 micro_steps=97200 elapsed=33.8s lr=3.000000e-04 loss_all=2.8656 acc_all=0.5855 loss_corrupt=2.8656 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8656 loss_recon=2.8656 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4996 init_acc_corrupt=0.4657 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=48700 micro_steps=97400 elapsed=34.0s lr=3.000000e-04 loss_all=2.8770 acc_all=0.5844 loss_corrupt=2.8770 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8770 loss_recon=2.8770 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4994 init_acc_corrupt=0.4663 init_gold_top10=0.4948 init_gold_top100=0.5007
+step=48800 micro_steps=97600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8922 acc_all=0.5819 loss_corrupt=2.8922 acc_corrupt=0.5819 corrupt_frac=1.0000 loss=2.8922 loss_recon=2.8922 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5034 init_acc_corrupt=0.4616 init_gold_top10=0.4910 init_gold_top100=0.4967
+step=48900 micro_steps=97800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8822 acc_all=0.5839 loss_corrupt=2.8822 acc_corrupt=0.5839 corrupt_frac=1.0000 loss=2.8822 loss_recon=2.8822 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4927 init_gold_top100=0.4986
+step=49000 micro_steps=98000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8701 acc_all=0.5856 loss_corrupt=2.8701 acc_corrupt=0.5856 corrupt_frac=1.0000 loss=2.8701 loss_recon=2.8701 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4988 init_acc_corrupt=0.4666 init_gold_top10=0.4955 init_gold_top100=0.5013
+step=49100 micro_steps=98200 elapsed=35.3s lr=3.000000e-04 loss_all=2.8678 acc_all=0.5852 loss_corrupt=2.8678 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8678 loss_recon=2.8678 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4999 init_acc_corrupt=0.4661 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=49200 micro_steps=98400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8828 acc_all=0.5833 loss_corrupt=2.8828 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8828 loss_recon=2.8828 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5019 init_acc_corrupt=0.4636 init_gold_top10=0.4925 init_gold_top100=0.4983
+step=49300 micro_steps=98600 elapsed=34.3s lr=3.000000e-04 loss_all=2.8833 acc_all=0.5833 loss_corrupt=2.8833 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8833 loss_recon=2.8833 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5018 init_acc_corrupt=0.4641 init_gold_top10=0.4925 init_gold_top100=0.4983
+step=49400 micro_steps=98800 elapsed=34.0s lr=3.000000e-04 loss_all=2.8515 acc_all=0.5876 loss_corrupt=2.8515 acc_corrupt=0.5876 corrupt_frac=1.0000 loss=2.8515 loss_recon=2.8515 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4978 init_acc_corrupt=0.4682 init_gold_top10=0.4966 init_gold_top100=0.5023
+step=49500 micro_steps=99000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8940 acc_all=0.5819 loss_corrupt=2.8940 acc_corrupt=0.5819 corrupt_frac=1.0000 loss=2.8940 loss_recon=2.8940 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5027 init_acc_corrupt=0.4624 init_gold_top10=0.4915 init_gold_top100=0.4974
+step=49600 micro_steps=99200 elapsed=34.0s lr=3.000000e-04 loss_all=2.8545 acc_all=0.5870 loss_corrupt=2.8545 acc_corrupt=0.5870 corrupt_frac=1.0000 loss=2.8545 loss_recon=2.8545 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4981 init_acc_corrupt=0.4676 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=49700 micro_steps=99400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8687 acc_all=0.5850 loss_corrupt=2.8687 acc_corrupt=0.5850 corrupt_frac=1.0000 loss=2.8687 loss_recon=2.8687 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5004 init_acc_corrupt=0.4653 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=49800 micro_steps=99600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8695 acc_all=0.5852 loss_corrupt=2.8695 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8695 loss_recon=2.8695 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4988 init_acc_corrupt=0.4662 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=49900 micro_steps=99800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8736 acc_all=0.5844 loss_corrupt=2.8736 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8736 loss_recon=2.8736 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5009 init_acc_corrupt=0.4641 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=50000 micro_steps=100000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8833 acc_all=0.5830 loss_corrupt=2.8833 acc_corrupt=0.5830 corrupt_frac=1.0000 loss=2.8833 loss_recon=2.8833 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5013 init_acc_corrupt=0.4640 init_gold_top10=0.4930 init_gold_top100=0.4989
+step=50100 micro_steps=100200 elapsed=35.7s lr=3.000000e-04 loss_all=2.8563 acc_all=0.5870 loss_corrupt=2.8563 acc_corrupt=0.5870 corrupt_frac=1.0000 loss=2.8563 loss_recon=2.8563 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4979 init_acc_corrupt=0.4680 init_gold_top10=0.4965 init_gold_top100=0.5021
+step=50200 micro_steps=100400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8790 acc_all=0.5837 loss_corrupt=2.8790 acc_corrupt=0.5837 corrupt_frac=1.0000 loss=2.8790 loss_recon=2.8790 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5007 init_acc_corrupt=0.4640 init_gold_top10=0.4935 init_gold_top100=0.4995
+step=50300 micro_steps=100600 elapsed=34.2s lr=3.000000e-04 loss_all=2.9184 acc_all=0.5782 loss_corrupt=2.9184 acc_corrupt=0.5782 corrupt_frac=1.0000 loss=2.9184 loss_recon=2.9184 loss_meanflow=0.0000 mean_model_t=0.4927 mean_corrupt_t=0.4927 wrong_frac=0.5072 init_acc_corrupt=0.4578 init_gold_top10=0.4870 init_gold_top100=0.4929
+step=50400 micro_steps=100800 elapsed=34.1s lr=3.000000e-04 loss_all=2.8668 acc_all=0.5860 loss_corrupt=2.8668 acc_corrupt=0.5860 corrupt_frac=1.0000 loss=2.8668 loss_recon=2.8668 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4995 init_acc_corrupt=0.4661 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=50500 micro_steps=101000 elapsed=33.6s lr=3.000000e-04 loss_all=2.8452 acc_all=0.5875 loss_corrupt=2.8452 acc_corrupt=0.5875 corrupt_frac=1.0000 loss=2.8452 loss_recon=2.8452 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4985 init_acc_corrupt=0.4675 init_gold_top10=0.4961 init_gold_top100=0.5017
+step=50600 micro_steps=101200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8561 acc_all=0.5867 loss_corrupt=2.8561 acc_corrupt=0.5867 corrupt_frac=1.0000 loss=2.8561 loss_recon=2.8561 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4993 init_acc_corrupt=0.4665 init_gold_top10=0.4952 init_gold_top100=0.5008
+step=50700 micro_steps=101400 elapsed=34.0s lr=3.000000e-04 loss_all=2.8651 acc_all=0.5856 loss_corrupt=2.8651 acc_corrupt=0.5856 corrupt_frac=1.0000 loss=2.8651 loss_recon=2.8651 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4995 init_acc_corrupt=0.4658 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=50800 micro_steps=101600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8557 acc_all=0.5868 loss_corrupt=2.8557 acc_corrupt=0.5868 corrupt_frac=1.0000 loss=2.8557 loss_recon=2.8557 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4985 init_acc_corrupt=0.4667 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=50900 micro_steps=101800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8663 acc_all=0.5858 loss_corrupt=2.8663 acc_corrupt=0.5858 corrupt_frac=1.0000 loss=2.8663 loss_recon=2.8663 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4994 init_acc_corrupt=0.4660 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=51000 micro_steps=102000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8789 acc_all=0.5843 loss_corrupt=2.8789 acc_corrupt=0.5843 corrupt_frac=1.0000 loss=2.8789 loss_recon=2.8789 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5003 init_acc_corrupt=0.4649 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=51100 micro_steps=102200 elapsed=35.3s lr=3.000000e-04 loss_all=2.8659 acc_all=0.5853 loss_corrupt=2.8659 acc_corrupt=0.5853 corrupt_frac=1.0000 loss=2.8659 loss_recon=2.8659 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5004 init_acc_corrupt=0.4647 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=51200 micro_steps=102400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8383 acc_all=0.5891 loss_corrupt=2.8383 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8383 loss_recon=2.8383 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4969 init_acc_corrupt=0.4692 init_gold_top10=0.4977 init_gold_top100=0.5034
+step=51300 micro_steps=102600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8722 acc_all=0.5847 loss_corrupt=2.8722 acc_corrupt=0.5847 corrupt_frac=1.0000 loss=2.8722 loss_recon=2.8722 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5011 init_acc_corrupt=0.4642 init_gold_top10=0.4934 init_gold_top100=0.4990
+step=51400 micro_steps=102800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8981 acc_all=0.5814 loss_corrupt=2.8981 acc_corrupt=0.5814 corrupt_frac=1.0000 loss=2.8981 loss_recon=2.8981 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5033 init_acc_corrupt=0.4610 init_gold_top10=0.4909 init_gold_top100=0.4968
+step=51500 micro_steps=103000 elapsed=34.0s lr=3.000000e-04 loss_all=2.8894 acc_all=0.5826 loss_corrupt=2.8894 acc_corrupt=0.5826 corrupt_frac=1.0000 loss=2.8894 loss_recon=2.8894 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5018 init_acc_corrupt=0.4636 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=51600 micro_steps=103200 elapsed=33.8s lr=3.000000e-04 loss_all=2.9326 acc_all=0.5771 loss_corrupt=2.9326 acc_corrupt=0.5771 corrupt_frac=1.0000 loss=2.9326 loss_recon=2.9326 loss_meanflow=0.0000 mean_model_t=0.4923 mean_corrupt_t=0.4923 wrong_frac=0.5079 init_acc_corrupt=0.4566 init_gold_top10=0.4862 init_gold_top100=0.4923
+step=51700 micro_steps=103400 elapsed=34.0s lr=3.000000e-04 loss_all=2.8838 acc_all=0.5835 loss_corrupt=2.8838 acc_corrupt=0.5835 corrupt_frac=1.0000 loss=2.8838 loss_recon=2.8838 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5011 init_acc_corrupt=0.4639 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=51800 micro_steps=103600 elapsed=34.1s lr=3.000000e-04 loss_all=2.8746 acc_all=0.5849 loss_corrupt=2.8746 acc_corrupt=0.5849 corrupt_frac=1.0000 loss=2.8746 loss_recon=2.8746 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5002 init_acc_corrupt=0.4653 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=51900 micro_steps=103800 elapsed=34.0s lr=3.000000e-04 loss_all=2.8283 acc_all=0.5906 loss_corrupt=2.8283 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8283 loss_recon=2.8283 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5045 wrong_frac=0.4954 init_acc_corrupt=0.4707 init_gold_top10=0.4992 init_gold_top100=0.5048
+step=52000 micro_steps=104000 elapsed=34.0s lr=3.000000e-04 loss_all=2.8966 acc_all=0.5818 loss_corrupt=2.8966 acc_corrupt=0.5818 corrupt_frac=1.0000 loss=2.8966 loss_recon=2.8966 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5035 init_acc_corrupt=0.4617 init_gold_top10=0.4908 init_gold_top100=0.4966
+step=52100 micro_steps=104200 elapsed=35.5s lr=3.000000e-04 loss_all=2.8312 acc_all=0.5899 loss_corrupt=2.8312 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8312 loss_recon=2.8312 loss_meanflow=0.0000 mean_model_t=0.5042 mean_corrupt_t=0.5042 wrong_frac=0.4957 init_acc_corrupt=0.4700 init_gold_top10=0.4988 init_gold_top100=0.5045
+step=52200 micro_steps=104400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8779 acc_all=0.5842 loss_corrupt=2.8779 acc_corrupt=0.5842 corrupt_frac=1.0000 loss=2.8779 loss_recon=2.8779 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5008 init_acc_corrupt=0.4642 init_gold_top10=0.4936 init_gold_top100=0.4994
+step=52300 micro_steps=104600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8245 acc_all=0.5906 loss_corrupt=2.8245 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8245 loss_recon=2.8245 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4962 init_acc_corrupt=0.4698 init_gold_top10=0.4983 init_gold_top100=0.5039
+step=52400 micro_steps=104800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8262 acc_all=0.5907 loss_corrupt=2.8262 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8262 loss_recon=2.8262 loss_meanflow=0.0000 mean_model_t=0.5040 mean_corrupt_t=0.5040 wrong_frac=0.4955 init_acc_corrupt=0.4706 init_gold_top10=0.4990 init_gold_top100=0.5045
+step=52500 micro_steps=105000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8450 acc_all=0.5880 loss_corrupt=2.8450 acc_corrupt=0.5880 corrupt_frac=1.0000 loss=2.8450 loss_recon=2.8450 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4986 init_acc_corrupt=0.4672 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=52600 micro_steps=105200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8700 acc_all=0.5846 loss_corrupt=2.8700 acc_corrupt=0.5846 corrupt_frac=1.0000 loss=2.8700 loss_recon=2.8700 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5012 init_acc_corrupt=0.4645 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=52700 micro_steps=105400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8872 acc_all=0.5833 loss_corrupt=2.8872 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8872 loss_recon=2.8872 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5019 init_acc_corrupt=0.4636 init_gold_top10=0.4923 init_gold_top100=0.4982
+step=52800 micro_steps=105600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8724 acc_all=0.5842 loss_corrupt=2.8724 acc_corrupt=0.5842 corrupt_frac=1.0000 loss=2.8724 loss_recon=2.8724 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5016 init_acc_corrupt=0.4636 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=52900 micro_steps=105800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8560 acc_all=0.5866 loss_corrupt=2.8560 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8560 loss_recon=2.8560 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4986 init_acc_corrupt=0.4670 init_gold_top10=0.4957 init_gold_top100=0.5016
+step=53000 micro_steps=106000 elapsed=33.7s lr=3.000000e-04 loss_all=2.9036 acc_all=0.5805 loss_corrupt=2.9036 acc_corrupt=0.5805 corrupt_frac=1.0000 loss=2.9036 loss_recon=2.9036 loss_meanflow=0.0000 mean_model_t=0.4959 mean_corrupt_t=0.4959 wrong_frac=0.5046 init_acc_corrupt=0.4599 init_gold_top10=0.4899 init_gold_top100=0.4957
+step=53100 micro_steps=106200 elapsed=35.4s lr=3.000000e-04 loss_all=2.8754 acc_all=0.5838 loss_corrupt=2.8754 acc_corrupt=0.5838 corrupt_frac=1.0000 loss=2.8754 loss_recon=2.8754 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5024 init_acc_corrupt=0.4629 init_gold_top10=0.4921 init_gold_top100=0.4977
+step=53200 micro_steps=106400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8740 acc_all=0.5842 loss_corrupt=2.8740 acc_corrupt=0.5842 corrupt_frac=1.0000 loss=2.8740 loss_recon=2.8740 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5015 init_acc_corrupt=0.4635 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=53300 micro_steps=106600 elapsed=34.0s lr=3.000000e-04 loss_all=2.8502 acc_all=0.5874 loss_corrupt=2.8502 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8502 loss_recon=2.8502 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4988 init_acc_corrupt=0.4667 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=53400 micro_steps=106800 elapsed=33.9s lr=3.000000e-04 loss_all=2.8901 acc_all=0.5822 loss_corrupt=2.8901 acc_corrupt=0.5822 corrupt_frac=1.0000 loss=2.8901 loss_recon=2.8901 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5025 init_acc_corrupt=0.4620 init_gold_top10=0.4917 init_gold_top100=0.4976
+step=53500 micro_steps=107000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8731 acc_all=0.5847 loss_corrupt=2.8731 acc_corrupt=0.5847 corrupt_frac=1.0000 loss=2.8731 loss_recon=2.8731 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5015 init_acc_corrupt=0.4638 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=53600 micro_steps=107200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8492 acc_all=0.5876 loss_corrupt=2.8492 acc_corrupt=0.5876 corrupt_frac=1.0000 loss=2.8492 loss_recon=2.8492 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4980 init_acc_corrupt=0.4671 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=53700 micro_steps=107400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8720 acc_all=0.5842 loss_corrupt=2.8720 acc_corrupt=0.5842 corrupt_frac=1.0000 loss=2.8720 loss_recon=2.8720 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5024 init_acc_corrupt=0.4624 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=53800 micro_steps=107600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8394 acc_all=0.5889 loss_corrupt=2.8394 acc_corrupt=0.5889 corrupt_frac=1.0000 loss=2.8394 loss_recon=2.8394 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4972 init_acc_corrupt=0.4689 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=53900 micro_steps=107800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8315 acc_all=0.5904 loss_corrupt=2.8315 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8315 loss_recon=2.8315 loss_meanflow=0.0000 mean_model_t=0.5038 mean_corrupt_t=0.5038 wrong_frac=0.4955 init_acc_corrupt=0.4710 init_gold_top10=0.4989 init_gold_top100=0.5046
+step=54000 micro_steps=108000 elapsed=34.0s lr=3.000000e-04 loss_all=2.8915 acc_all=0.5828 loss_corrupt=2.8915 acc_corrupt=0.5828 corrupt_frac=1.0000 loss=2.8915 loss_recon=2.8915 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5020 init_acc_corrupt=0.4629 init_gold_top10=0.4923 init_gold_top100=0.4982
+step=54100 micro_steps=108200 elapsed=35.3s lr=3.000000e-04 loss_all=2.8688 acc_all=0.5853 loss_corrupt=2.8688 acc_corrupt=0.5853 corrupt_frac=1.0000 loss=2.8688 loss_recon=2.8688 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4996 init_acc_corrupt=0.4654 init_gold_top10=0.4947 init_gold_top100=0.5006
+step=54200 micro_steps=108400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8676 acc_all=0.5857 loss_corrupt=2.8676 acc_corrupt=0.5857 corrupt_frac=1.0000 loss=2.8676 loss_recon=2.8676 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4998 init_acc_corrupt=0.4654 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=54300 micro_steps=108600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8556 acc_all=0.5867 loss_corrupt=2.8556 acc_corrupt=0.5867 corrupt_frac=1.0000 loss=2.8556 loss_recon=2.8556 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4992 init_acc_corrupt=0.4663 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=54400 micro_steps=108800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8235 acc_all=0.5908 loss_corrupt=2.8235 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8235 loss_recon=2.8235 loss_meanflow=0.0000 mean_model_t=0.5054 mean_corrupt_t=0.5054 wrong_frac=0.4951 init_acc_corrupt=0.4706 init_gold_top10=0.4994 init_gold_top100=0.5051
+step=54500 micro_steps=109000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8585 acc_all=0.5863 loss_corrupt=2.8585 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8585 loss_recon=2.8585 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4995 init_acc_corrupt=0.4661 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=54600 micro_steps=109200 elapsed=34.1s lr=3.000000e-04 loss_all=2.8695 acc_all=0.5847 loss_corrupt=2.8695 acc_corrupt=0.5847 corrupt_frac=1.0000 loss=2.8695 loss_recon=2.8695 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5013 init_acc_corrupt=0.4636 init_gold_top10=0.4930 init_gold_top100=0.4989
+step=54700 micro_steps=109400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8933 acc_all=0.5815 loss_corrupt=2.8933 acc_corrupt=0.5815 corrupt_frac=1.0000 loss=2.8933 loss_recon=2.8933 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5030 init_acc_corrupt=0.4614 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=54800 micro_steps=109600 elapsed=34.2s lr=3.000000e-04 loss_all=2.8695 acc_all=0.5855 loss_corrupt=2.8695 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8695 loss_recon=2.8695 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5005 init_acc_corrupt=0.4654 init_gold_top10=0.4938 init_gold_top100=0.4996
+step=54900 micro_steps=109800 elapsed=34.1s lr=3.000000e-04 loss_all=2.8462 acc_all=0.5879 loss_corrupt=2.8462 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8462 loss_recon=2.8462 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4988 init_acc_corrupt=0.4670 init_gold_top10=0.4957 init_gold_top100=0.5013
+step=55000 micro_steps=110000 elapsed=34.1s lr=3.000000e-04 loss_all=2.8547 acc_all=0.5862 loss_corrupt=2.8547 acc_corrupt=0.5862 corrupt_frac=1.0000 loss=2.8547 loss_recon=2.8547 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5000 init_acc_corrupt=0.4657 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=55100 micro_steps=110200 elapsed=43.3s lr=3.000000e-04 loss_all=2.8845 acc_all=0.5829 loss_corrupt=2.8845 acc_corrupt=0.5829 corrupt_frac=1.0000 loss=2.8845 loss_recon=2.8845 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5024 init_acc_corrupt=0.4625 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=55200 micro_steps=110400 elapsed=48.4s lr=3.000000e-04 loss_all=2.8399 acc_all=0.5884 loss_corrupt=2.8399 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8399 loss_recon=2.8399 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4983 init_acc_corrupt=0.4674 init_gold_top10=0.4962 init_gold_top100=0.5018
+step=55300 micro_steps=110600 elapsed=44.8s lr=3.000000e-04 loss_all=2.8640 acc_all=0.5861 loss_corrupt=2.8640 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8640 loss_recon=2.8640 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4993 init_acc_corrupt=0.4661 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=55400 micro_steps=110800 elapsed=41.5s lr=3.000000e-04 loss_all=2.8878 acc_all=0.5824 loss_corrupt=2.8878 acc_corrupt=0.5824 corrupt_frac=1.0000 loss=2.8878 loss_recon=2.8878 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5021 init_acc_corrupt=0.4623 init_gold_top10=0.4922 init_gold_top100=0.4980
+step=55500 micro_steps=111000 elapsed=42.3s lr=3.000000e-04 loss_all=2.8650 acc_all=0.5860 loss_corrupt=2.8650 acc_corrupt=0.5860 corrupt_frac=1.0000 loss=2.8650 loss_recon=2.8650 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5002 init_acc_corrupt=0.4650 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=55600 micro_steps=111200 elapsed=42.3s lr=3.000000e-04 loss_all=2.8751 acc_all=0.5842 loss_corrupt=2.8751 acc_corrupt=0.5842 corrupt_frac=1.0000 loss=2.8751 loss_recon=2.8751 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5017 init_acc_corrupt=0.4637 init_gold_top10=0.4927 init_gold_top100=0.4984
+step=55700 micro_steps=111400 elapsed=45.0s lr=3.000000e-04 loss_all=2.8701 acc_all=0.5843 loss_corrupt=2.8701 acc_corrupt=0.5843 corrupt_frac=1.0000 loss=2.8701 loss_recon=2.8701 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4647 init_gold_top10=0.4936 init_gold_top100=0.4994
+step=55800 micro_steps=111600 elapsed=41.2s lr=3.000000e-04 loss_all=2.8643 acc_all=0.5849 loss_corrupt=2.8643 acc_corrupt=0.5849 corrupt_frac=1.0000 loss=2.8643 loss_recon=2.8643 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5022 init_acc_corrupt=0.4636 init_gold_top10=0.4923 init_gold_top100=0.4980
+step=55900 micro_steps=111800 elapsed=45.4s lr=3.000000e-04 loss_all=2.8634 acc_all=0.5855 loss_corrupt=2.8634 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8634 loss_recon=2.8634 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5000 init_acc_corrupt=0.4652 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=56000 micro_steps=112000 elapsed=47.5s lr=3.000000e-04 loss_all=2.8291 acc_all=0.5903 loss_corrupt=2.8291 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8291 loss_recon=2.8291 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4965 init_acc_corrupt=0.4689 init_gold_top10=0.4980 init_gold_top100=0.5037
+step=56100 micro_steps=112200 elapsed=42.8s lr=3.000000e-04 loss_all=2.9128 acc_all=0.5798 loss_corrupt=2.9128 acc_corrupt=0.5798 corrupt_frac=1.0000 loss=2.9128 loss_recon=2.9128 loss_meanflow=0.0000 mean_model_t=0.4952 mean_corrupt_t=0.4952 wrong_frac=0.5051 init_acc_corrupt=0.4598 init_gold_top10=0.4891 init_gold_top100=0.4951
+step=56200 micro_steps=112400 elapsed=41.0s lr=3.000000e-04 loss_all=2.8666 acc_all=0.5851 loss_corrupt=2.8666 acc_corrupt=0.5851 corrupt_frac=1.0000 loss=2.8666 loss_recon=2.8666 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5008 init_acc_corrupt=0.4645 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=56300 micro_steps=112600 elapsed=40.1s lr=3.000000e-04 loss_all=2.8530 acc_all=0.5869 loss_corrupt=2.8530 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8530 loss_recon=2.8530 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4998 init_acc_corrupt=0.4661 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=56400 micro_steps=112800 elapsed=43.5s lr=3.000000e-04 loss_all=2.8930 acc_all=0.5816 loss_corrupt=2.8930 acc_corrupt=0.5816 corrupt_frac=1.0000 loss=2.8930 loss_recon=2.8930 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5046 init_acc_corrupt=0.4606 init_gold_top10=0.4897 init_gold_top100=0.4955
+step=56500 micro_steps=113000 elapsed=47.2s lr=3.000000e-04 loss_all=2.8634 acc_all=0.5856 loss_corrupt=2.8634 acc_corrupt=0.5856 corrupt_frac=1.0000 loss=2.8634 loss_recon=2.8634 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5008 init_acc_corrupt=0.4648 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=56600 micro_steps=113200 elapsed=45.6s lr=3.000000e-04 loss_all=2.8398 acc_all=0.5888 loss_corrupt=2.8398 acc_corrupt=0.5888 corrupt_frac=1.0000 loss=2.8398 loss_recon=2.8398 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4976 init_acc_corrupt=0.4675 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=56700 micro_steps=113400 elapsed=40.4s lr=3.000000e-04 loss_all=2.8498 acc_all=0.5874 loss_corrupt=2.8498 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8498 loss_recon=2.8498 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4983 init_acc_corrupt=0.4671 init_gold_top10=0.4960 init_gold_top100=0.5018
+step=56800 micro_steps=113600 elapsed=40.9s lr=3.000000e-04 loss_all=2.8200 acc_all=0.5915 loss_corrupt=2.8200 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8200 loss_recon=2.8200 loss_meanflow=0.0000 mean_model_t=0.5048 mean_corrupt_t=0.5048 wrong_frac=0.4951 init_acc_corrupt=0.4710 init_gold_top10=0.4994 init_gold_top100=0.5050
+step=56900 micro_steps=113800 elapsed=41.5s lr=3.000000e-04 loss_all=2.8653 acc_all=0.5859 loss_corrupt=2.8653 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8653 loss_recon=2.8653 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5004 init_acc_corrupt=0.4648 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=57000 micro_steps=114000 elapsed=44.0s lr=3.000000e-04 loss_all=2.8950 acc_all=0.5812 loss_corrupt=2.8950 acc_corrupt=0.5812 corrupt_frac=1.0000 loss=2.8950 loss_recon=2.8950 loss_meanflow=0.0000 mean_model_t=0.4950 mean_corrupt_t=0.4950 wrong_frac=0.5050 init_acc_corrupt=0.4599 init_gold_top10=0.4893 init_gold_top100=0.4951
+step=57100 micro_steps=114200 elapsed=48.2s lr=3.000000e-04 loss_all=2.8380 acc_all=0.5892 loss_corrupt=2.8380 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8380 loss_recon=2.8380 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4975 init_acc_corrupt=0.4688 init_gold_top10=0.4969 init_gold_top100=0.5027
+step=57200 micro_steps=114400 elapsed=54.2s lr=3.000000e-04 loss_all=2.8395 acc_all=0.5883 loss_corrupt=2.8395 acc_corrupt=0.5883 corrupt_frac=1.0000 loss=2.8395 loss_recon=2.8395 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4983 init_acc_corrupt=0.4672 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=57300 micro_steps=114600 elapsed=50.3s lr=3.000000e-04 loss_all=2.8439 acc_all=0.5882 loss_corrupt=2.8439 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8439 loss_recon=2.8439 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4984 init_acc_corrupt=0.4667 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=57400 micro_steps=114800 elapsed=40.6s lr=3.000000e-04 loss_all=2.8748 acc_all=0.5842 loss_corrupt=2.8748 acc_corrupt=0.5842 corrupt_frac=1.0000 loss=2.8748 loss_recon=2.8748 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5012 init_acc_corrupt=0.4641 init_gold_top10=0.4930 init_gold_top100=0.4989
+step=57500 micro_steps=115000 elapsed=40.9s lr=3.000000e-04 loss_all=2.8497 acc_all=0.5874 loss_corrupt=2.8497 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8497 loss_recon=2.8497 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4992 init_acc_corrupt=0.4664 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=57600 micro_steps=115200 elapsed=41.0s lr=3.000000e-04 loss_all=2.8421 acc_all=0.5885 loss_corrupt=2.8421 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8421 loss_recon=2.8421 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4981 init_acc_corrupt=0.4678 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=57700 micro_steps=115400 elapsed=40.8s lr=3.000000e-04 loss_all=2.8656 acc_all=0.5849 loss_corrupt=2.8656 acc_corrupt=0.5849 corrupt_frac=1.0000 loss=2.8656 loss_recon=2.8656 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5015 init_acc_corrupt=0.4639 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=57800 micro_steps=115600 elapsed=42.6s lr=3.000000e-04 loss_all=2.8165 acc_all=0.5917 loss_corrupt=2.8165 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8165 loss_recon=2.8165 loss_meanflow=0.0000 mean_model_t=0.5060 mean_corrupt_t=0.5060 wrong_frac=0.4943 init_acc_corrupt=0.4717 init_gold_top10=0.5001 init_gold_top100=0.5058
+step=57900 micro_steps=115800 elapsed=47.3s lr=3.000000e-04 loss_all=2.8531 acc_all=0.5871 loss_corrupt=2.8531 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8531 loss_recon=2.8531 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4992 init_acc_corrupt=0.4659 init_gold_top10=0.4954 init_gold_top100=0.5010
+step=58000 micro_steps=116000 elapsed=47.9s lr=3.000000e-04 loss_all=2.8838 acc_all=0.5833 loss_corrupt=2.8838 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8838 loss_recon=2.8838 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5024 init_acc_corrupt=0.4625 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=58100 micro_steps=116200 elapsed=64.9s lr=3.000000e-04 loss_all=2.8875 acc_all=0.5826 loss_corrupt=2.8875 acc_corrupt=0.5826 corrupt_frac=1.0000 loss=2.8875 loss_recon=2.8875 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5031 init_acc_corrupt=0.4619 init_gold_top10=0.4912 init_gold_top100=0.4971
+step=58200 micro_steps=116400 elapsed=56.7s lr=3.000000e-04 loss_all=2.8512 acc_all=0.5871 loss_corrupt=2.8512 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8512 loss_recon=2.8512 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4997 init_acc_corrupt=0.4660 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=58300 micro_steps=116600 elapsed=42.7s lr=3.000000e-04 loss_all=2.8743 acc_all=0.5841 loss_corrupt=2.8743 acc_corrupt=0.5841 corrupt_frac=1.0000 loss=2.8743 loss_recon=2.8743 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5027 init_acc_corrupt=0.4621 init_gold_top10=0.4916 init_gold_top100=0.4974
+step=58400 micro_steps=116800 elapsed=41.2s lr=3.000000e-04 loss_all=2.8274 acc_all=0.5900 loss_corrupt=2.8274 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8274 loss_recon=2.8274 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4970 init_acc_corrupt=0.4689 init_gold_top10=0.4977 init_gold_top100=0.5032
+step=58500 micro_steps=117000 elapsed=41.2s lr=3.000000e-04 loss_all=2.8535 acc_all=0.5869 loss_corrupt=2.8535 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8535 loss_recon=2.8535 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.5000 init_acc_corrupt=0.4657 init_gold_top10=0.4945 init_gold_top100=0.5001
+step=58600 micro_steps=117200 elapsed=41.2s lr=3.000000e-04 loss_all=2.8422 acc_all=0.5884 loss_corrupt=2.8422 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8422 loss_recon=2.8422 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4986 init_acc_corrupt=0.4671 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=58700 micro_steps=117400 elapsed=41.2s lr=3.000000e-04 loss_all=2.8450 acc_all=0.5879 loss_corrupt=2.8450 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8450 loss_recon=2.8450 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4994 init_acc_corrupt=0.4668 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=58800 micro_steps=117600 elapsed=41.2s lr=3.000000e-04 loss_all=2.8562 acc_all=0.5867 loss_corrupt=2.8562 acc_corrupt=0.5867 corrupt_frac=1.0000 loss=2.8562 loss_recon=2.8562 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5000 init_acc_corrupt=0.4655 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=58900 micro_steps=117800 elapsed=41.7s lr=3.000000e-04 loss_all=2.8508 acc_all=0.5873 loss_corrupt=2.8508 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8508 loss_recon=2.8508 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4999 init_acc_corrupt=0.4657 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=59000 micro_steps=118000 elapsed=47.8s lr=3.000000e-04 loss_all=2.8497 acc_all=0.5877 loss_corrupt=2.8497 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8497 loss_recon=2.8497 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4990 init_acc_corrupt=0.4668 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=59100 micro_steps=118200 elapsed=50.0s lr=3.000000e-04 loss_all=2.8728 acc_all=0.5844 loss_corrupt=2.8728 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8728 loss_recon=2.8728 loss_meanflow=0.0000 mean_model_t=0.4966 mean_corrupt_t=0.4966 wrong_frac=0.5031 init_acc_corrupt=0.4622 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=59200 micro_steps=118400 elapsed=57.9s lr=3.000000e-04 loss_all=2.8604 acc_all=0.5861 loss_corrupt=2.8604 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8604 loss_recon=2.8604 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5004 init_acc_corrupt=0.4651 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=59300 micro_steps=118600 elapsed=68.0s lr=3.000000e-04 loss_all=2.8565 acc_all=0.5863 loss_corrupt=2.8565 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8565 loss_recon=2.8565 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4999 init_acc_corrupt=0.4653 init_gold_top10=0.4944 init_gold_top100=0.5003
+step=59400 micro_steps=118800 elapsed=42.5s lr=3.000000e-04 loss_all=2.8418 acc_all=0.5879 loss_corrupt=2.8418 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8418 loss_recon=2.8418 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4995 init_acc_corrupt=0.4662 init_gold_top10=0.4951 init_gold_top100=0.5007
+step=59500 micro_steps=119000 elapsed=41.7s lr=3.000000e-04 loss_all=2.8178 acc_all=0.5916 loss_corrupt=2.8178 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8178 loss_recon=2.8178 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4956 init_acc_corrupt=0.4705 init_gold_top10=0.4990 init_gold_top100=0.5045
+step=59600 micro_steps=119200 elapsed=41.6s lr=3.000000e-04 loss_all=2.8784 acc_all=0.5833 loss_corrupt=2.8784 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8784 loss_recon=2.8784 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5032 init_acc_corrupt=0.4613 init_gold_top10=0.4912 init_gold_top100=0.4969
+step=59700 micro_steps=119400 elapsed=47.4s lr=3.000000e-04 loss_all=2.8592 acc_all=0.5865 loss_corrupt=2.8592 acc_corrupt=0.5865 corrupt_frac=1.0000 loss=2.8592 loss_recon=2.8592 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4999 init_acc_corrupt=0.4655 init_gold_top10=0.4945 init_gold_top100=0.5003
+step=59800 micro_steps=119600 elapsed=69.1s lr=3.000000e-04 loss_all=2.8590 acc_all=0.5861 loss_corrupt=2.8590 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8590 loss_recon=2.8590 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5012 init_acc_corrupt=0.4642 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=59900 micro_steps=119800 elapsed=57.4s lr=3.000000e-04 loss_all=2.8616 acc_all=0.5861 loss_corrupt=2.8616 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8616 loss_recon=2.8616 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5004 init_acc_corrupt=0.4651 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=60000 micro_steps=120000 elapsed=43.4s lr=3.000000e-04 loss_all=2.8428 acc_all=0.5882 loss_corrupt=2.8428 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8428 loss_recon=2.8428 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.4996 init_acc_corrupt=0.4657 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=60100 micro_steps=120200 elapsed=42.4s lr=3.000000e-04 loss_all=2.9038 acc_all=0.5802 loss_corrupt=2.9038 acc_corrupt=0.5802 corrupt_frac=1.0000 loss=2.9038 loss_recon=2.9038 loss_meanflow=0.0000 mean_model_t=0.4945 mean_corrupt_t=0.4945 wrong_frac=0.5061 init_acc_corrupt=0.4580 init_gold_top10=0.4882 init_gold_top100=0.4940
+step=60200 micro_steps=120400 elapsed=59.4s lr=3.000000e-04 loss_all=2.9072 acc_all=0.5800 loss_corrupt=2.9072 acc_corrupt=0.5800 corrupt_frac=1.0000 loss=2.9072 loss_recon=2.9072 loss_meanflow=0.0000 mean_model_t=0.4932 mean_corrupt_t=0.4932 wrong_frac=0.5064 init_acc_corrupt=0.4578 init_gold_top10=0.4879 init_gold_top100=0.4937
+step=60300 micro_steps=120600 elapsed=56.0s lr=3.000000e-04 loss_all=2.8334 acc_all=0.5892 loss_corrupt=2.8334 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8334 loss_recon=2.8334 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4986 init_acc_corrupt=0.4675 init_gold_top10=0.4960 init_gold_top100=0.5016
+step=60400 micro_steps=120800 elapsed=43.5s lr=3.000000e-04 loss_all=2.8581 acc_all=0.5868 loss_corrupt=2.8581 acc_corrupt=0.5868 corrupt_frac=1.0000 loss=2.8581 loss_recon=2.8581 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5001 init_acc_corrupt=0.4652 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=60500 micro_steps=121000 elapsed=36.7s lr=3.000000e-04 loss_all=2.8369 acc_all=0.5893 loss_corrupt=2.8369 acc_corrupt=0.5893 corrupt_frac=1.0000 loss=2.8369 loss_recon=2.8369 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4972 init_acc_corrupt=0.4687 init_gold_top10=0.4973 init_gold_top100=0.5030
+step=60600 micro_steps=121200 elapsed=46.5s lr=3.000000e-04 loss_all=2.8525 acc_all=0.5868 loss_corrupt=2.8525 acc_corrupt=0.5868 corrupt_frac=1.0000 loss=2.8525 loss_recon=2.8525 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4998 init_acc_corrupt=0.4657 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=60700 micro_steps=121400 elapsed=38.2s lr=3.000000e-04 loss_all=2.8267 acc_all=0.5907 loss_corrupt=2.8267 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8267 loss_recon=2.8267 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4962 init_acc_corrupt=0.4699 init_gold_top10=0.4982 init_gold_top100=0.5039
+step=60800 micro_steps=121600 elapsed=37.0s lr=3.000000e-04 loss_all=2.8827 acc_all=0.5829 loss_corrupt=2.8827 acc_corrupt=0.5829 corrupt_frac=1.0000 loss=2.8827 loss_recon=2.8827 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5034 init_acc_corrupt=0.4614 init_gold_top10=0.4908 init_gold_top100=0.4967
+step=60900 micro_steps=121800 elapsed=43.6s lr=3.000000e-04 loss_all=2.8569 acc_all=0.5863 loss_corrupt=2.8569 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8569 loss_recon=2.8569 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5003 init_acc_corrupt=0.4652 init_gold_top10=0.4942 init_gold_top100=0.4998
+step=61000 micro_steps=122000 elapsed=37.7s lr=3.000000e-04 loss_all=2.8661 acc_all=0.5852 loss_corrupt=2.8661 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8661 loss_recon=2.8661 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5018 init_acc_corrupt=0.4627 init_gold_top10=0.4925 init_gold_top100=0.4984
+step=61100 micro_steps=122200 elapsed=35.5s lr=3.000000e-04 loss_all=2.8579 acc_all=0.5861 loss_corrupt=2.8579 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8579 loss_recon=2.8579 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5009 init_acc_corrupt=0.4647 init_gold_top10=0.4933 init_gold_top100=0.4992
+step=61200 micro_steps=122400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8164 acc_all=0.5914 loss_corrupt=2.8164 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8164 loss_recon=2.8164 loss_meanflow=0.0000 mean_model_t=0.5046 mean_corrupt_t=0.5046 wrong_frac=0.4955 init_acc_corrupt=0.4710 init_gold_top10=0.4990 init_gold_top100=0.5047
+step=61300 micro_steps=122600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8414 acc_all=0.5887 loss_corrupt=2.8414 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8414 loss_recon=2.8414 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4986 init_acc_corrupt=0.4668 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=61400 micro_steps=122800 elapsed=34.9s lr=3.000000e-04 loss_all=2.8658 acc_all=0.5854 loss_corrupt=2.8658 acc_corrupt=0.5854 corrupt_frac=1.0000 loss=2.8658 loss_recon=2.8658 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5002 init_acc_corrupt=0.4648 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=61500 micro_steps=123000 elapsed=40.3s lr=3.000000e-04 loss_all=2.8555 acc_all=0.5866 loss_corrupt=2.8555 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8555 loss_recon=2.8555 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5006 init_acc_corrupt=0.4650 init_gold_top10=0.4938 init_gold_top100=0.4995
+step=61600 micro_steps=123200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8651 acc_all=0.5855 loss_corrupt=2.8651 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8651 loss_recon=2.8651 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4632 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=61700 micro_steps=123400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8481 acc_all=0.5877 loss_corrupt=2.8481 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8481 loss_recon=2.8481 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4989 init_acc_corrupt=0.4667 init_gold_top10=0.4956 init_gold_top100=0.5012
+step=61800 micro_steps=123600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8375 acc_all=0.5885 loss_corrupt=2.8375 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8375 loss_recon=2.8375 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4991 init_acc_corrupt=0.4664 init_gold_top10=0.4955 init_gold_top100=0.5011
+step=61900 micro_steps=123800 elapsed=35.7s lr=3.000000e-04 loss_all=2.8397 acc_all=0.5881 loss_corrupt=2.8397 acc_corrupt=0.5881 corrupt_frac=1.0000 loss=2.8397 loss_recon=2.8397 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5003 init_acc_corrupt=0.4652 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=62000 micro_steps=124000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8451 acc_all=0.5875 loss_corrupt=2.8451 acc_corrupt=0.5875 corrupt_frac=1.0000 loss=2.8451 loss_recon=2.8451 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4998 init_acc_corrupt=0.4657 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=62100 micro_steps=124200 elapsed=35.3s lr=3.000000e-04 loss_all=2.8076 acc_all=0.5925 loss_corrupt=2.8076 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8076 loss_recon=2.8076 loss_meanflow=0.0000 mean_model_t=0.5046 mean_corrupt_t=0.5046 wrong_frac=0.4954 init_acc_corrupt=0.4712 init_gold_top10=0.4993 init_gold_top100=0.5048
+step=62200 micro_steps=124400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8255 acc_all=0.5909 loss_corrupt=2.8255 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8255 loss_recon=2.8255 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4971 init_acc_corrupt=0.4687 init_gold_top10=0.4974 init_gold_top100=0.5031
+step=62300 micro_steps=124600 elapsed=39.7s lr=3.000000e-04 loss_all=2.8792 acc_all=0.5833 loss_corrupt=2.8792 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8792 loss_recon=2.8792 loss_meanflow=0.0000 mean_model_t=0.4958 mean_corrupt_t=0.4958 wrong_frac=0.5041 init_acc_corrupt=0.4615 init_gold_top10=0.4901 init_gold_top100=0.4959
+step=62400 micro_steps=124800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8249 acc_all=0.5899 loss_corrupt=2.8249 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8249 loss_recon=2.8249 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4974 init_acc_corrupt=0.4687 init_gold_top10=0.4971 init_gold_top100=0.5027
+step=62500 micro_steps=125000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8909 acc_all=0.5821 loss_corrupt=2.8909 acc_corrupt=0.5821 corrupt_frac=1.0000 loss=2.8909 loss_recon=2.8909 loss_meanflow=0.0000 mean_model_t=0.4954 mean_corrupt_t=0.4954 wrong_frac=0.5044 init_acc_corrupt=0.4609 init_gold_top10=0.4898 init_gold_top100=0.4956
+step=62600 micro_steps=125200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8066 acc_all=0.5933 loss_corrupt=2.8066 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.8066 loss_recon=2.8066 loss_meanflow=0.0000 mean_model_t=0.5064 mean_corrupt_t=0.5064 wrong_frac=0.4933 init_acc_corrupt=0.4726 init_gold_top10=0.5012 init_gold_top100=0.5068
+step=62700 micro_steps=125400 elapsed=33.6s lr=3.000000e-04 loss_all=2.8728 acc_all=0.5840 loss_corrupt=2.8728 acc_corrupt=0.5840 corrupt_frac=1.0000 loss=2.8728 loss_recon=2.8728 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5032 init_acc_corrupt=0.4621 init_gold_top10=0.4913 init_gold_top100=0.4970
+step=62800 micro_steps=125600 elapsed=34.9s lr=3.000000e-04 loss_all=2.8323 acc_all=0.5897 loss_corrupt=2.8323 acc_corrupt=0.5897 corrupt_frac=1.0000 loss=2.8323 loss_recon=2.8323 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4974 init_acc_corrupt=0.4682 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=62900 micro_steps=125800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8853 acc_all=0.5829 loss_corrupt=2.8853 acc_corrupt=0.5829 corrupt_frac=1.0000 loss=2.8853 loss_recon=2.8853 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5034 init_acc_corrupt=0.4616 init_gold_top10=0.4907 init_gold_top100=0.4966
+step=63000 micro_steps=126000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8561 acc_all=0.5863 loss_corrupt=2.8561 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8561 loss_recon=2.8561 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4645 init_gold_top10=0.4935 init_gold_top100=0.4996
+step=63100 micro_steps=126200 elapsed=35.2s lr=3.000000e-04 loss_all=2.8646 acc_all=0.5854 loss_corrupt=2.8646 acc_corrupt=0.5854 corrupt_frac=1.0000 loss=2.8646 loss_recon=2.8646 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5016 init_acc_corrupt=0.4642 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=63200 micro_steps=126400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8661 acc_all=0.5854 loss_corrupt=2.8661 acc_corrupt=0.5854 corrupt_frac=1.0000 loss=2.8661 loss_recon=2.8661 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5010 init_acc_corrupt=0.4638 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=63300 micro_steps=126600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8569 acc_all=0.5866 loss_corrupt=2.8569 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8569 loss_recon=2.8569 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5005 init_acc_corrupt=0.4645 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=63400 micro_steps=126800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8562 acc_all=0.5873 loss_corrupt=2.8562 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8562 loss_recon=2.8562 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4997 init_acc_corrupt=0.4658 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=63500 micro_steps=127000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8503 acc_all=0.5867 loss_corrupt=2.8503 acc_corrupt=0.5867 corrupt_frac=1.0000 loss=2.8503 loss_recon=2.8503 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5015 init_acc_corrupt=0.4641 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=63600 micro_steps=127200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8629 acc_all=0.5854 loss_corrupt=2.8629 acc_corrupt=0.5854 corrupt_frac=1.0000 loss=2.8629 loss_recon=2.8629 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5021 init_acc_corrupt=0.4631 init_gold_top10=0.4923 init_gold_top100=0.4980
+step=63700 micro_steps=127400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8410 acc_all=0.5885 loss_corrupt=2.8410 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8410 loss_recon=2.8410 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4992 init_acc_corrupt=0.4665 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=63800 micro_steps=127600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8451 acc_all=0.5874 loss_corrupt=2.8451 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8451 loss_recon=2.8451 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5005 init_acc_corrupt=0.4652 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=63900 micro_steps=127800 elapsed=33.8s lr=3.000000e-04 loss_all=2.7918 acc_all=0.5947 loss_corrupt=2.7918 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7918 loss_recon=2.7918 loss_meanflow=0.0000 mean_model_t=0.5066 mean_corrupt_t=0.5066 wrong_frac=0.4936 init_acc_corrupt=0.4735 init_gold_top10=0.5011 init_gold_top100=0.5065
+step=64000 micro_steps=128000 elapsed=33.6s lr=3.000000e-04 loss_all=2.8393 acc_all=0.5888 loss_corrupt=2.8393 acc_corrupt=0.5888 corrupt_frac=1.0000 loss=2.8393 loss_recon=2.8393 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4986 init_acc_corrupt=0.4672 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=64100 micro_steps=128200 elapsed=35.1s lr=3.000000e-04 loss_all=2.8326 acc_all=0.5892 loss_corrupt=2.8326 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8326 loss_recon=2.8326 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4985 init_acc_corrupt=0.4675 init_gold_top10=0.4960 init_gold_top100=0.5016
+step=64200 micro_steps=128400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8753 acc_all=0.5838 loss_corrupt=2.8753 acc_corrupt=0.5838 corrupt_frac=1.0000 loss=2.8753 loss_recon=2.8753 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5032 init_acc_corrupt=0.4618 init_gold_top10=0.4910 init_gold_top100=0.4969
+step=64300 micro_steps=128600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8322 acc_all=0.5896 loss_corrupt=2.8322 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8322 loss_recon=2.8322 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4979 init_acc_corrupt=0.4672 init_gold_top10=0.4964 init_gold_top100=0.5024
+step=64400 micro_steps=128800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8392 acc_all=0.5887 loss_corrupt=2.8392 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8392 loss_recon=2.8392 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4994 init_acc_corrupt=0.4664 init_gold_top10=0.4952 init_gold_top100=0.5007
+step=64500 micro_steps=129000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8453 acc_all=0.5879 loss_corrupt=2.8453 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8453 loss_recon=2.8453 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4989 init_acc_corrupt=0.4663 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=64600 micro_steps=129200 elapsed=33.8s lr=3.000000e-04 loss_all=2.8589 acc_all=0.5866 loss_corrupt=2.8589 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8589 loss_recon=2.8589 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5006 init_acc_corrupt=0.4649 init_gold_top10=0.4938 init_gold_top100=0.4995
+step=64700 micro_steps=129400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8373 acc_all=0.5887 loss_corrupt=2.8373 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8373 loss_recon=2.8373 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4993 init_acc_corrupt=0.4660 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=64800 micro_steps=129600 elapsed=34.0s lr=3.000000e-04 loss_all=2.8454 acc_all=0.5875 loss_corrupt=2.8454 acc_corrupt=0.5875 corrupt_frac=1.0000 loss=2.8454 loss_recon=2.8454 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5006 init_acc_corrupt=0.4652 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=64900 micro_steps=129800 elapsed=33.9s lr=3.000000e-04 loss_all=2.8631 acc_all=0.5857 loss_corrupt=2.8631 acc_corrupt=0.5857 corrupt_frac=1.0000 loss=2.8631 loss_recon=2.8631 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5013 init_acc_corrupt=0.4635 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=65000 micro_steps=130000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8608 acc_all=0.5855 loss_corrupt=2.8608 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8608 loss_recon=2.8608 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5028 init_acc_corrupt=0.4629 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=65100 micro_steps=130200 elapsed=35.3s lr=3.000000e-04 loss_all=2.8524 acc_all=0.5870 loss_corrupt=2.8524 acc_corrupt=0.5870 corrupt_frac=1.0000 loss=2.8524 loss_recon=2.8524 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4992 init_acc_corrupt=0.4658 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=65200 micro_steps=130400 elapsed=34.0s lr=3.000000e-04 loss_all=2.8629 acc_all=0.5859 loss_corrupt=2.8629 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8629 loss_recon=2.8629 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5011 init_acc_corrupt=0.4640 init_gold_top10=0.4932 init_gold_top100=0.4991
+step=65300 micro_steps=130600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8737 acc_all=0.5840 loss_corrupt=2.8737 acc_corrupt=0.5840 corrupt_frac=1.0000 loss=2.8737 loss_recon=2.8737 loss_meanflow=0.0000 mean_model_t=0.4961 mean_corrupt_t=0.4961 wrong_frac=0.5035 init_acc_corrupt=0.4617 init_gold_top10=0.4909 init_gold_top100=0.4966
+step=65400 micro_steps=130800 elapsed=33.6s lr=3.000000e-04 loss_all=2.8459 acc_all=0.5883 loss_corrupt=2.8459 acc_corrupt=0.5883 corrupt_frac=1.0000 loss=2.8459 loss_recon=2.8459 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4992 init_acc_corrupt=0.4666 init_gold_top10=0.4953 init_gold_top100=0.5009
+step=65500 micro_steps=131000 elapsed=33.6s lr=3.000000e-04 loss_all=2.8484 acc_all=0.5876 loss_corrupt=2.8484 acc_corrupt=0.5876 corrupt_frac=1.0000 loss=2.8484 loss_recon=2.8484 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4996 init_acc_corrupt=0.4660 init_gold_top10=0.4949 init_gold_top100=0.5006
+step=65600 micro_steps=131200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8309 acc_all=0.5900 loss_corrupt=2.8309 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8309 loss_recon=2.8309 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4979 init_acc_corrupt=0.4680 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=65700 micro_steps=131400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8254 acc_all=0.5909 loss_corrupt=2.8254 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8254 loss_recon=2.8254 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4969 init_acc_corrupt=0.4686 init_gold_top10=0.4974 init_gold_top100=0.5032
+step=65800 micro_steps=131600 elapsed=34.0s lr=3.000000e-04 loss_all=2.9038 acc_all=0.5805 loss_corrupt=2.9038 acc_corrupt=0.5805 corrupt_frac=1.0000 loss=2.9038 loss_recon=2.9038 loss_meanflow=0.0000 mean_model_t=0.4938 mean_corrupt_t=0.4938 wrong_frac=0.5059 init_acc_corrupt=0.4585 init_gold_top10=0.4883 init_gold_top100=0.4942
+step=65900 micro_steps=131800 elapsed=33.9s lr=3.000000e-04 loss_all=2.8595 acc_all=0.5863 loss_corrupt=2.8595 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8595 loss_recon=2.8595 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5010 init_acc_corrupt=0.4637 init_gold_top10=0.4935 init_gold_top100=0.4991
+step=66000 micro_steps=132000 elapsed=34.2s lr=3.000000e-04 loss_all=2.8374 acc_all=0.5890 loss_corrupt=2.8374 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8374 loss_recon=2.8374 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4983 init_acc_corrupt=0.4677 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=66100 micro_steps=132200 elapsed=35.8s lr=3.000000e-04 loss_all=2.8261 acc_all=0.5906 loss_corrupt=2.8261 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8261 loss_recon=2.8261 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4962 init_acc_corrupt=0.4700 init_gold_top10=0.4982 init_gold_top100=0.5039
+step=66200 micro_steps=132400 elapsed=33.9s lr=3.000000e-04 loss_all=2.8379 acc_all=0.5882 loss_corrupt=2.8379 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8379 loss_recon=2.8379 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4995 init_acc_corrupt=0.4658 init_gold_top10=0.4949 init_gold_top100=0.5006
+step=66300 micro_steps=132600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8428 acc_all=0.5883 loss_corrupt=2.8428 acc_corrupt=0.5883 corrupt_frac=1.0000 loss=2.8428 loss_recon=2.8428 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4983 init_acc_corrupt=0.4673 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=66400 micro_steps=132800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8703 acc_all=0.5844 loss_corrupt=2.8703 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8703 loss_recon=2.8703 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5030 init_acc_corrupt=0.4621 init_gold_top10=0.4912 init_gold_top100=0.4971
+step=66500 micro_steps=133000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8362 acc_all=0.5894 loss_corrupt=2.8362 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8362 loss_recon=2.8362 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4974 init_acc_corrupt=0.4682 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=66600 micro_steps=133200 elapsed=33.8s lr=3.000000e-04 loss_all=2.8588 acc_all=0.5857 loss_corrupt=2.8588 acc_corrupt=0.5857 corrupt_frac=1.0000 loss=2.8588 loss_recon=2.8588 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5015 init_acc_corrupt=0.4638 init_gold_top10=0.4930 init_gold_top100=0.4986
+step=66700 micro_steps=133400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8276 acc_all=0.5902 loss_corrupt=2.8276 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8276 loss_recon=2.8276 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4983 init_acc_corrupt=0.4671 init_gold_top10=0.4963 init_gold_top100=0.5019
+step=66800 micro_steps=133600 elapsed=34.0s lr=3.000000e-04 loss_all=2.8883 acc_all=0.5819 loss_corrupt=2.8883 acc_corrupt=0.5819 corrupt_frac=1.0000 loss=2.8883 loss_recon=2.8883 loss_meanflow=0.0000 mean_model_t=0.4946 mean_corrupt_t=0.4946 wrong_frac=0.5057 init_acc_corrupt=0.4591 init_gold_top10=0.4885 init_gold_top100=0.4944
+step=66900 micro_steps=133800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8295 acc_all=0.5898 loss_corrupt=2.8295 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8295 loss_recon=2.8295 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4978 init_acc_corrupt=0.4677 init_gold_top10=0.4966 init_gold_top100=0.5023
+step=67000 micro_steps=134000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8581 acc_all=0.5861 loss_corrupt=2.8581 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8581 loss_recon=2.8581 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5012 init_acc_corrupt=0.4644 init_gold_top10=0.4932 init_gold_top100=0.4989
+step=67100 micro_steps=134200 elapsed=35.7s lr=3.000000e-04 loss_all=2.8942 acc_all=0.5813 loss_corrupt=2.8942 acc_corrupt=0.5813 corrupt_frac=1.0000 loss=2.8942 loss_recon=2.8942 loss_meanflow=0.0000 mean_model_t=0.4945 mean_corrupt_t=0.4945 wrong_frac=0.5057 init_acc_corrupt=0.4589 init_gold_top10=0.4885 init_gold_top100=0.4944
+step=67200 micro_steps=134400 elapsed=34.0s lr=3.000000e-04 loss_all=2.8570 acc_all=0.5865 loss_corrupt=2.8570 acc_corrupt=0.5865 corrupt_frac=1.0000 loss=2.8570 loss_recon=2.8570 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5019 init_acc_corrupt=0.4632 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=67300 micro_steps=134600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8132 acc_all=0.5920 loss_corrupt=2.8132 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8132 loss_recon=2.8132 loss_meanflow=0.0000 mean_model_t=0.5038 mean_corrupt_t=0.5038 wrong_frac=0.4959 init_acc_corrupt=0.4699 init_gold_top10=0.4986 init_gold_top100=0.5042
+step=67400 micro_steps=134800 elapsed=34.0s lr=3.000000e-04 loss_all=2.8750 acc_all=0.5840 loss_corrupt=2.8750 acc_corrupt=0.5840 corrupt_frac=1.0000 loss=2.8750 loss_recon=2.8750 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5028 init_acc_corrupt=0.4619 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=67500 micro_steps=135000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8417 acc_all=0.5885 loss_corrupt=2.8417 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8417 loss_recon=2.8417 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4987 init_acc_corrupt=0.4669 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=67600 micro_steps=135200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8314 acc_all=0.5896 loss_corrupt=2.8314 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8314 loss_recon=2.8314 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4977 init_acc_corrupt=0.4681 init_gold_top10=0.4968 init_gold_top100=0.5024
+step=67700 micro_steps=135400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8534 acc_all=0.5869 loss_corrupt=2.8534 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8534 loss_recon=2.8534 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5015 init_acc_corrupt=0.4644 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=67800 micro_steps=135600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8472 acc_all=0.5875 loss_corrupt=2.8472 acc_corrupt=0.5875 corrupt_frac=1.0000 loss=2.8472 loss_recon=2.8472 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5007 init_acc_corrupt=0.4651 init_gold_top10=0.4938 init_gold_top100=0.4994
+step=67900 micro_steps=135800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8543 acc_all=0.5865 loss_corrupt=2.8543 acc_corrupt=0.5865 corrupt_frac=1.0000 loss=2.8543 loss_recon=2.8543 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5009 init_acc_corrupt=0.4642 init_gold_top10=0.4937 init_gold_top100=0.4993
+step=68000 micro_steps=136000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8627 acc_all=0.5855 loss_corrupt=2.8627 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8627 loss_recon=2.8627 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4637 init_gold_top10=0.4925 init_gold_top100=0.4985
+step=68100 micro_steps=136200 elapsed=35.2s lr=3.000000e-04 loss_all=2.8591 acc_all=0.5861 loss_corrupt=2.8591 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8591 loss_recon=2.8591 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5008 init_acc_corrupt=0.4649 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=68200 micro_steps=136400 elapsed=34.0s lr=3.000000e-04 loss_all=2.8373 acc_all=0.5890 loss_corrupt=2.8373 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8373 loss_recon=2.8373 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4986 init_acc_corrupt=0.4670 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=68300 micro_steps=136600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8353 acc_all=0.5887 loss_corrupt=2.8353 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8353 loss_recon=2.8353 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4993 init_acc_corrupt=0.4664 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=68400 micro_steps=136800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8575 acc_all=0.5860 loss_corrupt=2.8575 acc_corrupt=0.5860 corrupt_frac=1.0000 loss=2.8575 loss_recon=2.8575 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5023 init_acc_corrupt=0.4634 init_gold_top10=0.4921 init_gold_top100=0.4978
+step=68500 micro_steps=137000 elapsed=34.0s lr=3.000000e-04 loss_all=2.7890 acc_all=0.5952 loss_corrupt=2.7890 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7890 loss_recon=2.7890 loss_meanflow=0.0000 mean_model_t=0.5068 mean_corrupt_t=0.5068 wrong_frac=0.4926 init_acc_corrupt=0.4735 init_gold_top10=0.5019 init_gold_top100=0.5075
+step=68600 micro_steps=137200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8646 acc_all=0.5858 loss_corrupt=2.8646 acc_corrupt=0.5858 corrupt_frac=1.0000 loss=2.8646 loss_recon=2.8646 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=68700 micro_steps=137400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8756 acc_all=0.5844 loss_corrupt=2.8756 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8756 loss_recon=2.8756 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5033 init_acc_corrupt=0.4621 init_gold_top10=0.4910 init_gold_top100=0.4969
+step=68800 micro_steps=137600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8380 acc_all=0.5887 loss_corrupt=2.8380 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8380 loss_recon=2.8380 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4991 init_acc_corrupt=0.4668 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=68900 micro_steps=137800 elapsed=34.0s lr=3.000000e-04 loss_all=2.8207 acc_all=0.5914 loss_corrupt=2.8207 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8207 loss_recon=2.8207 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4969 init_acc_corrupt=0.4690 init_gold_top10=0.4973 init_gold_top100=0.5033
+step=69000 micro_steps=138000 elapsed=34.1s lr=3.000000e-04 loss_all=2.8973 acc_all=0.5811 loss_corrupt=2.8973 acc_corrupt=0.5811 corrupt_frac=1.0000 loss=2.8973 loss_recon=2.8973 loss_meanflow=0.0000 mean_model_t=0.4939 mean_corrupt_t=0.4939 wrong_frac=0.5060 init_acc_corrupt=0.4584 init_gold_top10=0.4882 init_gold_top100=0.4941
+step=69100 micro_steps=138200 elapsed=35.5s lr=3.000000e-04 loss_all=2.8208 acc_all=0.5913 loss_corrupt=2.8208 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8208 loss_recon=2.8208 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4970 init_acc_corrupt=0.4687 init_gold_top10=0.4973 init_gold_top100=0.5032
+step=69200 micro_steps=138400 elapsed=33.9s lr=3.000000e-04 loss_all=2.9025 acc_all=0.5804 loss_corrupt=2.9025 acc_corrupt=0.5804 corrupt_frac=1.0000 loss=2.9025 loss_recon=2.9025 loss_meanflow=0.0000 mean_model_t=0.4939 mean_corrupt_t=0.4939 wrong_frac=0.5060 init_acc_corrupt=0.4585 init_gold_top10=0.4881 init_gold_top100=0.4941
+step=69300 micro_steps=138600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8768 acc_all=0.5833 loss_corrupt=2.8768 acc_corrupt=0.5833 corrupt_frac=1.0000 loss=2.8768 loss_recon=2.8768 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5035 init_acc_corrupt=0.4610 init_gold_top10=0.4908 init_gold_top100=0.4966
+step=69400 micro_steps=138800 elapsed=36.1s lr=3.000000e-04 loss_all=2.8764 acc_all=0.5844 loss_corrupt=2.8764 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8764 loss_recon=2.8764 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5025 init_acc_corrupt=0.4621 init_gold_top10=0.4917 init_gold_top100=0.4976
+step=69500 micro_steps=139000 elapsed=34.0s lr=3.000000e-04 loss_all=2.8478 acc_all=0.5876 loss_corrupt=2.8478 acc_corrupt=0.5876 corrupt_frac=1.0000 loss=2.8478 loss_recon=2.8478 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5004 init_acc_corrupt=0.4652 init_gold_top10=0.4941 init_gold_top100=0.4997
+step=69600 micro_steps=139200 elapsed=34.2s lr=3.000000e-04 loss_all=2.8441 acc_all=0.5880 loss_corrupt=2.8441 acc_corrupt=0.5880 corrupt_frac=1.0000 loss=2.8441 loss_recon=2.8441 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4994 init_acc_corrupt=0.4665 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=69700 micro_steps=139400 elapsed=34.1s lr=3.000000e-04 loss_all=2.8104 acc_all=0.5924 loss_corrupt=2.8104 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8104 loss_recon=2.8104 loss_meanflow=0.0000 mean_model_t=0.5046 mean_corrupt_t=0.5046 wrong_frac=0.4951 init_acc_corrupt=0.4710 init_gold_top10=0.4994 init_gold_top100=0.5050
+step=69800 micro_steps=139600 elapsed=33.6s lr=3.000000e-04 loss_all=2.8351 acc_all=0.5890 loss_corrupt=2.8351 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8351 loss_recon=2.8351 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4989 init_acc_corrupt=0.4667 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=69900 micro_steps=139800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8298 acc_all=0.5898 loss_corrupt=2.8298 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8298 loss_recon=2.8298 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4982 init_acc_corrupt=0.4676 init_gold_top10=0.4964 init_gold_top100=0.5019
+step=70000 micro_steps=140000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8277 acc_all=0.5899 loss_corrupt=2.8277 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8277 loss_recon=2.8277 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4980 init_acc_corrupt=0.4679 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=70100 micro_steps=140200 elapsed=35.1s lr=3.000000e-04 loss_all=2.8286 acc_all=0.5901 loss_corrupt=2.8286 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8286 loss_recon=2.8286 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4975 init_acc_corrupt=0.4677 init_gold_top10=0.4969 init_gold_top100=0.5027
+step=70200 micro_steps=140400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8029 acc_all=0.5932 loss_corrupt=2.8029 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.8029 loss_recon=2.8029 loss_meanflow=0.0000 mean_model_t=0.5048 mean_corrupt_t=0.5048 wrong_frac=0.4952 init_acc_corrupt=0.4711 init_gold_top10=0.4994 init_gold_top100=0.5049
+step=70300 micro_steps=140600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8651 acc_all=0.5852 loss_corrupt=2.8651 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8651 loss_recon=2.8651 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5023 init_acc_corrupt=0.4622 init_gold_top10=0.4920 init_gold_top100=0.4979
+step=70400 micro_steps=140800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8064 acc_all=0.5926 loss_corrupt=2.8064 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8064 loss_recon=2.8064 loss_meanflow=0.0000 mean_model_t=0.5044 mean_corrupt_t=0.5044 wrong_frac=0.4949 init_acc_corrupt=0.4708 init_gold_top10=0.4996 init_gold_top100=0.5052
+step=70500 micro_steps=141000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8596 acc_all=0.5857 loss_corrupt=2.8596 acc_corrupt=0.5857 corrupt_frac=1.0000 loss=2.8596 loss_recon=2.8596 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5027 init_acc_corrupt=0.4627 init_gold_top10=0.4917 init_gold_top100=0.4974
+step=70600 micro_steps=141200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8460 acc_all=0.5878 loss_corrupt=2.8460 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8460 loss_recon=2.8460 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4996 init_acc_corrupt=0.4656 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=70700 micro_steps=141400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8487 acc_all=0.5871 loss_corrupt=2.8487 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8487 loss_recon=2.8487 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5005 init_acc_corrupt=0.4646 init_gold_top10=0.4938 init_gold_top100=0.4996
+step=70800 micro_steps=141600 elapsed=33.7s lr=3.000000e-04 loss_all=2.7941 acc_all=0.5952 loss_corrupt=2.7941 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7941 loss_recon=2.7941 loss_meanflow=0.0000 mean_model_t=0.5066 mean_corrupt_t=0.5066 wrong_frac=0.4929 init_acc_corrupt=0.4733 init_gold_top10=0.5017 init_gold_top100=0.5073
+step=70900 micro_steps=141800 elapsed=33.6s lr=3.000000e-04 loss_all=2.8175 acc_all=0.5916 loss_corrupt=2.8175 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8175 loss_recon=2.8175 loss_meanflow=0.0000 mean_model_t=0.5055 mean_corrupt_t=0.5055 wrong_frac=0.4951 init_acc_corrupt=0.4707 init_gold_top10=0.4995 init_gold_top100=0.5050
+step=71000 micro_steps=142000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8398 acc_all=0.5890 loss_corrupt=2.8398 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8398 loss_recon=2.8398 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4993 init_acc_corrupt=0.4659 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=71100 micro_steps=142200 elapsed=35.2s lr=3.000000e-04 loss_all=2.8330 acc_all=0.5896 loss_corrupt=2.8330 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8330 loss_recon=2.8330 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4983 init_acc_corrupt=0.4672 init_gold_top10=0.4960 init_gold_top100=0.5018
+step=71200 micro_steps=142400 elapsed=33.6s lr=3.000000e-04 loss_all=2.8776 acc_all=0.5836 loss_corrupt=2.8776 acc_corrupt=0.5836 corrupt_frac=1.0000 loss=2.8776 loss_recon=2.8776 loss_meanflow=0.0000 mean_model_t=0.4960 mean_corrupt_t=0.4960 wrong_frac=0.5039 init_acc_corrupt=0.4604 init_gold_top10=0.4904 init_gold_top100=0.4963
+step=71300 micro_steps=142600 elapsed=33.6s lr=3.000000e-04 loss_all=2.8774 acc_all=0.5837 loss_corrupt=2.8774 acc_corrupt=0.5837 corrupt_frac=1.0000 loss=2.8774 loss_recon=2.8774 loss_meanflow=0.0000 mean_model_t=0.4957 mean_corrupt_t=0.4957 wrong_frac=0.5041 init_acc_corrupt=0.4613 init_gold_top10=0.4901 init_gold_top100=0.4959
+step=71400 micro_steps=142800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8463 acc_all=0.5873 loss_corrupt=2.8463 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8463 loss_recon=2.8463 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5014 init_acc_corrupt=0.4643 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=71500 micro_steps=143000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8497 acc_all=0.5874 loss_corrupt=2.8497 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8497 loss_recon=2.8497 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4995 init_acc_corrupt=0.4652 init_gold_top10=0.4948 init_gold_top100=0.5007
+step=71600 micro_steps=143200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8598 acc_all=0.5862 loss_corrupt=2.8598 acc_corrupt=0.5862 corrupt_frac=1.0000 loss=2.8598 loss_recon=2.8598 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4638 init_gold_top10=0.4928 init_gold_top100=0.4984
+step=71700 micro_steps=143400 elapsed=34.1s lr=3.000000e-04 loss_all=2.8025 acc_all=0.5932 loss_corrupt=2.8025 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.8025 loss_recon=2.8025 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4958 init_acc_corrupt=0.4703 init_gold_top10=0.4988 init_gold_top100=0.5045
+step=71800 micro_steps=143600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8485 acc_all=0.5872 loss_corrupt=2.8485 acc_corrupt=0.5872 corrupt_frac=1.0000 loss=2.8485 loss_recon=2.8485 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5008 init_acc_corrupt=0.4648 init_gold_top10=0.4936 init_gold_top100=0.4992
+step=71900 micro_steps=143800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8256 acc_all=0.5904 loss_corrupt=2.8256 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8256 loss_recon=2.8256 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4988 init_acc_corrupt=0.4668 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=72000 micro_steps=144000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8469 acc_all=0.5878 loss_corrupt=2.8469 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8469 loss_recon=2.8469 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5004 init_acc_corrupt=0.4646 init_gold_top10=0.4939 init_gold_top100=0.4998
+step=72100 micro_steps=144200 elapsed=35.2s lr=3.000000e-04 loss_all=2.8508 acc_all=0.5870 loss_corrupt=2.8508 acc_corrupt=0.5870 corrupt_frac=1.0000 loss=2.8508 loss_recon=2.8508 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5009 init_acc_corrupt=0.4640 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=72200 micro_steps=144400 elapsed=33.6s lr=3.000000e-04 loss_all=2.8357 acc_all=0.5885 loss_corrupt=2.8357 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8357 loss_recon=2.8357 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4997 init_acc_corrupt=0.4658 init_gold_top10=0.4948 init_gold_top100=0.5004
+step=72300 micro_steps=144600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8494 acc_all=0.5869 loss_corrupt=2.8494 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8494 loss_recon=2.8494 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5023 init_acc_corrupt=0.4631 init_gold_top10=0.4923 init_gold_top100=0.4979
+step=72400 micro_steps=144800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8264 acc_all=0.5904 loss_corrupt=2.8264 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8264 loss_recon=2.8264 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4976 init_acc_corrupt=0.4679 init_gold_top10=0.4968 init_gold_top100=0.5026
+step=72500 micro_steps=145000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8672 acc_all=0.5845 loss_corrupt=2.8672 acc_corrupt=0.5845 corrupt_frac=1.0000 loss=2.8672 loss_recon=2.8672 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5041 init_acc_corrupt=0.4607 init_gold_top10=0.4903 init_gold_top100=0.4961
+step=72600 micro_steps=145200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8183 acc_all=0.5908 loss_corrupt=2.8183 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8183 loss_recon=2.8183 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4986 init_acc_corrupt=0.4672 init_gold_top10=0.4960 init_gold_top100=0.5015
+step=72700 micro_steps=145400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8538 acc_all=0.5866 loss_corrupt=2.8538 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8538 loss_recon=2.8538 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5019 init_acc_corrupt=0.4635 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=72800 micro_steps=145600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8359 acc_all=0.5891 loss_corrupt=2.8359 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8359 loss_recon=2.8359 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4993 init_acc_corrupt=0.4664 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=72900 micro_steps=145800 elapsed=33.7s lr=3.000000e-04 loss_all=2.8574 acc_all=0.5861 loss_corrupt=2.8574 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8574 loss_recon=2.8574 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5026 init_acc_corrupt=0.4621 init_gold_top10=0.4919 init_gold_top100=0.4977
+step=73000 micro_steps=146000 elapsed=34.0s lr=3.000000e-04 loss_all=2.8283 acc_all=0.5901 loss_corrupt=2.8283 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8283 loss_recon=2.8283 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4977 init_acc_corrupt=0.4676 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=73100 micro_steps=146200 elapsed=35.2s lr=3.000000e-04 loss_all=2.8229 acc_all=0.5908 loss_corrupt=2.8229 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8229 loss_recon=2.8229 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4967 init_acc_corrupt=0.4687 init_gold_top10=0.4976 init_gold_top100=0.5034
+step=73200 micro_steps=146400 elapsed=33.6s lr=3.000000e-04 loss_all=2.8148 acc_all=0.5917 loss_corrupt=2.8148 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8148 loss_recon=2.8148 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4967 init_acc_corrupt=0.4694 init_gold_top10=0.4978 init_gold_top100=0.5034
+step=73300 micro_steps=146600 elapsed=33.7s lr=3.000000e-04 loss_all=2.8192 acc_all=0.5907 loss_corrupt=2.8192 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8192 loss_recon=2.8192 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4977 init_acc_corrupt=0.4683 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=73400 micro_steps=146800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8296 acc_all=0.5901 loss_corrupt=2.8296 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8296 loss_recon=2.8296 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4977 init_acc_corrupt=0.4679 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=73500 micro_steps=147000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8713 acc_all=0.5841 loss_corrupt=2.8713 acc_corrupt=0.5841 corrupt_frac=1.0000 loss=2.8713 loss_recon=2.8713 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5031 init_acc_corrupt=0.4623 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=73600 micro_steps=147200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8367 acc_all=0.5891 loss_corrupt=2.8367 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8367 loss_recon=2.8367 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4988 init_acc_corrupt=0.4669 init_gold_top10=0.4956 init_gold_top100=0.5012
+step=73700 micro_steps=147400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8261 acc_all=0.5903 loss_corrupt=2.8261 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8261 loss_recon=2.8261 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4977 init_acc_corrupt=0.4677 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=73800 micro_steps=147600 elapsed=33.8s lr=3.000000e-04 loss_all=2.8241 acc_all=0.5906 loss_corrupt=2.8241 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8241 loss_recon=2.8241 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4982 init_acc_corrupt=0.4675 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=73900 micro_steps=147800 elapsed=33.9s lr=3.000000e-04 loss_all=2.8308 acc_all=0.5894 loss_corrupt=2.8308 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8308 loss_recon=2.8308 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.4997 init_acc_corrupt=0.4651 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=74000 micro_steps=148000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8454 acc_all=0.5878 loss_corrupt=2.8454 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8454 loss_recon=2.8454 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5003 init_acc_corrupt=0.4654 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=74100 micro_steps=148200 elapsed=51.3s lr=3.000000e-04 loss_all=2.8378 acc_all=0.5890 loss_corrupt=2.8378 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8378 loss_recon=2.8378 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4985 init_acc_corrupt=0.4672 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=74200 micro_steps=148400 elapsed=85.8s lr=3.000000e-04 loss_all=2.8425 acc_all=0.5879 loss_corrupt=2.8425 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8425 loss_recon=2.8425 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5018 init_acc_corrupt=0.4635 init_gold_top10=0.4928 init_gold_top100=0.4984
+step=74300 micro_steps=148600 elapsed=41.1s lr=3.000000e-04 loss_all=2.8365 acc_all=0.5887 loss_corrupt=2.8365 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8365 loss_recon=2.8365 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5003 init_acc_corrupt=0.4654 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=74400 micro_steps=148800 elapsed=95.7s lr=3.000000e-04 loss_all=2.8356 acc_all=0.5889 loss_corrupt=2.8356 acc_corrupt=0.5889 corrupt_frac=1.0000 loss=2.8356 loss_recon=2.8356 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4989 init_acc_corrupt=0.4668 init_gold_top10=0.4957 init_gold_top100=0.5012
+step=74500 micro_steps=149000 elapsed=97.0s lr=3.000000e-04 loss_all=2.8443 acc_all=0.5879 loss_corrupt=2.8443 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8443 loss_recon=2.8443 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5008 init_acc_corrupt=0.4638 init_gold_top10=0.4937 init_gold_top100=0.4995
+step=74600 micro_steps=149200 elapsed=89.8s lr=3.000000e-04 loss_all=2.8077 acc_all=0.5928 loss_corrupt=2.8077 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8077 loss_recon=2.8077 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5045 wrong_frac=0.4961 init_acc_corrupt=0.4704 init_gold_top10=0.4986 init_gold_top100=0.5041
+step=74700 micro_steps=149400 elapsed=34.9s lr=3.000000e-04 loss_all=2.8017 acc_all=0.5936 loss_corrupt=2.8017 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.8017 loss_recon=2.8017 loss_meanflow=0.0000 mean_model_t=0.5042 mean_corrupt_t=0.5042 wrong_frac=0.4957 init_acc_corrupt=0.4707 init_gold_top10=0.4989 init_gold_top100=0.5045
+step=74800 micro_steps=149600 elapsed=51.8s lr=3.000000e-04 loss_all=2.8644 acc_all=0.5856 loss_corrupt=2.8644 acc_corrupt=0.5856 corrupt_frac=1.0000 loss=2.8644 loss_recon=2.8644 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5018 init_acc_corrupt=0.4630 init_gold_top10=0.4924 init_gold_top100=0.4984
+step=74900 micro_steps=149800 elapsed=51.5s lr=3.000000e-04 loss_all=2.8565 acc_all=0.5861 loss_corrupt=2.8565 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8565 loss_recon=2.8565 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5019 init_acc_corrupt=0.4632 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=75000 micro_steps=150000 elapsed=51.7s lr=3.000000e-04 loss_all=2.8570 acc_all=0.5859 loss_corrupt=2.8570 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8570 loss_recon=2.8570 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5025 init_acc_corrupt=0.4626 init_gold_top10=0.4918 init_gold_top100=0.4976
+step=75100 micro_steps=150200 elapsed=53.6s lr=3.000000e-04 loss_all=2.8170 acc_all=0.5913 loss_corrupt=2.8170 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8170 loss_recon=2.8170 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4986 init_acc_corrupt=0.4671 init_gold_top10=0.4960 init_gold_top100=0.5015
+step=75200 micro_steps=150400 elapsed=52.2s lr=3.000000e-04 loss_all=2.8414 acc_all=0.5879 loss_corrupt=2.8414 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8414 loss_recon=2.8414 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5022 init_acc_corrupt=0.4638 init_gold_top10=0.4922 init_gold_top100=0.4979
+step=75300 micro_steps=150600 elapsed=72.9s lr=3.000000e-04 loss_all=2.8122 acc_all=0.5923 loss_corrupt=2.8122 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8122 loss_recon=2.8122 loss_meanflow=0.0000 mean_model_t=0.5038 mean_corrupt_t=0.5038 wrong_frac=0.4959 init_acc_corrupt=0.4700 init_gold_top10=0.4986 init_gold_top100=0.5042
+step=75400 micro_steps=150800 elapsed=53.3s lr=3.000000e-04 loss_all=2.8143 acc_all=0.5919 loss_corrupt=2.8143 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8143 loss_recon=2.8143 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4971 init_acc_corrupt=0.4691 init_gold_top10=0.4974 init_gold_top100=0.5030
+step=75500 micro_steps=151000 elapsed=33.9s lr=3.000000e-04 loss_all=2.8520 acc_all=0.5867 loss_corrupt=2.8520 acc_corrupt=0.5867 corrupt_frac=1.0000 loss=2.8520 loss_recon=2.8520 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5021 init_acc_corrupt=0.4635 init_gold_top10=0.4925 init_gold_top100=0.4981
+step=75600 micro_steps=151200 elapsed=33.7s lr=3.000000e-04 loss_all=2.8140 acc_all=0.5916 loss_corrupt=2.8140 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8140 loss_recon=2.8140 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4976 init_acc_corrupt=0.4685 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=75700 micro_steps=151400 elapsed=33.7s lr=3.000000e-04 loss_all=2.8355 acc_all=0.5884 loss_corrupt=2.8355 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8355 loss_recon=2.8355 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5008 init_acc_corrupt=0.4644 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=75800 micro_steps=151600 elapsed=33.6s lr=3.000000e-04 loss_all=2.8294 acc_all=0.5894 loss_corrupt=2.8294 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8294 loss_recon=2.8294 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4996 init_acc_corrupt=0.4665 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=75900 micro_steps=151800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8197 acc_all=0.5909 loss_corrupt=2.8197 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8197 loss_recon=2.8197 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4980 init_acc_corrupt=0.4673 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=76000 micro_steps=152000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8110 acc_all=0.5918 loss_corrupt=2.8110 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8110 loss_recon=2.8110 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4976 init_acc_corrupt=0.4681 init_gold_top10=0.4970 init_gold_top100=0.5026
+step=76100 micro_steps=152200 elapsed=35.3s lr=3.000000e-04 loss_all=2.8417 acc_all=0.5883 loss_corrupt=2.8417 acc_corrupt=0.5883 corrupt_frac=1.0000 loss=2.8417 loss_recon=2.8417 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5006 init_acc_corrupt=0.4647 init_gold_top10=0.4938 init_gold_top100=0.4996
+step=76200 micro_steps=152400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8316 acc_all=0.5891 loss_corrupt=2.8316 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8316 loss_recon=2.8316 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5005 init_acc_corrupt=0.4657 init_gold_top10=0.4941 init_gold_top100=0.4997
+step=76300 micro_steps=152600 elapsed=34.0s lr=3.000000e-04 loss_all=2.8416 acc_all=0.5879 loss_corrupt=2.8416 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8416 loss_recon=2.8416 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5010 init_acc_corrupt=0.4646 init_gold_top10=0.4933 init_gold_top100=0.4991
+step=76400 micro_steps=152800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8359 acc_all=0.5891 loss_corrupt=2.8359 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8359 loss_recon=2.8359 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4992 init_acc_corrupt=0.4660 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=76500 micro_steps=153000 elapsed=33.7s lr=3.000000e-04 loss_all=2.8127 acc_all=0.5918 loss_corrupt=2.8127 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8127 loss_recon=2.8127 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4976 init_acc_corrupt=0.4678 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=76600 micro_steps=153200 elapsed=33.8s lr=3.000000e-04 loss_all=2.8415 acc_all=0.5880 loss_corrupt=2.8415 acc_corrupt=0.5880 corrupt_frac=1.0000 loss=2.8415 loss_recon=2.8415 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5010 init_acc_corrupt=0.4648 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=76700 micro_steps=153400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8072 acc_all=0.5926 loss_corrupt=2.8072 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8072 loss_recon=2.8072 loss_meanflow=0.0000 mean_model_t=0.5044 mean_corrupt_t=0.5044 wrong_frac=0.4951 init_acc_corrupt=0.4706 init_gold_top10=0.4994 init_gold_top100=0.5051
+step=76800 micro_steps=153600 elapsed=33.9s lr=3.000000e-04 loss_all=2.8212 acc_all=0.5908 loss_corrupt=2.8212 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8212 loss_recon=2.8212 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4980 init_acc_corrupt=0.4678 init_gold_top10=0.4964 init_gold_top100=0.5022
+step=76900 micro_steps=153800 elapsed=34.2s lr=3.000000e-04 loss_all=2.8479 acc_all=0.5872 loss_corrupt=2.8479 acc_corrupt=0.5872 corrupt_frac=1.0000 loss=2.8479 loss_recon=2.8479 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5017 init_acc_corrupt=0.4635 init_gold_top10=0.4926 init_gold_top100=0.4984
+step=77000 micro_steps=154000 elapsed=34.3s lr=3.000000e-04 loss_all=2.8452 acc_all=0.5878 loss_corrupt=2.8452 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8452 loss_recon=2.8452 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5011 init_acc_corrupt=0.4640 init_gold_top10=0.4931 init_gold_top100=0.4990
+step=77100 micro_steps=154200 elapsed=35.7s lr=3.000000e-04 loss_all=2.8514 acc_all=0.5873 loss_corrupt=2.8514 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8514 loss_recon=2.8514 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5010 init_acc_corrupt=0.4645 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=77200 micro_steps=154400 elapsed=34.2s lr=3.000000e-04 loss_all=2.8131 acc_all=0.5917 loss_corrupt=2.8131 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8131 loss_recon=2.8131 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4973 init_acc_corrupt=0.4687 init_gold_top10=0.4970 init_gold_top100=0.5028
+step=77300 micro_steps=154600 elapsed=34.2s lr=3.000000e-04 loss_all=2.8600 acc_all=0.5859 loss_corrupt=2.8600 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8600 loss_recon=2.8600 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5024 init_acc_corrupt=0.4623 init_gold_top10=0.4918 init_gold_top100=0.4976
+step=77400 micro_steps=154800 elapsed=34.4s lr=3.000000e-04 loss_all=2.8508 acc_all=0.5873 loss_corrupt=2.8508 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8508 loss_recon=2.8508 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5017 init_acc_corrupt=0.4638 init_gold_top10=0.4927 init_gold_top100=0.4984
+step=77500 micro_steps=155000 elapsed=34.2s lr=3.000000e-04 loss_all=2.8108 acc_all=0.5920 loss_corrupt=2.8108 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8108 loss_recon=2.8108 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4971 init_acc_corrupt=0.4686 init_gold_top10=0.4975 init_gold_top100=0.5031
+step=77600 micro_steps=155200 elapsed=33.9s lr=3.000000e-04 loss_all=2.8222 acc_all=0.5907 loss_corrupt=2.8222 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8222 loss_recon=2.8222 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4983 init_acc_corrupt=0.4676 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=77700 micro_steps=155400 elapsed=33.8s lr=3.000000e-04 loss_all=2.8086 acc_all=0.5923 loss_corrupt=2.8086 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8086 loss_recon=2.8086 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4967 init_acc_corrupt=0.4695 init_gold_top10=0.4977 init_gold_top100=0.5035
+step=77800 micro_steps=155600 elapsed=34.1s lr=3.000000e-04 loss_all=2.8271 acc_all=0.5904 loss_corrupt=2.8271 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8271 loss_recon=2.8271 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4980 init_acc_corrupt=0.4670 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=77900 micro_steps=155800 elapsed=33.8s lr=3.000000e-04 loss_all=2.8218 acc_all=0.5909 loss_corrupt=2.8218 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8218 loss_recon=2.8218 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4980 init_acc_corrupt=0.4676 init_gold_top10=0.4963 init_gold_top100=0.5021
+step=78000 micro_steps=156000 elapsed=33.8s lr=3.000000e-04 loss_all=2.8241 acc_all=0.5904 loss_corrupt=2.8241 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8241 loss_recon=2.8241 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4984 init_acc_corrupt=0.4667 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=78100 micro_steps=156200 elapsed=35.5s lr=3.000000e-04 loss_all=2.8365 acc_all=0.5887 loss_corrupt=2.8365 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8365 loss_recon=2.8365 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5016 init_acc_corrupt=0.4645 init_gold_top10=0.4929 init_gold_top100=0.4985
+step=78200 micro_steps=156400 elapsed=34.0s lr=3.000000e-04 loss_all=2.8451 acc_all=0.5877 loss_corrupt=2.8451 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8451 loss_recon=2.8451 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5017 init_acc_corrupt=0.4638 init_gold_top10=0.4926 init_gold_top100=0.4984
+step=78300 micro_steps=156600 elapsed=39.6s lr=3.000000e-04 loss_all=2.8355 acc_all=0.5892 loss_corrupt=2.8355 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8355 loss_recon=2.8355 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4992 init_acc_corrupt=0.4663 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=78400 micro_steps=156800 elapsed=45.0s lr=3.000000e-04 loss_all=2.8701 acc_all=0.5844 loss_corrupt=2.8701 acc_corrupt=0.5844 corrupt_frac=1.0000 loss=2.8701 loss_recon=2.8701 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5035 init_acc_corrupt=0.4616 init_gold_top10=0.4907 init_gold_top100=0.4966
+step=78500 micro_steps=157000 elapsed=51.1s lr=3.000000e-04 loss_all=2.8573 acc_all=0.5863 loss_corrupt=2.8573 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8573 loss_recon=2.8573 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5023 init_acc_corrupt=0.4629 init_gold_top10=0.4920 init_gold_top100=0.4978
+step=78600 micro_steps=157200 elapsed=49.6s lr=3.000000e-04 loss_all=2.8095 acc_all=0.5924 loss_corrupt=2.8095 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8095 loss_recon=2.8095 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4965 init_acc_corrupt=0.4692 init_gold_top10=0.4981 init_gold_top100=0.5037
+step=78700 micro_steps=157400 elapsed=49.4s lr=3.000000e-04 loss_all=2.8176 acc_all=0.5909 loss_corrupt=2.8176 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8176 loss_recon=2.8176 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4983 init_acc_corrupt=0.4678 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=78800 micro_steps=157600 elapsed=49.8s lr=3.000000e-04 loss_all=2.8467 acc_all=0.5877 loss_corrupt=2.8467 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8467 loss_recon=2.8467 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5013 init_acc_corrupt=0.4641 init_gold_top10=0.4931 init_gold_top100=0.4988
+step=78900 micro_steps=157800 elapsed=49.6s lr=3.000000e-04 loss_all=2.8471 acc_all=0.5877 loss_corrupt=2.8471 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8471 loss_recon=2.8471 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5010 init_acc_corrupt=0.4641 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=79000 micro_steps=158000 elapsed=61.7s lr=3.000000e-04 loss_all=2.8401 acc_all=0.5882 loss_corrupt=2.8401 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8401 loss_recon=2.8401 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5001 init_acc_corrupt=0.4652 init_gold_top10=0.4945 init_gold_top100=0.5001
+step=79100 micro_steps=158200 elapsed=53.5s lr=3.000000e-04 loss_all=2.8595 acc_all=0.5856 loss_corrupt=2.8595 acc_corrupt=0.5856 corrupt_frac=1.0000 loss=2.8595 loss_recon=2.8595 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5027 init_acc_corrupt=0.4624 init_gold_top10=0.4917 init_gold_top100=0.4975
+step=79200 micro_steps=158400 elapsed=52.1s lr=3.000000e-04 loss_all=2.8897 acc_all=0.5818 loss_corrupt=2.8897 acc_corrupt=0.5818 corrupt_frac=1.0000 loss=2.8897 loss_recon=2.8897 loss_meanflow=0.0000 mean_model_t=0.4929 mean_corrupt_t=0.4929 wrong_frac=0.5070 init_acc_corrupt=0.4577 init_gold_top10=0.4872 init_gold_top100=0.4930
+step=79300 micro_steps=158600 elapsed=52.2s lr=3.000000e-04 loss_all=2.8386 acc_all=0.5881 loss_corrupt=2.8386 acc_corrupt=0.5881 corrupt_frac=1.0000 loss=2.8386 loss_recon=2.8386 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5010 init_acc_corrupt=0.4645 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=79400 micro_steps=158800 elapsed=51.6s lr=3.000000e-04 loss_all=2.8385 acc_all=0.5887 loss_corrupt=2.8385 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8385 loss_recon=2.8385 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5003 init_acc_corrupt=0.4656 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=79500 micro_steps=159000 elapsed=51.5s lr=3.000000e-04 loss_all=2.8219 acc_all=0.5903 loss_corrupt=2.8219 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8219 loss_recon=2.8219 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4987 init_acc_corrupt=0.4668 init_gold_top10=0.4957 init_gold_top100=0.5016
+step=79600 micro_steps=159200 elapsed=51.4s lr=3.000000e-04 loss_all=2.8346 acc_all=0.5889 loss_corrupt=2.8346 acc_corrupt=0.5889 corrupt_frac=1.0000 loss=2.8346 loss_recon=2.8346 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5011 init_acc_corrupt=0.4643 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=79700 micro_steps=159400 elapsed=51.4s lr=3.000000e-04 loss_all=2.8268 acc_all=0.5904 loss_corrupt=2.8268 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8268 loss_recon=2.8268 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4988 init_acc_corrupt=0.4668 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=79800 micro_steps=159600 elapsed=51.7s lr=3.000000e-04 loss_all=2.8591 acc_all=0.5864 loss_corrupt=2.8591 acc_corrupt=0.5864 corrupt_frac=1.0000 loss=2.8591 loss_recon=2.8591 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5017 init_acc_corrupt=0.4635 init_gold_top10=0.4926 init_gold_top100=0.4984
+step=79900 micro_steps=159800 elapsed=52.0s lr=3.000000e-04 loss_all=2.8066 acc_all=0.5924 loss_corrupt=2.8066 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8066 loss_recon=2.8066 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4979 init_acc_corrupt=0.4677 init_gold_top10=0.4968 init_gold_top100=0.5024
+step=80000 micro_steps=160000 elapsed=52.0s lr=3.000000e-04 loss_all=2.8230 acc_all=0.5902 loss_corrupt=2.8230 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8230 loss_recon=2.8230 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4992 init_acc_corrupt=0.4665 init_gold_top10=0.4954 init_gold_top100=0.5010
+step=80100 micro_steps=160200 elapsed=54.6s lr=3.000000e-04 loss_all=2.8402 acc_all=0.5883 loss_corrupt=2.8402 acc_corrupt=0.5883 corrupt_frac=1.0000 loss=2.8402 loss_recon=2.8402 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4636 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=80200 micro_steps=160400 elapsed=51.3s lr=3.000000e-04 loss_all=2.8538 acc_all=0.5867 loss_corrupt=2.8538 acc_corrupt=0.5867 corrupt_frac=1.0000 loss=2.8538 loss_recon=2.8538 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5021 init_acc_corrupt=0.4631 init_gold_top10=0.4924 init_gold_top100=0.4980
+step=80300 micro_steps=160600 elapsed=51.9s lr=3.000000e-04 loss_all=2.8409 acc_all=0.5884 loss_corrupt=2.8409 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8409 loss_recon=2.8409 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5007 init_acc_corrupt=0.4645 init_gold_top10=0.4938 init_gold_top100=0.4995
+step=80400 micro_steps=160800 elapsed=51.5s lr=3.000000e-04 loss_all=2.8402 acc_all=0.5879 loss_corrupt=2.8402 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8402 loss_recon=2.8402 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5002 init_acc_corrupt=0.4652 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=80500 micro_steps=161000 elapsed=51.4s lr=3.000000e-04 loss_all=2.8213 acc_all=0.5909 loss_corrupt=2.8213 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8213 loss_recon=2.8213 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4986 init_acc_corrupt=0.4671 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=80600 micro_steps=161200 elapsed=50.9s lr=3.000000e-04 loss_all=2.8423 acc_all=0.5881 loss_corrupt=2.8423 acc_corrupt=0.5881 corrupt_frac=1.0000 loss=2.8423 loss_recon=2.8423 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5002 init_acc_corrupt=0.4651 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=80700 micro_steps=161400 elapsed=50.7s lr=3.000000e-04 loss_all=2.8063 acc_all=0.5935 loss_corrupt=2.8063 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.8063 loss_recon=2.8063 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5045 wrong_frac=0.4957 init_acc_corrupt=0.4699 init_gold_top10=0.4988 init_gold_top100=0.5044
+step=80800 micro_steps=161600 elapsed=50.7s lr=3.000000e-04 loss_all=2.8161 acc_all=0.5916 loss_corrupt=2.8161 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8161 loss_recon=2.8161 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4981 init_acc_corrupt=0.4678 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=80900 micro_steps=161800 elapsed=85.3s lr=3.000000e-04 loss_all=2.8320 acc_all=0.5891 loss_corrupt=2.8320 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8320 loss_recon=2.8320 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5000 init_acc_corrupt=0.4650 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=81000 micro_steps=162000 elapsed=95.2s lr=3.000000e-04 loss_all=2.8340 acc_all=0.5890 loss_corrupt=2.8340 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8340 loss_recon=2.8340 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5002 init_acc_corrupt=0.4649 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=81100 micro_steps=162200 elapsed=92.4s lr=3.000000e-04 loss_all=2.8568 acc_all=0.5859 loss_corrupt=2.8568 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8568 loss_recon=2.8568 loss_meanflow=0.0000 mean_model_t=0.4959 mean_corrupt_t=0.4959 wrong_frac=0.5039 init_acc_corrupt=0.4616 init_gold_top10=0.4905 init_gold_top100=0.4962
+step=81200 micro_steps=162400 elapsed=84.6s lr=3.000000e-04 loss_all=2.8186 acc_all=0.5911 loss_corrupt=2.8186 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8186 loss_recon=2.8186 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4995 init_acc_corrupt=0.4662 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=81300 micro_steps=162600 elapsed=68.6s lr=3.000000e-04 loss_all=2.8064 acc_all=0.5933 loss_corrupt=2.8064 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.8064 loss_recon=2.8064 loss_meanflow=0.0000 mean_model_t=0.5044 mean_corrupt_t=0.5044 wrong_frac=0.4954 init_acc_corrupt=0.4703 init_gold_top10=0.4991 init_gold_top100=0.5048
+step=81400 micro_steps=162800 elapsed=92.1s lr=3.000000e-04 loss_all=2.8487 acc_all=0.5864 loss_corrupt=2.8487 acc_corrupt=0.5864 corrupt_frac=1.0000 loss=2.8487 loss_recon=2.8487 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5035 init_acc_corrupt=0.4616 init_gold_top10=0.4909 init_gold_top100=0.4967
+step=81500 micro_steps=163000 elapsed=91.7s lr=3.000000e-04 loss_all=2.8880 acc_all=0.5819 loss_corrupt=2.8880 acc_corrupt=0.5819 corrupt_frac=1.0000 loss=2.8880 loss_recon=2.8880 loss_meanflow=0.0000 mean_model_t=0.4927 mean_corrupt_t=0.4927 wrong_frac=0.5070 init_acc_corrupt=0.4576 init_gold_top10=0.4872 init_gold_top100=0.4930
+step=81600 micro_steps=163200 elapsed=85.8s lr=3.000000e-04 loss_all=2.8370 acc_all=0.5885 loss_corrupt=2.8370 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8370 loss_recon=2.8370 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5004 init_acc_corrupt=0.4652 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=81700 micro_steps=163400 elapsed=96.5s lr=3.000000e-04 loss_all=2.8682 acc_all=0.5849 loss_corrupt=2.8682 acc_corrupt=0.5849 corrupt_frac=1.0000 loss=2.8682 loss_recon=2.8682 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5038 init_acc_corrupt=0.4612 init_gold_top10=0.4905 init_gold_top100=0.4963
+step=81800 micro_steps=163600 elapsed=75.9s lr=3.000000e-04 loss_all=2.8174 acc_all=0.5911 loss_corrupt=2.8174 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8174 loss_recon=2.8174 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4984 init_acc_corrupt=0.4667 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=81900 micro_steps=163800 elapsed=57.6s lr=3.000000e-04 loss_all=2.8568 acc_all=0.5859 loss_corrupt=2.8568 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8568 loss_recon=2.8568 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5032 init_acc_corrupt=0.4622 init_gold_top10=0.4911 init_gold_top100=0.4969
+step=82000 micro_steps=164000 elapsed=57.2s lr=3.000000e-04 loss_all=2.8464 acc_all=0.5876 loss_corrupt=2.8464 acc_corrupt=0.5876 corrupt_frac=1.0000 loss=2.8464 loss_recon=2.8464 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=82100 micro_steps=164200 elapsed=99.6s lr=3.000000e-04 loss_all=2.8220 acc_all=0.5909 loss_corrupt=2.8220 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8220 loss_recon=2.8220 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4986 init_acc_corrupt=0.4666 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=82200 micro_steps=164400 elapsed=86.1s lr=3.000000e-04 loss_all=2.8083 acc_all=0.5921 loss_corrupt=2.8083 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8083 loss_recon=2.8083 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4975 init_acc_corrupt=0.4682 init_gold_top10=0.4968 init_gold_top100=0.5027
+step=82300 micro_steps=164600 elapsed=91.6s lr=3.000000e-04 loss_all=2.8387 acc_all=0.5885 loss_corrupt=2.8387 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8387 loss_recon=2.8387 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5014 init_acc_corrupt=0.4637 init_gold_top10=0.4930 init_gold_top100=0.4987
+step=82400 micro_steps=164800 elapsed=57.9s lr=3.000000e-04 loss_all=2.8130 acc_all=0.5922 loss_corrupt=2.8130 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8130 loss_recon=2.8130 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4973 init_acc_corrupt=0.4688 init_gold_top10=0.4972 init_gold_top100=0.5028
+step=82500 micro_steps=165000 elapsed=57.0s lr=3.000000e-04 loss_all=2.8353 acc_all=0.5894 loss_corrupt=2.8353 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8353 loss_recon=2.8353 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4992 init_acc_corrupt=0.4655 init_gold_top10=0.4950 init_gold_top100=0.5009
+step=82600 micro_steps=165200 elapsed=57.1s lr=3.000000e-04 loss_all=2.8530 acc_all=0.5869 loss_corrupt=2.8530 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8530 loss_recon=2.8530 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5021 init_acc_corrupt=0.4628 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=82700 micro_steps=165400 elapsed=57.5s lr=3.000000e-04 loss_all=2.8467 acc_all=0.5874 loss_corrupt=2.8467 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8467 loss_recon=2.8467 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5019 init_acc_corrupt=0.4635 init_gold_top10=0.4923 init_gold_top100=0.4982
+step=82800 micro_steps=165600 elapsed=57.8s lr=3.000000e-04 loss_all=2.8410 acc_all=0.5882 loss_corrupt=2.8410 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8410 loss_recon=2.8410 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5009 init_acc_corrupt=0.4642 init_gold_top10=0.4935 init_gold_top100=0.4991
+step=82900 micro_steps=165800 elapsed=57.9s lr=3.000000e-04 loss_all=2.8390 acc_all=0.5885 loss_corrupt=2.8390 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8390 loss_recon=2.8390 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5011 init_acc_corrupt=0.4647 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=83000 micro_steps=166000 elapsed=57.9s lr=3.000000e-04 loss_all=2.8515 acc_all=0.5869 loss_corrupt=2.8515 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8515 loss_recon=2.8515 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5023 init_acc_corrupt=0.4625 init_gold_top10=0.4920 init_gold_top100=0.4979
+step=83100 micro_steps=166200 elapsed=59.4s lr=3.000000e-04 loss_all=2.8569 acc_all=0.5861 loss_corrupt=2.8569 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8569 loss_recon=2.8569 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5026 init_acc_corrupt=0.4623 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=83200 micro_steps=166400 elapsed=57.3s lr=3.000000e-04 loss_all=2.8215 acc_all=0.5903 loss_corrupt=2.8215 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8215 loss_recon=2.8215 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5000 init_acc_corrupt=0.4657 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=83300 micro_steps=166600 elapsed=59.1s lr=3.000000e-04 loss_all=2.8131 acc_all=0.5921 loss_corrupt=2.8131 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8131 loss_recon=2.8131 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4974 init_acc_corrupt=0.4687 init_gold_top10=0.4971 init_gold_top100=0.5027
+step=83400 micro_steps=166800 elapsed=57.4s lr=3.000000e-04 loss_all=2.7976 acc_all=0.5943 loss_corrupt=2.7976 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7976 loss_recon=2.7976 loss_meanflow=0.0000 mean_model_t=0.5056 mean_corrupt_t=0.5056 wrong_frac=0.4944 init_acc_corrupt=0.4713 init_gold_top10=0.5000 init_gold_top100=0.5059
+step=83500 micro_steps=167000 elapsed=57.7s lr=3.000000e-04 loss_all=2.8494 acc_all=0.5873 loss_corrupt=2.8494 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8494 loss_recon=2.8494 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5021 init_acc_corrupt=0.4631 init_gold_top10=0.4922 init_gold_top100=0.4980
+step=83600 micro_steps=167200 elapsed=57.6s lr=3.000000e-04 loss_all=2.8602 acc_all=0.5857 loss_corrupt=2.8602 acc_corrupt=0.5857 corrupt_frac=1.0000 loss=2.8602 loss_recon=2.8602 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5028 init_acc_corrupt=0.4623 init_gold_top10=0.4915 init_gold_top100=0.4974
+step=83700 micro_steps=167400 elapsed=57.8s lr=3.000000e-04 loss_all=2.8226 acc_all=0.5904 loss_corrupt=2.8226 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8226 loss_recon=2.8226 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4984 init_acc_corrupt=0.4672 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=83800 micro_steps=167600 elapsed=57.4s lr=3.000000e-04 loss_all=2.8646 acc_all=0.5853 loss_corrupt=2.8646 acc_corrupt=0.5853 corrupt_frac=1.0000 loss=2.8646 loss_recon=2.8646 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5035 init_acc_corrupt=0.4614 init_gold_top10=0.4908 init_gold_top100=0.4967
+step=83900 micro_steps=167800 elapsed=57.4s lr=3.000000e-04 loss_all=2.7949 acc_all=0.5938 loss_corrupt=2.7949 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7949 loss_recon=2.7949 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4967 init_acc_corrupt=0.4699 init_gold_top10=0.4978 init_gold_top100=0.5034
+step=84000 micro_steps=168000 elapsed=57.1s lr=3.000000e-04 loss_all=2.8468 acc_all=0.5874 loss_corrupt=2.8468 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8468 loss_recon=2.8468 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5015 init_acc_corrupt=0.4639 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=84100 micro_steps=168200 elapsed=58.8s lr=3.000000e-04 loss_all=2.8496 acc_all=0.5868 loss_corrupt=2.8496 acc_corrupt=0.5868 corrupt_frac=1.0000 loss=2.8496 loss_recon=2.8496 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5021 init_acc_corrupt=0.4632 init_gold_top10=0.4923 init_gold_top100=0.4979
+step=84200 micro_steps=168400 elapsed=57.6s lr=3.000000e-04 loss_all=2.8356 acc_all=0.5890 loss_corrupt=2.8356 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8356 loss_recon=2.8356 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5004 init_acc_corrupt=0.4650 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=84300 micro_steps=168600 elapsed=57.8s lr=3.000000e-04 loss_all=2.8529 acc_all=0.5868 loss_corrupt=2.8529 acc_corrupt=0.5868 corrupt_frac=1.0000 loss=2.8529 loss_recon=2.8529 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5026 init_acc_corrupt=0.4627 init_gold_top10=0.4916 init_gold_top100=0.4974
+step=84400 micro_steps=168800 elapsed=58.0s lr=3.000000e-04 loss_all=2.8127 acc_all=0.5916 loss_corrupt=2.8127 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8127 loss_recon=2.8127 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4976 init_acc_corrupt=0.4681 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=84500 micro_steps=169000 elapsed=57.9s lr=3.000000e-04 loss_all=2.8447 acc_all=0.5873 loss_corrupt=2.8447 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8447 loss_recon=2.8447 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5017 init_acc_corrupt=0.4636 init_gold_top10=0.4927 init_gold_top100=0.4984
+step=84600 micro_steps=169200 elapsed=58.0s lr=3.000000e-04 loss_all=2.8522 acc_all=0.5865 loss_corrupt=2.8522 acc_corrupt=0.5865 corrupt_frac=1.0000 loss=2.8522 loss_recon=2.8522 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5033 init_acc_corrupt=0.4616 init_gold_top10=0.4912 init_gold_top100=0.4969
+step=84700 micro_steps=169400 elapsed=58.1s lr=3.000000e-04 loss_all=2.8272 acc_all=0.5902 loss_corrupt=2.8272 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8272 loss_recon=2.8272 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4991 init_acc_corrupt=0.4666 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=84800 micro_steps=169600 elapsed=58.2s lr=3.000000e-04 loss_all=2.8214 acc_all=0.5904 loss_corrupt=2.8214 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8214 loss_recon=2.8214 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4980 init_acc_corrupt=0.4675 init_gold_top10=0.4965 init_gold_top100=0.5021
+step=84900 micro_steps=169800 elapsed=58.2s lr=3.000000e-04 loss_all=2.8366 acc_all=0.5884 loss_corrupt=2.8366 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8366 loss_recon=2.8366 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5012 init_acc_corrupt=0.4643 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=85000 micro_steps=170000 elapsed=57.7s lr=3.000000e-04 loss_all=2.8744 acc_all=0.5839 loss_corrupt=2.8744 acc_corrupt=0.5839 corrupt_frac=1.0000 loss=2.8744 loss_recon=2.8744 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5048 init_acc_corrupt=0.4596 init_gold_top10=0.4894 init_gold_top100=0.4952
+step=85100 micro_steps=170200 elapsed=59.7s lr=3.000000e-04 loss_all=2.8132 acc_all=0.5919 loss_corrupt=2.8132 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8132 loss_recon=2.8132 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4979 init_acc_corrupt=0.4681 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=85200 micro_steps=170400 elapsed=58.2s lr=3.000000e-04 loss_all=2.8443 acc_all=0.5873 loss_corrupt=2.8443 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8443 loss_recon=2.8443 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5026 init_acc_corrupt=0.4633 init_gold_top10=0.4917 init_gold_top100=0.4976
+step=85300 micro_steps=170600 elapsed=58.1s lr=3.000000e-04 loss_all=2.8419 acc_all=0.5874 loss_corrupt=2.8419 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8419 loss_recon=2.8419 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5024 init_acc_corrupt=0.4629 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=85400 micro_steps=170800 elapsed=58.1s lr=3.000000e-04 loss_all=2.8134 acc_all=0.5922 loss_corrupt=2.8134 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8134 loss_recon=2.8134 loss_meanflow=0.0000 mean_model_t=0.5039 mean_corrupt_t=0.5039 wrong_frac=0.4961 init_acc_corrupt=0.4693 init_gold_top10=0.4982 init_gold_top100=0.5040
+step=85500 micro_steps=171000 elapsed=73.1s lr=3.000000e-04 loss_all=2.8383 acc_all=0.5887 loss_corrupt=2.8383 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8383 loss_recon=2.8383 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4999 init_acc_corrupt=0.4653 init_gold_top10=0.4944 init_gold_top100=0.5003
+step=85600 micro_steps=171200 elapsed=83.8s lr=3.000000e-04 loss_all=2.8244 acc_all=0.5905 loss_corrupt=2.8244 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8244 loss_recon=2.8244 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4991 init_acc_corrupt=0.4665 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=85700 micro_steps=171400 elapsed=96.1s lr=3.000000e-04 loss_all=2.8412 acc_all=0.5884 loss_corrupt=2.8412 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8412 loss_recon=2.8412 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5006 init_acc_corrupt=0.4648 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=85800 micro_steps=171600 elapsed=66.3s lr=3.000000e-04 loss_all=2.8132 acc_all=0.5917 loss_corrupt=2.8132 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8132 loss_recon=2.8132 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4989 init_acc_corrupt=0.4673 init_gold_top10=0.4957 init_gold_top100=0.5012
+step=85900 micro_steps=171800 elapsed=57.4s lr=3.000000e-04 loss_all=2.8318 acc_all=0.5890 loss_corrupt=2.8318 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8318 loss_recon=2.8318 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4996 init_acc_corrupt=0.4658 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=86000 micro_steps=172000 elapsed=57.4s lr=3.000000e-04 loss_all=2.8249 acc_all=0.5902 loss_corrupt=2.8249 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8249 loss_recon=2.8249 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4992 init_acc_corrupt=0.4664 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=86100 micro_steps=172200 elapsed=59.1s lr=3.000000e-04 loss_all=2.8223 acc_all=0.5910 loss_corrupt=2.8223 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8223 loss_recon=2.8223 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4987 init_acc_corrupt=0.4666 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=86200 micro_steps=172400 elapsed=57.5s lr=3.000000e-04 loss_all=2.8184 acc_all=0.5910 loss_corrupt=2.8184 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8184 loss_recon=2.8184 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4985 init_acc_corrupt=0.4672 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=86300 micro_steps=172600 elapsed=57.8s lr=3.000000e-04 loss_all=2.8631 acc_all=0.5855 loss_corrupt=2.8631 acc_corrupt=0.5855 corrupt_frac=1.0000 loss=2.8631 loss_recon=2.8631 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5030 init_acc_corrupt=0.4614 init_gold_top10=0.4913 init_gold_top100=0.4972
+step=86400 micro_steps=172800 elapsed=77.9s lr=3.000000e-04 loss_all=2.8730 acc_all=0.5841 loss_corrupt=2.8730 acc_corrupt=0.5841 corrupt_frac=1.0000 loss=2.8730 loss_recon=2.8730 loss_meanflow=0.0000 mean_model_t=0.4959 mean_corrupt_t=0.4959 wrong_frac=0.5041 init_acc_corrupt=0.4608 init_gold_top10=0.4902 init_gold_top100=0.4960
+step=86500 micro_steps=173000 elapsed=85.6s lr=3.000000e-04 loss_all=2.8124 acc_all=0.5922 loss_corrupt=2.8124 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8124 loss_recon=2.8124 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4974 init_acc_corrupt=0.4685 init_gold_top10=0.4970 init_gold_top100=0.5028
+step=86600 micro_steps=173200 elapsed=93.6s lr=3.000000e-04 loss_all=2.8142 acc_all=0.5918 loss_corrupt=2.8142 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8142 loss_recon=2.8142 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4981 init_acc_corrupt=0.4683 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=86700 micro_steps=173400 elapsed=76.9s lr=3.000000e-04 loss_all=2.8291 acc_all=0.5902 loss_corrupt=2.8291 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8291 loss_recon=2.8291 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4983 init_acc_corrupt=0.4674 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=86800 micro_steps=173600 elapsed=83.1s lr=3.000000e-04 loss_all=2.8077 acc_all=0.5925 loss_corrupt=2.8077 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8077 loss_recon=2.8077 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4979 init_acc_corrupt=0.4680 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=86900 micro_steps=173800 elapsed=92.9s lr=3.000000e-04 loss_all=2.8213 acc_all=0.5909 loss_corrupt=2.8213 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8213 loss_recon=2.8213 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4992 init_acc_corrupt=0.4661 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=87000 micro_steps=174000 elapsed=69.1s lr=3.000000e-04 loss_all=2.8550 acc_all=0.5866 loss_corrupt=2.8550 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8550 loss_recon=2.8550 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5022 init_acc_corrupt=0.4626 init_gold_top10=0.4920 init_gold_top100=0.4979
+step=87100 micro_steps=174200 elapsed=86.3s lr=3.000000e-04 loss_all=2.8525 acc_all=0.5863 loss_corrupt=2.8525 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8525 loss_recon=2.8525 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5025 init_acc_corrupt=0.4627 init_gold_top10=0.4919 init_gold_top100=0.4976
+step=87200 micro_steps=174400 elapsed=82.4s lr=3.000000e-04 loss_all=2.8298 acc_all=0.5892 loss_corrupt=2.8298 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8298 loss_recon=2.8298 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5006 init_acc_corrupt=0.4651 init_gold_top10=0.4939 init_gold_top100=0.4995
+step=87300 micro_steps=174600 elapsed=62.4s lr=3.000000e-04 loss_all=2.8432 acc_all=0.5878 loss_corrupt=2.8432 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8432 loss_recon=2.8432 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5017 init_acc_corrupt=0.4639 init_gold_top10=0.4924 init_gold_top100=0.4983
+step=87400 micro_steps=174800 elapsed=64.0s lr=3.000000e-04 loss_all=2.8198 acc_all=0.5910 loss_corrupt=2.8198 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8198 loss_recon=2.8198 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4987 init_acc_corrupt=0.4670 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=87500 micro_steps=175000 elapsed=64.1s lr=3.000000e-04 loss_all=2.8729 acc_all=0.5842 loss_corrupt=2.8729 acc_corrupt=0.5842 corrupt_frac=1.0000 loss=2.8729 loss_recon=2.8729 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5041 init_acc_corrupt=0.4605 init_gold_top10=0.4902 init_gold_top100=0.4960
+step=87600 micro_steps=175200 elapsed=69.2s lr=3.000000e-04 loss_all=2.8284 acc_all=0.5897 loss_corrupt=2.8284 acc_corrupt=0.5897 corrupt_frac=1.0000 loss=2.8284 loss_recon=2.8284 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4991 init_acc_corrupt=0.4657 init_gold_top10=0.4952 init_gold_top100=0.5011
+step=87700 micro_steps=175400 elapsed=80.8s lr=3.000000e-04 loss_all=2.8086 acc_all=0.5922 loss_corrupt=2.8086 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8086 loss_recon=2.8086 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4975 init_acc_corrupt=0.4684 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=87800 micro_steps=175600 elapsed=74.9s lr=3.000000e-04 loss_all=2.8100 acc_all=0.5918 loss_corrupt=2.8100 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8100 loss_recon=2.8100 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4990 init_acc_corrupt=0.4670 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=87900 micro_steps=175800 elapsed=74.2s lr=3.000000e-04 loss_all=2.8328 acc_all=0.5896 loss_corrupt=2.8328 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8328 loss_recon=2.8328 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4994 init_acc_corrupt=0.4659 init_gold_top10=0.4949 init_gold_top100=0.5006
+step=88000 micro_steps=176000 elapsed=63.1s lr=3.000000e-04 loss_all=2.8183 acc_all=0.5917 loss_corrupt=2.8183 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8183 loss_recon=2.8183 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4984 init_acc_corrupt=0.4674 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=88100 micro_steps=176200 elapsed=64.7s lr=3.000000e-04 loss_all=2.7986 acc_all=0.5937 loss_corrupt=2.7986 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7986 loss_recon=2.7986 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4964 init_acc_corrupt=0.4694 init_gold_top10=0.4981 init_gold_top100=0.5037
+step=88200 micro_steps=176400 elapsed=63.1s lr=3.000000e-04 loss_all=2.7919 acc_all=0.5946 loss_corrupt=2.7919 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7919 loss_recon=2.7919 loss_meanflow=0.0000 mean_model_t=0.5046 mean_corrupt_t=0.5046 wrong_frac=0.4952 init_acc_corrupt=0.4716 init_gold_top10=0.4993 init_gold_top100=0.5049
+step=88300 micro_steps=176600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8755 acc_all=0.5838 loss_corrupt=2.8755 acc_corrupt=0.5838 corrupt_frac=1.0000 loss=2.8755 loss_recon=2.8755 loss_meanflow=0.0000 mean_model_t=0.4939 mean_corrupt_t=0.4939 wrong_frac=0.5060 init_acc_corrupt=0.4589 init_gold_top10=0.4882 init_gold_top100=0.4941
+step=88400 micro_steps=176800 elapsed=68.2s lr=3.000000e-04 loss_all=2.8554 acc_all=0.5859 loss_corrupt=2.8554 acc_corrupt=0.5859 corrupt_frac=1.0000 loss=2.8554 loss_recon=2.8554 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5033 init_acc_corrupt=0.4618 init_gold_top10=0.4910 init_gold_top100=0.4968
+step=88500 micro_steps=177000 elapsed=81.7s lr=3.000000e-04 loss_all=2.8230 acc_all=0.5904 loss_corrupt=2.8230 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8230 loss_recon=2.8230 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5001 init_acc_corrupt=0.4660 init_gold_top10=0.4944 init_gold_top100=0.5000
+step=88600 micro_steps=177200 elapsed=87.6s lr=3.000000e-04 loss_all=2.8336 acc_all=0.5891 loss_corrupt=2.8336 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8336 loss_recon=2.8336 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5004 init_acc_corrupt=0.4648 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=88700 micro_steps=177400 elapsed=100.3s lr=3.000000e-04 loss_all=2.8176 acc_all=0.5917 loss_corrupt=2.8176 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8176 loss_recon=2.8176 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4982 init_acc_corrupt=0.4676 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=88800 micro_steps=177600 elapsed=91.4s lr=3.000000e-04 loss_all=2.8410 acc_all=0.5875 loss_corrupt=2.8410 acc_corrupt=0.5875 corrupt_frac=1.0000 loss=2.8410 loss_recon=2.8410 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5020 init_acc_corrupt=0.4627 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=88900 micro_steps=177800 elapsed=100.5s lr=3.000000e-04 loss_all=2.8296 acc_all=0.5897 loss_corrupt=2.8296 acc_corrupt=0.5897 corrupt_frac=1.0000 loss=2.8296 loss_recon=2.8296 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5002 init_acc_corrupt=0.4649 init_gold_top10=0.4943 init_gold_top100=0.5000
+step=89000 micro_steps=178000 elapsed=84.3s lr=3.000000e-04 loss_all=2.7824 acc_all=0.5958 loss_corrupt=2.7824 acc_corrupt=0.5958 corrupt_frac=1.0000 loss=2.7824 loss_recon=2.7824 loss_meanflow=0.0000 mean_model_t=0.5062 mean_corrupt_t=0.5062 wrong_frac=0.4938 init_acc_corrupt=0.4725 init_gold_top10=0.5007 init_gold_top100=0.5063
+step=89100 micro_steps=178200 elapsed=65.6s lr=3.000000e-04 loss_all=2.8127 acc_all=0.5922 loss_corrupt=2.8127 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8127 loss_recon=2.8127 loss_meanflow=0.0000 mean_model_t=0.5036 mean_corrupt_t=0.5036 wrong_frac=0.4967 init_acc_corrupt=0.4688 init_gold_top10=0.4977 init_gold_top100=0.5034
+step=89200 micro_steps=178400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8479 acc_all=0.5877 loss_corrupt=2.8479 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8479 loss_recon=2.8479 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5024 init_acc_corrupt=0.4625 init_gold_top10=0.4919 init_gold_top100=0.4977
+step=89300 micro_steps=178600 elapsed=63.8s lr=3.000000e-04 loss_all=2.8318 acc_all=0.5895 loss_corrupt=2.8318 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8318 loss_recon=2.8318 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4997 init_acc_corrupt=0.4665 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=89400 micro_steps=178800 elapsed=64.4s lr=3.000000e-04 loss_all=2.8109 acc_all=0.5919 loss_corrupt=2.8109 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8109 loss_recon=2.8109 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4984 init_acc_corrupt=0.4670 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=89500 micro_steps=179000 elapsed=64.7s lr=3.000000e-04 loss_all=2.8347 acc_all=0.5885 loss_corrupt=2.8347 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8347 loss_recon=2.8347 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5005 init_acc_corrupt=0.4647 init_gold_top10=0.4938 init_gold_top100=0.4997
+step=89600 micro_steps=179200 elapsed=79.2s lr=3.000000e-04 loss_all=2.7841 acc_all=0.5949 loss_corrupt=2.7841 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7841 loss_recon=2.7841 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4965 init_acc_corrupt=0.4703 init_gold_top10=0.4981 init_gold_top100=0.5036
+step=89700 micro_steps=179400 elapsed=92.2s lr=3.000000e-04 loss_all=2.8049 acc_all=0.5932 loss_corrupt=2.8049 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.8049 loss_recon=2.8049 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4972 init_acc_corrupt=0.4683 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=89800 micro_steps=179600 elapsed=91.7s lr=3.000000e-04 loss_all=2.8258 acc_all=0.5902 loss_corrupt=2.8258 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8258 loss_recon=2.8258 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5004 init_acc_corrupt=0.4659 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=89900 micro_steps=179800 elapsed=94.4s lr=3.000000e-04 loss_all=2.8376 acc_all=0.5880 loss_corrupt=2.8376 acc_corrupt=0.5880 corrupt_frac=1.0000 loss=2.8376 loss_recon=2.8376 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5025 init_acc_corrupt=0.4628 init_gold_top10=0.4918 init_gold_top100=0.4976
+step=90000 micro_steps=180000 elapsed=80.7s lr=3.000000e-04 loss_all=2.8281 acc_all=0.5894 loss_corrupt=2.8281 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8281 loss_recon=2.8281 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5005 init_acc_corrupt=0.4644 init_gold_top10=0.4938 init_gold_top100=0.4997
+step=90100 micro_steps=180200 elapsed=94.1s lr=3.000000e-04 loss_all=2.8357 acc_all=0.5895 loss_corrupt=2.8357 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8357 loss_recon=2.8357 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4994 init_acc_corrupt=0.4661 init_gold_top10=0.4949 init_gold_top100=0.5007
+step=90200 micro_steps=180400 elapsed=95.4s lr=3.000000e-04 loss_all=2.8301 acc_all=0.5890 loss_corrupt=2.8301 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8301 loss_recon=2.8301 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5012 init_acc_corrupt=0.4643 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=90300 micro_steps=180600 elapsed=90.4s lr=3.000000e-04 loss_all=2.8581 acc_all=0.5858 loss_corrupt=2.8581 acc_corrupt=0.5858 corrupt_frac=1.0000 loss=2.8581 loss_recon=2.8581 loss_meanflow=0.0000 mean_model_t=0.4957 mean_corrupt_t=0.4957 wrong_frac=0.5039 init_acc_corrupt=0.4608 init_gold_top10=0.4903 init_gold_top100=0.4962
+step=90400 micro_steps=180800 elapsed=83.0s lr=3.000000e-04 loss_all=2.8504 acc_all=0.5869 loss_corrupt=2.8504 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8504 loss_recon=2.8504 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5031 init_acc_corrupt=0.4623 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=90500 micro_steps=181000 elapsed=89.8s lr=3.000000e-04 loss_all=2.7830 acc_all=0.5955 loss_corrupt=2.7830 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7830 loss_recon=2.7830 loss_meanflow=0.0000 mean_model_t=0.5050 mean_corrupt_t=0.5050 wrong_frac=0.4945 init_acc_corrupt=0.4716 init_gold_top10=0.5000 init_gold_top100=0.5057
+step=90600 micro_steps=181200 elapsed=97.9s lr=3.000000e-04 loss_all=2.8104 acc_all=0.5919 loss_corrupt=2.8104 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8104 loss_recon=2.8104 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4984 init_acc_corrupt=0.4671 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=90700 micro_steps=181400 elapsed=88.2s lr=3.000000e-04 loss_all=2.8492 acc_all=0.5871 loss_corrupt=2.8492 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8492 loss_recon=2.8492 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5019 init_acc_corrupt=0.4631 init_gold_top10=0.4924 init_gold_top100=0.4983
+step=90800 micro_steps=181600 elapsed=85.4s lr=3.000000e-04 loss_all=2.8283 acc_all=0.5895 loss_corrupt=2.8283 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8283 loss_recon=2.8283 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5007 init_acc_corrupt=0.4647 init_gold_top10=0.4938 init_gold_top100=0.4995
+step=90900 micro_steps=181800 elapsed=87.9s lr=3.000000e-04 loss_all=2.8180 acc_all=0.5908 loss_corrupt=2.8180 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8180 loss_recon=2.8180 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4989 init_acc_corrupt=0.4666 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=91000 micro_steps=182000 elapsed=100.1s lr=3.000000e-04 loss_all=2.8274 acc_all=0.5903 loss_corrupt=2.8274 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8274 loss_recon=2.8274 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4990 init_acc_corrupt=0.4659 init_gold_top10=0.4955 init_gold_top100=0.5011
+step=91100 micro_steps=182200 elapsed=87.1s lr=3.000000e-04 loss_all=2.8204 acc_all=0.5910 loss_corrupt=2.8204 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8204 loss_recon=2.8204 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4990 init_acc_corrupt=0.4663 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=91200 micro_steps=182400 elapsed=86.9s lr=3.000000e-04 loss_all=2.8328 acc_all=0.5895 loss_corrupt=2.8328 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8328 loss_recon=2.8328 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5004 init_acc_corrupt=0.4649 init_gold_top10=0.4938 init_gold_top100=0.4997
+step=91300 micro_steps=182600 elapsed=87.0s lr=3.000000e-04 loss_all=2.8052 acc_all=0.5925 loss_corrupt=2.8052 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8052 loss_recon=2.8052 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4980 init_acc_corrupt=0.4679 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=91400 micro_steps=182800 elapsed=99.9s lr=3.000000e-04 loss_all=2.8480 acc_all=0.5869 loss_corrupt=2.8480 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8480 loss_recon=2.8480 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5035 init_acc_corrupt=0.4616 init_gold_top10=0.4908 init_gold_top100=0.4967
+step=91500 micro_steps=183000 elapsed=83.1s lr=3.000000e-04 loss_all=2.8012 acc_all=0.5933 loss_corrupt=2.8012 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.8012 loss_recon=2.8012 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4966 init_acc_corrupt=0.4690 init_gold_top10=0.4980 init_gold_top100=0.5035
+step=91600 micro_steps=183200 elapsed=63.4s lr=3.000000e-04 loss_all=2.8022 acc_all=0.5930 loss_corrupt=2.8022 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.8022 loss_recon=2.8022 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4972 init_acc_corrupt=0.4685 init_gold_top10=0.4973 init_gold_top100=0.5030
+step=91700 micro_steps=183400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8230 acc_all=0.5904 loss_corrupt=2.8230 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8230 loss_recon=2.8230 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4990 init_acc_corrupt=0.4664 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=91800 micro_steps=183600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8351 acc_all=0.5886 loss_corrupt=2.8351 acc_corrupt=0.5886 corrupt_frac=1.0000 loss=2.8351 loss_recon=2.8351 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5020 init_acc_corrupt=0.4635 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=91900 micro_steps=183800 elapsed=77.7s lr=3.000000e-04 loss_all=2.7998 acc_all=0.5933 loss_corrupt=2.7998 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7998 loss_recon=2.7998 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4969 init_acc_corrupt=0.4684 init_gold_top10=0.4975 init_gold_top100=0.5033
+step=92000 micro_steps=184000 elapsed=83.2s lr=3.000000e-04 loss_all=2.8342 acc_all=0.5889 loss_corrupt=2.8342 acc_corrupt=0.5889 corrupt_frac=1.0000 loss=2.8342 loss_recon=2.8342 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5025 init_acc_corrupt=0.4632 init_gold_top10=0.4920 init_gold_top100=0.4976
+step=92100 micro_steps=184200 elapsed=89.6s lr=3.000000e-04 loss_all=2.8325 acc_all=0.5888 loss_corrupt=2.8325 acc_corrupt=0.5888 corrupt_frac=1.0000 loss=2.8325 loss_recon=2.8325 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5016 init_acc_corrupt=0.4642 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=92200 micro_steps=184400 elapsed=79.8s lr=3.000000e-04 loss_all=2.7806 acc_all=0.5957 loss_corrupt=2.7806 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7806 loss_recon=2.7806 loss_meanflow=0.0000 mean_model_t=0.5055 mean_corrupt_t=0.5055 wrong_frac=0.4950 init_acc_corrupt=0.4710 init_gold_top10=0.4996 init_gold_top100=0.5052
+step=92300 micro_steps=184600 elapsed=89.5s lr=3.000000e-04 loss_all=2.8570 acc_all=0.5863 loss_corrupt=2.8570 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8570 loss_recon=2.8570 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5030 init_acc_corrupt=0.4624 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=92400 micro_steps=184800 elapsed=69.4s lr=3.000000e-04 loss_all=2.8323 acc_all=0.5893 loss_corrupt=2.8323 acc_corrupt=0.5893 corrupt_frac=1.0000 loss=2.8323 loss_recon=2.8323 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.5000 init_acc_corrupt=0.4656 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=92500 micro_steps=185000 elapsed=71.3s lr=3.000000e-04 loss_all=2.8514 acc_all=0.5871 loss_corrupt=2.8514 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8514 loss_recon=2.8514 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5020 init_acc_corrupt=0.4632 init_gold_top10=0.4922 init_gold_top100=0.4980
+step=92600 micro_steps=185200 elapsed=73.2s lr=3.000000e-04 loss_all=2.8422 acc_all=0.5883 loss_corrupt=2.8422 acc_corrupt=0.5883 corrupt_frac=1.0000 loss=2.8422 loss_recon=2.8422 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5013 init_acc_corrupt=0.4640 init_gold_top10=0.4931 init_gold_top100=0.4988
+step=92700 micro_steps=185400 elapsed=76.2s lr=3.000000e-04 loss_all=2.8236 acc_all=0.5904 loss_corrupt=2.8236 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8236 loss_recon=2.8236 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4998 init_acc_corrupt=0.4658 init_gold_top10=0.4944 init_gold_top100=0.5003
+step=92800 micro_steps=185600 elapsed=73.8s lr=3.000000e-04 loss_all=2.8217 acc_all=0.5905 loss_corrupt=2.8217 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8217 loss_recon=2.8217 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5001 init_acc_corrupt=0.4659 init_gold_top10=0.4944 init_gold_top100=0.4999
+step=92900 micro_steps=185800 elapsed=75.3s lr=3.000000e-04 loss_all=2.7949 acc_all=0.5934 loss_corrupt=2.7949 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7949 loss_recon=2.7949 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4977 init_acc_corrupt=0.4684 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=93000 micro_steps=186000 elapsed=79.1s lr=3.000000e-04 loss_all=2.7932 acc_all=0.5945 loss_corrupt=2.7932 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7932 loss_recon=2.7932 loss_meanflow=0.0000 mean_model_t=0.5040 mean_corrupt_t=0.5040 wrong_frac=0.4961 init_acc_corrupt=0.4701 init_gold_top10=0.4984 init_gold_top100=0.5040
+step=93100 micro_steps=186200 elapsed=67.8s lr=3.000000e-04 loss_all=2.8164 acc_all=0.5911 loss_corrupt=2.8164 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8164 loss_recon=2.8164 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4991 init_acc_corrupt=0.4666 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=93200 micro_steps=186400 elapsed=78.6s lr=3.000000e-04 loss_all=2.8209 acc_all=0.5907 loss_corrupt=2.8209 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8209 loss_recon=2.8209 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4989 init_acc_corrupt=0.4665 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=93300 micro_steps=186600 elapsed=75.8s lr=3.000000e-04 loss_all=2.8254 acc_all=0.5900 loss_corrupt=2.8254 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8254 loss_recon=2.8254 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4989 init_acc_corrupt=0.4667 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=93400 micro_steps=186800 elapsed=76.1s lr=3.000000e-04 loss_all=2.8304 acc_all=0.5892 loss_corrupt=2.8304 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8304 loss_recon=2.8304 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5012 init_acc_corrupt=0.4643 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=93500 micro_steps=187000 elapsed=76.7s lr=3.000000e-04 loss_all=2.7973 acc_all=0.5933 loss_corrupt=2.7973 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7973 loss_recon=2.7973 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4977 init_acc_corrupt=0.4687 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=93600 micro_steps=187200 elapsed=79.3s lr=3.000000e-04 loss_all=2.8071 acc_all=0.5931 loss_corrupt=2.8071 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.8071 loss_recon=2.8071 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4971 init_acc_corrupt=0.4681 init_gold_top10=0.4974 init_gold_top100=0.5031
+step=93700 micro_steps=187400 elapsed=76.1s lr=3.000000e-04 loss_all=2.8521 acc_all=0.5866 loss_corrupt=2.8521 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8521 loss_recon=2.8521 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5026 init_acc_corrupt=0.4625 init_gold_top10=0.4917 init_gold_top100=0.4976
+step=93800 micro_steps=187600 elapsed=76.6s lr=3.000000e-04 loss_all=2.8278 acc_all=0.5899 loss_corrupt=2.8278 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8278 loss_recon=2.8278 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5001 init_acc_corrupt=0.4651 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=93900 micro_steps=187800 elapsed=75.7s lr=3.000000e-04 loss_all=2.8237 acc_all=0.5903 loss_corrupt=2.8237 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8237 loss_recon=2.8237 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5003 init_acc_corrupt=0.4653 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=94000 micro_steps=188000 elapsed=79.8s lr=3.000000e-04 loss_all=2.7923 acc_all=0.5942 loss_corrupt=2.7923 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7923 loss_recon=2.7923 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4966 init_acc_corrupt=0.4688 init_gold_top10=0.4979 init_gold_top100=0.5036
+step=94100 micro_steps=188200 elapsed=81.4s lr=3.000000e-04 loss_all=2.8072 acc_all=0.5927 loss_corrupt=2.8072 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8072 loss_recon=2.8072 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4983 init_acc_corrupt=0.4675 init_gold_top10=0.4962 init_gold_top100=0.5018
+step=94200 micro_steps=188400 elapsed=79.3s lr=3.000000e-04 loss_all=2.8368 acc_all=0.5884 loss_corrupt=2.8368 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8368 loss_recon=2.8368 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5019 init_acc_corrupt=0.4640 init_gold_top10=0.4925 init_gold_top100=0.4983
+step=94300 micro_steps=188600 elapsed=79.3s lr=3.000000e-04 loss_all=2.8329 acc_all=0.5891 loss_corrupt=2.8329 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8329 loss_recon=2.8329 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5016 init_acc_corrupt=0.4634 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=94400 micro_steps=188800 elapsed=78.6s lr=3.000000e-04 loss_all=2.7994 acc_all=0.5937 loss_corrupt=2.7994 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7994 loss_recon=2.7994 loss_meanflow=0.0000 mean_model_t=0.5036 mean_corrupt_t=0.5036 wrong_frac=0.4964 init_acc_corrupt=0.4695 init_gold_top10=0.4981 init_gold_top100=0.5038
+step=94500 micro_steps=189000 elapsed=79.3s lr=3.000000e-04 loss_all=2.8455 acc_all=0.5875 loss_corrupt=2.8455 acc_corrupt=0.5875 corrupt_frac=1.0000 loss=2.8455 loss_recon=2.8455 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5029 init_acc_corrupt=0.4621 init_gold_top10=0.4914 init_gold_top100=0.4972
+step=94600 micro_steps=189200 elapsed=79.2s lr=3.000000e-04 loss_all=2.8017 acc_all=0.5931 loss_corrupt=2.8017 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.8017 loss_recon=2.8017 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4979 init_acc_corrupt=0.4680 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=94700 micro_steps=189400 elapsed=78.8s lr=3.000000e-04 loss_all=2.8271 acc_all=0.5897 loss_corrupt=2.8271 acc_corrupt=0.5897 corrupt_frac=1.0000 loss=2.8271 loss_recon=2.8271 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5004 init_acc_corrupt=0.4646 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=94800 micro_steps=189600 elapsed=78.8s lr=3.000000e-04 loss_all=2.8357 acc_all=0.5887 loss_corrupt=2.8357 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8357 loss_recon=2.8357 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5004 init_acc_corrupt=0.4647 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=94900 micro_steps=189800 elapsed=79.0s lr=3.000000e-04 loss_all=2.7893 acc_all=0.5952 loss_corrupt=2.7893 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7893 loss_recon=2.7893 loss_meanflow=0.0000 mean_model_t=0.5042 mean_corrupt_t=0.5042 wrong_frac=0.4955 init_acc_corrupt=0.4703 init_gold_top10=0.4991 init_gold_top100=0.5047
+step=95000 micro_steps=190000 elapsed=79.2s lr=3.000000e-04 loss_all=2.8276 acc_all=0.5898 loss_corrupt=2.8276 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8276 loss_recon=2.8276 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5011 init_acc_corrupt=0.4651 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=95100 micro_steps=190200 elapsed=83.3s lr=3.000000e-04 loss_all=2.8252 acc_all=0.5899 loss_corrupt=2.8252 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8252 loss_recon=2.8252 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5012 init_acc_corrupt=0.4648 init_gold_top10=0.4932 init_gold_top100=0.4989
+step=95200 micro_steps=190400 elapsed=85.2s lr=3.000000e-04 loss_all=2.8230 acc_all=0.5905 loss_corrupt=2.8230 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8230 loss_recon=2.8230 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4995 init_acc_corrupt=0.4652 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=95300 micro_steps=190600 elapsed=79.6s lr=3.000000e-04 loss_all=2.8091 acc_all=0.5919 loss_corrupt=2.8091 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8091 loss_recon=2.8091 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4988 init_acc_corrupt=0.4671 init_gold_top10=0.4958 init_gold_top100=0.5014
+step=95400 micro_steps=190800 elapsed=82.4s lr=3.000000e-04 loss_all=2.8598 acc_all=0.5852 loss_corrupt=2.8598 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8598 loss_recon=2.8598 loss_meanflow=0.0000 mean_model_t=0.4955 mean_corrupt_t=0.4955 wrong_frac=0.5049 init_acc_corrupt=0.4594 init_gold_top10=0.4893 init_gold_top100=0.4952
+step=95500 micro_steps=191000 elapsed=85.6s lr=3.000000e-04 loss_all=2.8180 acc_all=0.5911 loss_corrupt=2.8180 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8180 loss_recon=2.8180 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4991 init_acc_corrupt=0.4663 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=95600 micro_steps=191200 elapsed=80.7s lr=3.000000e-04 loss_all=2.8054 acc_all=0.5925 loss_corrupt=2.8054 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8054 loss_recon=2.8054 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4985 init_acc_corrupt=0.4672 init_gold_top10=0.4960 init_gold_top100=0.5016
+step=95700 micro_steps=191400 elapsed=81.4s lr=3.000000e-04 loss_all=2.8451 acc_all=0.5877 loss_corrupt=2.8451 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8451 loss_recon=2.8451 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5016 init_acc_corrupt=0.4636 init_gold_top10=0.4926 init_gold_top100=0.4986
+step=95800 micro_steps=191600 elapsed=85.7s lr=3.000000e-04 loss_all=2.8221 acc_all=0.5902 loss_corrupt=2.8221 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8221 loss_recon=2.8221 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5002 init_acc_corrupt=0.4650 init_gold_top10=0.4944 init_gold_top100=0.4999
+step=95900 micro_steps=191800 elapsed=81.8s lr=3.000000e-04 loss_all=2.8593 acc_all=0.5860 loss_corrupt=2.8593 acc_corrupt=0.5860 corrupt_frac=1.0000 loss=2.8593 loss_recon=2.8593 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5032 init_acc_corrupt=0.4618 init_gold_top10=0.4911 init_gold_top100=0.4970
+step=96000 micro_steps=192000 elapsed=80.6s lr=3.000000e-04 loss_all=2.7803 acc_all=0.5963 loss_corrupt=2.7803 acc_corrupt=0.5963 corrupt_frac=1.0000 loss=2.7803 loss_recon=2.7803 loss_meanflow=0.0000 mean_model_t=0.5050 mean_corrupt_t=0.5050 wrong_frac=0.4949 init_acc_corrupt=0.4710 init_gold_top10=0.4995 init_gold_top100=0.5052
+step=96100 micro_steps=192200 elapsed=87.1s lr=3.000000e-04 loss_all=2.8243 acc_all=0.5901 loss_corrupt=2.8243 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8243 loss_recon=2.8243 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4999 init_acc_corrupt=0.4655 init_gold_top10=0.4946 init_gold_top100=0.5002
+step=96200 micro_steps=192400 elapsed=65.2s lr=3.000000e-04 loss_all=2.8463 acc_all=0.5873 loss_corrupt=2.8463 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8463 loss_recon=2.8463 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5027 init_acc_corrupt=0.4622 init_gold_top10=0.4916 init_gold_top100=0.4974
+step=96300 micro_steps=192600 elapsed=63.9s lr=3.000000e-04 loss_all=2.8073 acc_all=0.5924 loss_corrupt=2.8073 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8073 loss_recon=2.8073 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4972 init_acc_corrupt=0.4684 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=96400 micro_steps=192800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8112 acc_all=0.5917 loss_corrupt=2.8112 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8112 loss_recon=2.8112 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4990 init_acc_corrupt=0.4668 init_gold_top10=0.4954 init_gold_top100=0.5012
+step=96500 micro_steps=193000 elapsed=63.7s lr=3.000000e-04 loss_all=2.8461 acc_all=0.5878 loss_corrupt=2.8461 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8461 loss_recon=2.8461 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5031 init_acc_corrupt=0.4619 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=96600 micro_steps=193200 elapsed=63.7s lr=3.000000e-04 loss_all=2.8188 acc_all=0.5910 loss_corrupt=2.8188 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8188 loss_recon=2.8188 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4997 init_acc_corrupt=0.4662 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=96700 micro_steps=193400 elapsed=63.7s lr=3.000000e-04 loss_all=2.7877 acc_all=0.5955 loss_corrupt=2.7877 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7877 loss_recon=2.7877 loss_meanflow=0.0000 mean_model_t=0.5055 mean_corrupt_t=0.5055 wrong_frac=0.4947 init_acc_corrupt=0.4715 init_gold_top10=0.4997 init_gold_top100=0.5055
+step=96800 micro_steps=193600 elapsed=87.6s lr=3.000000e-04 loss_all=2.7995 acc_all=0.5932 loss_corrupt=2.7995 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7995 loss_recon=2.7995 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4969 init_acc_corrupt=0.4687 init_gold_top10=0.4975 init_gold_top100=0.5032
+step=96900 micro_steps=193800 elapsed=85.7s lr=3.000000e-04 loss_all=2.8225 acc_all=0.5902 loss_corrupt=2.8225 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8225 loss_recon=2.8225 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5003 init_acc_corrupt=0.4655 init_gold_top10=0.4942 init_gold_top100=0.4998
+step=97000 micro_steps=194000 elapsed=91.9s lr=3.000000e-04 loss_all=2.8212 acc_all=0.5906 loss_corrupt=2.8212 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8212 loss_recon=2.8212 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4987 init_acc_corrupt=0.4665 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=97100 micro_steps=194200 elapsed=69.0s lr=3.000000e-04 loss_all=2.8266 acc_all=0.5896 loss_corrupt=2.8266 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8266 loss_recon=2.8266 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5006 init_acc_corrupt=0.4642 init_gold_top10=0.4937 init_gold_top100=0.4996
+step=97200 micro_steps=194400 elapsed=65.6s lr=3.000000e-04 loss_all=2.7834 acc_all=0.5954 loss_corrupt=2.7834 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7834 loss_recon=2.7834 loss_meanflow=0.0000 mean_model_t=0.5049 mean_corrupt_t=0.5049 wrong_frac=0.4954 init_acc_corrupt=0.4703 init_gold_top10=0.4991 init_gold_top100=0.5048
+step=97300 micro_steps=194600 elapsed=64.3s lr=3.000000e-04 loss_all=2.8108 acc_all=0.5916 loss_corrupt=2.8108 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8108 loss_recon=2.8108 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4986 init_acc_corrupt=0.4670 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=97400 micro_steps=194800 elapsed=63.8s lr=3.000000e-04 loss_all=2.7801 acc_all=0.5961 loss_corrupt=2.7801 acc_corrupt=0.5961 corrupt_frac=1.0000 loss=2.7801 loss_recon=2.7801 loss_meanflow=0.0000 mean_model_t=0.5053 mean_corrupt_t=0.5053 wrong_frac=0.4950 init_acc_corrupt=0.4716 init_gold_top10=0.4996 init_gold_top100=0.5051
+step=97500 micro_steps=195000 elapsed=64.4s lr=3.000000e-04 loss_all=2.8266 acc_all=0.5900 loss_corrupt=2.8266 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8266 loss_recon=2.8266 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5004 init_acc_corrupt=0.4649 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=97600 micro_steps=195200 elapsed=64.4s lr=3.000000e-04 loss_all=2.8304 acc_all=0.5892 loss_corrupt=2.8304 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8304 loss_recon=2.8304 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5020 init_acc_corrupt=0.4636 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=97700 micro_steps=195400 elapsed=64.3s lr=3.000000e-04 loss_all=2.8207 acc_all=0.5907 loss_corrupt=2.8207 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8207 loss_recon=2.8207 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4998 init_acc_corrupt=0.4655 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=97800 micro_steps=195600 elapsed=64.5s lr=3.000000e-04 loss_all=2.8148 acc_all=0.5913 loss_corrupt=2.8148 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8148 loss_recon=2.8148 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4990 init_acc_corrupt=0.4672 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=97900 micro_steps=195800 elapsed=64.5s lr=3.000000e-04 loss_all=2.8382 acc_all=0.5878 loss_corrupt=2.8382 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8382 loss_recon=2.8382 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5030 init_acc_corrupt=0.4624 init_gold_top10=0.4915 init_gold_top100=0.4972
+step=98000 micro_steps=196000 elapsed=64.4s lr=3.000000e-04 loss_all=2.8116 acc_all=0.5923 loss_corrupt=2.8116 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8116 loss_recon=2.8116 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4975 init_acc_corrupt=0.4678 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=98100 micro_steps=196200 elapsed=64.6s lr=3.000000e-04 loss_all=2.8134 acc_all=0.5916 loss_corrupt=2.8134 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8134 loss_recon=2.8134 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4987 init_acc_corrupt=0.4676 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=98200 micro_steps=196400 elapsed=63.8s lr=3.000000e-04 loss_all=2.8082 acc_all=0.5921 loss_corrupt=2.8082 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8082 loss_recon=2.8082 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4985 init_acc_corrupt=0.4674 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=98300 micro_steps=196600 elapsed=64.0s lr=3.000000e-04 loss_all=2.7830 acc_all=0.5954 loss_corrupt=2.7830 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7830 loss_recon=2.7830 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4952 init_acc_corrupt=0.4711 init_gold_top10=0.4994 init_gold_top100=0.5050
+step=98400 micro_steps=196800 elapsed=64.0s lr=3.000000e-04 loss_all=2.8285 acc_all=0.5898 loss_corrupt=2.8285 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8285 loss_recon=2.8285 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.5000 init_acc_corrupt=0.4659 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=98500 micro_steps=197000 elapsed=64.2s lr=3.000000e-04 loss_all=2.8257 acc_all=0.5899 loss_corrupt=2.8257 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8257 loss_recon=2.8257 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5005 init_acc_corrupt=0.4652 init_gold_top10=0.4940 init_gold_top100=0.4996
+step=98600 micro_steps=197200 elapsed=64.0s lr=3.000000e-04 loss_all=2.8224 acc_all=0.5908 loss_corrupt=2.8224 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8224 loss_recon=2.8224 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4995 init_acc_corrupt=0.4655 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=98700 micro_steps=197400 elapsed=64.3s lr=3.000000e-04 loss_all=2.8263 acc_all=0.5896 loss_corrupt=2.8263 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8263 loss_recon=2.8263 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5017 init_acc_corrupt=0.4638 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=98800 micro_steps=197600 elapsed=64.2s lr=3.000000e-04 loss_all=2.8096 acc_all=0.5924 loss_corrupt=2.8096 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8096 loss_recon=2.8096 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4984 init_acc_corrupt=0.4675 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=98900 micro_steps=197800 elapsed=65.1s lr=3.000000e-04 loss_all=2.8028 acc_all=0.5931 loss_corrupt=2.8028 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.8028 loss_recon=2.8028 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4984 init_acc_corrupt=0.4675 init_gold_top10=0.4962 init_gold_top100=0.5018
+step=99000 micro_steps=198000 elapsed=63.9s lr=3.000000e-04 loss_all=2.8425 acc_all=0.5878 loss_corrupt=2.8425 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8425 loss_recon=2.8425 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5021 init_acc_corrupt=0.4627 init_gold_top10=0.4921 init_gold_top100=0.4980
+step=99100 micro_steps=198200 elapsed=65.8s lr=3.000000e-04 loss_all=2.8182 acc_all=0.5913 loss_corrupt=2.8182 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8182 loss_recon=2.8182 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4993 init_acc_corrupt=0.4664 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=99200 micro_steps=198400 elapsed=64.3s lr=3.000000e-04 loss_all=2.7946 acc_all=0.5939 loss_corrupt=2.7946 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7946 loss_recon=2.7946 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4969 init_acc_corrupt=0.4693 init_gold_top10=0.4976 init_gold_top100=0.5032
+step=99300 micro_steps=198600 elapsed=63.8s lr=3.000000e-04 loss_all=2.8258 acc_all=0.5896 loss_corrupt=2.8258 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8258 loss_recon=2.8258 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5010 init_acc_corrupt=0.4646 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=99400 micro_steps=198800 elapsed=64.3s lr=3.000000e-04 loss_all=2.8117 acc_all=0.5917 loss_corrupt=2.8117 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8117 loss_recon=2.8117 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4993 init_acc_corrupt=0.4663 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=99500 micro_steps=199000 elapsed=63.9s lr=3.000000e-04 loss_all=2.7966 acc_all=0.5933 loss_corrupt=2.7966 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7966 loss_recon=2.7966 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4984 init_acc_corrupt=0.4677 init_gold_top10=0.4962 init_gold_top100=0.5018
+step=99600 micro_steps=199200 elapsed=64.0s lr=3.000000e-04 loss_all=2.8404 acc_all=0.5883 loss_corrupt=2.8404 acc_corrupt=0.5883 corrupt_frac=1.0000 loss=2.8404 loss_recon=2.8404 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5012 init_acc_corrupt=0.4637 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=99700 micro_steps=199400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8193 acc_all=0.5908 loss_corrupt=2.8193 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8193 loss_recon=2.8193 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4994 init_acc_corrupt=0.4656 init_gold_top10=0.4949 init_gold_top100=0.5008
+step=99800 micro_steps=199600 elapsed=64.0s lr=3.000000e-04 loss_all=2.8483 acc_all=0.5875 loss_corrupt=2.8483 acc_corrupt=0.5875 corrupt_frac=1.0000 loss=2.8483 loss_recon=2.8483 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5026 init_acc_corrupt=0.4622 init_gold_top10=0.4916 init_gold_top100=0.4975
+step=99900 micro_steps=199800 elapsed=63.9s lr=3.000000e-04 loss_all=2.8056 acc_all=0.5927 loss_corrupt=2.8056 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8056 loss_recon=2.8056 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4977 init_acc_corrupt=0.4679 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=100000 micro_steps=200000 elapsed=64.1s lr=3.000000e-04 loss_all=2.8056 acc_all=0.5925 loss_corrupt=2.8056 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8056 loss_recon=2.8056 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4986 init_acc_corrupt=0.4672 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=100100 micro_steps=200200 elapsed=67.0s lr=3.000000e-04 loss_all=2.8069 acc_all=0.5923 loss_corrupt=2.8069 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8069 loss_recon=2.8069 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4986 init_acc_corrupt=0.4675 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=100200 micro_steps=200400 elapsed=64.1s lr=3.000000e-04 loss_all=2.7972 acc_all=0.5932 loss_corrupt=2.7972 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7972 loss_recon=2.7972 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4980 init_acc_corrupt=0.4680 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=100300 micro_steps=200600 elapsed=62.7s lr=3.000000e-04 loss_all=2.8075 acc_all=0.5923 loss_corrupt=2.8075 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8075 loss_recon=2.8075 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4983 init_acc_corrupt=0.4672 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=100400 micro_steps=200800 elapsed=63.9s lr=3.000000e-04 loss_all=2.7871 acc_all=0.5955 loss_corrupt=2.7871 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7871 loss_recon=2.7871 loss_meanflow=0.0000 mean_model_t=0.5044 mean_corrupt_t=0.5044 wrong_frac=0.4954 init_acc_corrupt=0.4709 init_gold_top10=0.4990 init_gold_top100=0.5047
+step=100500 micro_steps=201000 elapsed=63.7s lr=3.000000e-04 loss_all=2.8124 acc_all=0.5917 loss_corrupt=2.8124 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8124 loss_recon=2.8124 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.4998 init_acc_corrupt=0.4661 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=100600 micro_steps=201200 elapsed=63.6s lr=3.000000e-04 loss_all=2.7992 acc_all=0.5936 loss_corrupt=2.7992 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7992 loss_recon=2.7992 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4974 init_acc_corrupt=0.4684 init_gold_top10=0.4969 init_gold_top100=0.5027
+step=100700 micro_steps=201400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7931 acc_all=0.5945 loss_corrupt=2.7931 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7931 loss_recon=2.7931 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4965 init_acc_corrupt=0.4696 init_gold_top10=0.4980 init_gold_top100=0.5037
+step=100800 micro_steps=201600 elapsed=63.7s lr=3.000000e-04 loss_all=2.8393 acc_all=0.5879 loss_corrupt=2.8393 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8393 loss_recon=2.8393 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5031 init_acc_corrupt=0.4626 init_gold_top10=0.4913 init_gold_top100=0.4969
+step=100900 micro_steps=201800 elapsed=63.6s lr=3.000000e-04 loss_all=2.8494 acc_all=0.5869 loss_corrupt=2.8494 acc_corrupt=0.5869 corrupt_frac=1.0000 loss=2.8494 loss_recon=2.8494 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5030 init_acc_corrupt=0.4616 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=101000 micro_steps=202000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8513 acc_all=0.5864 loss_corrupt=2.8513 acc_corrupt=0.5864 corrupt_frac=1.0000 loss=2.8513 loss_recon=2.8513 loss_meanflow=0.0000 mean_model_t=0.4957 mean_corrupt_t=0.4957 wrong_frac=0.5041 init_acc_corrupt=0.4604 init_gold_top10=0.4901 init_gold_top100=0.4961
+step=101100 micro_steps=202200 elapsed=64.9s lr=3.000000e-04 loss_all=2.8306 acc_all=0.5891 loss_corrupt=2.8306 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8306 loss_recon=2.8306 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5010 init_acc_corrupt=0.4644 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=101200 micro_steps=202400 elapsed=63.6s lr=3.000000e-04 loss_all=2.8331 acc_all=0.5889 loss_corrupt=2.8331 acc_corrupt=0.5889 corrupt_frac=1.0000 loss=2.8331 loss_recon=2.8331 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5023 init_acc_corrupt=0.4633 init_gold_top10=0.4920 init_gold_top100=0.4979
+step=101300 micro_steps=202600 elapsed=64.0s lr=3.000000e-04 loss_all=2.8257 acc_all=0.5899 loss_corrupt=2.8257 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8257 loss_recon=2.8257 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5001 init_acc_corrupt=0.4650 init_gold_top10=0.4943 init_gold_top100=0.5000
+step=101400 micro_steps=202800 elapsed=64.5s lr=3.000000e-04 loss_all=2.8190 acc_all=0.5911 loss_corrupt=2.8190 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8190 loss_recon=2.8190 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4995 init_acc_corrupt=0.4656 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=101500 micro_steps=203000 elapsed=64.3s lr=3.000000e-04 loss_all=2.7624 acc_all=0.5978 loss_corrupt=2.7624 acc_corrupt=0.5978 corrupt_frac=1.0000 loss=2.7624 loss_recon=2.7624 loss_meanflow=0.0000 mean_model_t=0.5066 mean_corrupt_t=0.5066 wrong_frac=0.4935 init_acc_corrupt=0.4728 init_gold_top10=0.5011 init_gold_top100=0.5067
+step=101600 micro_steps=203200 elapsed=64.5s lr=3.000000e-04 loss_all=2.8020 acc_all=0.5929 loss_corrupt=2.8020 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.8020 loss_recon=2.8020 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4988 init_acc_corrupt=0.4670 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=101700 micro_steps=203400 elapsed=64.4s lr=3.000000e-04 loss_all=2.8215 acc_all=0.5906 loss_corrupt=2.8215 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8215 loss_recon=2.8215 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4993 init_acc_corrupt=0.4660 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=101800 micro_steps=203600 elapsed=64.1s lr=3.000000e-04 loss_all=2.7955 acc_all=0.5935 loss_corrupt=2.7955 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7955 loss_recon=2.7955 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4979 init_acc_corrupt=0.4682 init_gold_top10=0.4966 init_gold_top100=0.5023
+step=101900 micro_steps=203800 elapsed=64.3s lr=3.000000e-04 loss_all=2.8180 acc_all=0.5908 loss_corrupt=2.8180 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8180 loss_recon=2.8180 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4999 init_acc_corrupt=0.4655 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=102000 micro_steps=204000 elapsed=64.5s lr=3.000000e-04 loss_all=2.8064 acc_all=0.5927 loss_corrupt=2.8064 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8064 loss_recon=2.8064 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4986 init_acc_corrupt=0.4668 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=102100 micro_steps=204200 elapsed=67.0s lr=3.000000e-04 loss_all=2.7694 acc_all=0.5970 loss_corrupt=2.7694 acc_corrupt=0.5970 corrupt_frac=1.0000 loss=2.7694 loss_recon=2.7694 loss_meanflow=0.0000 mean_model_t=0.5052 mean_corrupt_t=0.5052 wrong_frac=0.4946 init_acc_corrupt=0.4718 init_gold_top10=0.5000 init_gold_top100=0.5056
+step=102200 micro_steps=204400 elapsed=66.9s lr=3.000000e-04 loss_all=2.8568 acc_all=0.5861 loss_corrupt=2.8568 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8568 loss_recon=2.8568 loss_meanflow=0.0000 mean_model_t=0.4958 mean_corrupt_t=0.4958 wrong_frac=0.5040 init_acc_corrupt=0.4611 init_gold_top10=0.4901 init_gold_top100=0.4960
+step=102300 micro_steps=204600 elapsed=67.2s lr=3.000000e-04 loss_all=2.8337 acc_all=0.5888 loss_corrupt=2.8337 acc_corrupt=0.5888 corrupt_frac=1.0000 loss=2.8337 loss_recon=2.8337 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5025 init_acc_corrupt=0.4626 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=102400 micro_steps=204800 elapsed=67.2s lr=3.000000e-04 loss_all=2.8387 acc_all=0.5878 loss_corrupt=2.8387 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8387 loss_recon=2.8387 loss_meanflow=0.0000 mean_model_t=0.4958 mean_corrupt_t=0.4958 wrong_frac=0.5041 init_acc_corrupt=0.4613 init_gold_top10=0.4903 init_gold_top100=0.4960
+step=102500 micro_steps=205000 elapsed=67.5s lr=3.000000e-04 loss_all=2.7856 acc_all=0.5955 loss_corrupt=2.7856 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7856 loss_recon=2.7856 loss_meanflow=0.0000 mean_model_t=0.5052 mean_corrupt_t=0.5052 wrong_frac=0.4946 init_acc_corrupt=0.4716 init_gold_top10=0.4999 init_gold_top100=0.5055
+step=102600 micro_steps=205200 elapsed=67.8s lr=3.000000e-04 loss_all=2.8543 acc_all=0.5860 loss_corrupt=2.8543 acc_corrupt=0.5860 corrupt_frac=1.0000 loss=2.8543 loss_recon=2.8543 loss_meanflow=0.0000 mean_model_t=0.4957 mean_corrupt_t=0.4957 wrong_frac=0.5048 init_acc_corrupt=0.4601 init_gold_top10=0.4894 init_gold_top100=0.4953
+step=102700 micro_steps=205400 elapsed=67.8s lr=3.000000e-04 loss_all=2.8164 acc_all=0.5911 loss_corrupt=2.8164 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8164 loss_recon=2.8164 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4997 init_acc_corrupt=0.4662 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=102800 micro_steps=205600 elapsed=67.3s lr=3.000000e-04 loss_all=2.8022 acc_all=0.5927 loss_corrupt=2.8022 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8022 loss_recon=2.8022 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4996 init_acc_corrupt=0.4666 init_gold_top10=0.4950 init_gold_top100=0.5006
+step=102900 micro_steps=205800 elapsed=67.4s lr=3.000000e-04 loss_all=2.8154 acc_all=0.5911 loss_corrupt=2.8154 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8154 loss_recon=2.8154 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4996 init_acc_corrupt=0.4661 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=103000 micro_steps=206000 elapsed=67.5s lr=3.000000e-04 loss_all=2.7913 acc_all=0.5940 loss_corrupt=2.7913 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7913 loss_recon=2.7913 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4982 init_acc_corrupt=0.4681 init_gold_top10=0.4963 init_gold_top100=0.5019
+step=103100 micro_steps=206200 elapsed=68.8s lr=3.000000e-04 loss_all=2.8368 acc_all=0.5886 loss_corrupt=2.8368 acc_corrupt=0.5886 corrupt_frac=1.0000 loss=2.8368 loss_recon=2.8368 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5014 init_acc_corrupt=0.4634 init_gold_top10=0.4930 init_gold_top100=0.4987
+step=103200 micro_steps=206400 elapsed=66.7s lr=3.000000e-04 loss_all=2.7902 acc_all=0.5942 loss_corrupt=2.7902 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7902 loss_recon=2.7902 loss_meanflow=0.0000 mean_model_t=0.5040 mean_corrupt_t=0.5040 wrong_frac=0.4970 init_acc_corrupt=0.4695 init_gold_top10=0.4975 init_gold_top100=0.5032
+step=103300 micro_steps=206600 elapsed=67.6s lr=3.000000e-04 loss_all=2.8409 acc_all=0.5877 loss_corrupt=2.8409 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8409 loss_recon=2.8409 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5028 init_acc_corrupt=0.4621 init_gold_top10=0.4917 init_gold_top100=0.4975
+step=103400 micro_steps=206800 elapsed=102.9s lr=3.000000e-04 loss_all=2.8085 acc_all=0.5918 loss_corrupt=2.8085 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8085 loss_recon=2.8085 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4989 init_acc_corrupt=0.4668 init_gold_top10=0.4957 init_gold_top100=0.5013
+step=103500 micro_steps=207000 elapsed=107.8s lr=3.000000e-04 loss_all=2.7694 acc_all=0.5972 loss_corrupt=2.7694 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7694 loss_recon=2.7694 loss_meanflow=0.0000 mean_model_t=0.5051 mean_corrupt_t=0.5051 wrong_frac=0.4946 init_acc_corrupt=0.4711 init_gold_top10=0.4999 init_gold_top100=0.5055
+step=103600 micro_steps=207200 elapsed=79.7s lr=3.000000e-04 loss_all=2.8089 acc_all=0.5921 loss_corrupt=2.8089 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8089 loss_recon=2.8089 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4993 init_acc_corrupt=0.4660 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=103700 micro_steps=207400 elapsed=96.3s lr=3.000000e-04 loss_all=2.8101 acc_all=0.5918 loss_corrupt=2.8101 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8101 loss_recon=2.8101 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4992 init_acc_corrupt=0.4663 init_gold_top10=0.4950 init_gold_top100=0.5009
+step=103800 micro_steps=207600 elapsed=71.1s lr=3.000000e-04 loss_all=2.8198 acc_all=0.5906 loss_corrupt=2.8198 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8198 loss_recon=2.8198 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5008 init_acc_corrupt=0.4649 init_gold_top10=0.4938 init_gold_top100=0.4994
+step=103900 micro_steps=207800 elapsed=143.8s lr=3.000000e-04 loss_all=2.8295 acc_all=0.5893 loss_corrupt=2.8295 acc_corrupt=0.5893 corrupt_frac=1.0000 loss=2.8295 loss_recon=2.8295 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5018 init_acc_corrupt=0.4634 init_gold_top10=0.4927 init_gold_top100=0.4984
+step=104000 micro_steps=208000 elapsed=143.4s lr=3.000000e-04 loss_all=2.8102 acc_all=0.5922 loss_corrupt=2.8102 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8102 loss_recon=2.8102 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4987 init_acc_corrupt=0.4670 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=104100 micro_steps=208200 elapsed=113.0s lr=3.000000e-04 loss_all=2.7973 acc_all=0.5938 loss_corrupt=2.7973 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7973 loss_recon=2.7973 loss_meanflow=0.0000 mean_model_t=0.5036 mean_corrupt_t=0.5036 wrong_frac=0.4966 init_acc_corrupt=0.4686 init_gold_top10=0.4977 init_gold_top100=0.5035
+step=104200 micro_steps=208400 elapsed=102.8s lr=3.000000e-04 loss_all=2.8075 acc_all=0.5928 loss_corrupt=2.8075 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8075 loss_recon=2.8075 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4978 init_acc_corrupt=0.4678 init_gold_top10=0.4966 init_gold_top100=0.5023
+step=104300 micro_steps=208600 elapsed=103.2s lr=3.000000e-04 loss_all=2.7949 acc_all=0.5942 loss_corrupt=2.7949 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7949 loss_recon=2.7949 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4970 init_acc_corrupt=0.4689 init_gold_top10=0.4974 init_gold_top100=0.5031
+step=104400 micro_steps=208800 elapsed=94.7s lr=3.000000e-04 loss_all=2.7861 acc_all=0.5952 loss_corrupt=2.7861 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7861 loss_recon=2.7861 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4972 init_acc_corrupt=0.4696 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=104500 micro_steps=209000 elapsed=99.4s lr=3.000000e-04 loss_all=2.8604 acc_all=0.5852 loss_corrupt=2.8604 acc_corrupt=0.5852 corrupt_frac=1.0000 loss=2.8604 loss_recon=2.8604 loss_meanflow=0.0000 mean_model_t=0.4950 mean_corrupt_t=0.4950 wrong_frac=0.5049 init_acc_corrupt=0.4600 init_gold_top10=0.4894 init_gold_top100=0.4952
+step=104600 micro_steps=209200 elapsed=133.2s lr=3.000000e-04 loss_all=2.7976 acc_all=0.5939 loss_corrupt=2.7976 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7976 loss_recon=2.7976 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4969 init_acc_corrupt=0.4690 init_gold_top10=0.4976 init_gold_top100=0.5033
+step=104700 micro_steps=209400 elapsed=133.3s lr=3.000000e-04 loss_all=2.7968 acc_all=0.5933 loss_corrupt=2.7968 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7968 loss_recon=2.7968 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4978 init_acc_corrupt=0.4675 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=104800 micro_steps=209600 elapsed=133.0s lr=3.000000e-04 loss_all=2.7706 acc_all=0.5967 loss_corrupt=2.7706 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7706 loss_recon=2.7706 loss_meanflow=0.0000 mean_model_t=0.5051 mean_corrupt_t=0.5051 wrong_frac=0.4949 init_acc_corrupt=0.4718 init_gold_top10=0.4997 init_gold_top100=0.5053
+step=104900 micro_steps=209800 elapsed=133.1s lr=3.000000e-04 loss_all=2.8046 acc_all=0.5925 loss_corrupt=2.8046 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8046 loss_recon=2.8046 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4997 init_acc_corrupt=0.4665 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=105000 micro_steps=210000 elapsed=133.8s lr=3.000000e-04 loss_all=2.8279 acc_all=0.5895 loss_corrupt=2.8279 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8279 loss_recon=2.8279 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5008 init_acc_corrupt=0.4646 init_gold_top10=0.4937 init_gold_top100=0.4993
+step=105100 micro_steps=210200 elapsed=134.7s lr=3.000000e-04 loss_all=2.8248 acc_all=0.5898 loss_corrupt=2.8248 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8248 loss_recon=2.8248 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5016 init_acc_corrupt=0.4640 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=105200 micro_steps=210400 elapsed=132.9s lr=3.000000e-04 loss_all=2.8441 acc_all=0.5874 loss_corrupt=2.8441 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8441 loss_recon=2.8441 loss_meanflow=0.0000 mean_model_t=0.4966 mean_corrupt_t=0.4966 wrong_frac=0.5029 init_acc_corrupt=0.4622 init_gold_top10=0.4914 init_gold_top100=0.4972
+step=105300 micro_steps=210600 elapsed=132.9s lr=3.000000e-04 loss_all=2.8778 acc_all=0.5824 loss_corrupt=2.8778 acc_corrupt=0.5824 corrupt_frac=1.0000 loss=2.8778 loss_recon=2.8778 loss_meanflow=0.0000 mean_model_t=0.4912 mean_corrupt_t=0.4912 wrong_frac=0.5086 init_acc_corrupt=0.4563 init_gold_top10=0.4855 init_gold_top100=0.4915
+step=105400 micro_steps=210800 elapsed=132.6s lr=3.000000e-04 loss_all=2.8021 acc_all=0.5926 loss_corrupt=2.8021 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8021 loss_recon=2.8021 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4987 init_acc_corrupt=0.4672 init_gold_top10=0.4957 init_gold_top100=0.5013
+step=105500 micro_steps=211000 elapsed=132.9s lr=3.000000e-04 loss_all=2.8359 acc_all=0.5882 loss_corrupt=2.8359 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8359 loss_recon=2.8359 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5032 init_acc_corrupt=0.4620 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=105600 micro_steps=211200 elapsed=132.9s lr=3.000000e-04 loss_all=2.8209 acc_all=0.5902 loss_corrupt=2.8209 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8209 loss_recon=2.8209 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5011 init_acc_corrupt=0.4647 init_gold_top10=0.4933 init_gold_top100=0.4991
+step=105700 micro_steps=211400 elapsed=132.4s lr=3.000000e-04 loss_all=2.8195 acc_all=0.5911 loss_corrupt=2.8195 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8195 loss_recon=2.8195 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4995 init_acc_corrupt=0.4656 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=105800 micro_steps=211600 elapsed=131.6s lr=3.000000e-04 loss_all=2.8059 acc_all=0.5923 loss_corrupt=2.8059 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8059 loss_recon=2.8059 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4982 init_acc_corrupt=0.4668 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=105900 micro_steps=211800 elapsed=131.4s lr=3.000000e-04 loss_all=2.8141 acc_all=0.5914 loss_corrupt=2.8141 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8141 loss_recon=2.8141 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5000 init_acc_corrupt=0.4656 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=106000 micro_steps=212000 elapsed=131.6s lr=3.000000e-04 loss_all=2.8123 acc_all=0.5918 loss_corrupt=2.8123 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8123 loss_recon=2.8123 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5003 init_acc_corrupt=0.4650 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=106100 micro_steps=212200 elapsed=107.0s lr=3.000000e-04 loss_all=2.7948 acc_all=0.5937 loss_corrupt=2.7948 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7948 loss_recon=2.7948 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4983 init_acc_corrupt=0.4679 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=106200 micro_steps=212400 elapsed=63.7s lr=3.000000e-04 loss_all=2.8131 acc_all=0.5917 loss_corrupt=2.8131 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8131 loss_recon=2.8131 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4991 init_acc_corrupt=0.4667 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=106300 micro_steps=212600 elapsed=63.7s lr=3.000000e-04 loss_all=2.8212 acc_all=0.5902 loss_corrupt=2.8212 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8212 loss_recon=2.8212 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5012 init_acc_corrupt=0.4642 init_gold_top10=0.4932 init_gold_top100=0.4989
+step=106400 micro_steps=212800 elapsed=63.9s lr=3.000000e-04 loss_all=2.8146 acc_all=0.5915 loss_corrupt=2.8146 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8146 loss_recon=2.8146 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4992 init_acc_corrupt=0.4669 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=106500 micro_steps=213000 elapsed=63.8s lr=3.000000e-04 loss_all=2.8309 acc_all=0.5890 loss_corrupt=2.8309 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8309 loss_recon=2.8309 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5014 init_acc_corrupt=0.4634 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=106600 micro_steps=213200 elapsed=63.9s lr=3.000000e-04 loss_all=2.8466 acc_all=0.5873 loss_corrupt=2.8466 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8466 loss_recon=2.8466 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5043 init_acc_corrupt=0.4611 init_gold_top10=0.4902 init_gold_top100=0.4958
+step=106700 micro_steps=213400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8300 acc_all=0.5892 loss_corrupt=2.8300 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8300 loss_recon=2.8300 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5015 init_acc_corrupt=0.4635 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=106800 micro_steps=213600 elapsed=63.8s lr=3.000000e-04 loss_all=2.8347 acc_all=0.5887 loss_corrupt=2.8347 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8347 loss_recon=2.8347 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5019 init_acc_corrupt=0.4630 init_gold_top10=0.4925 init_gold_top100=0.4983
+step=106900 micro_steps=213800 elapsed=64.1s lr=3.000000e-04 loss_all=2.8680 acc_all=0.5845 loss_corrupt=2.8680 acc_corrupt=0.5845 corrupt_frac=1.0000 loss=2.8680 loss_recon=2.8680 loss_meanflow=0.0000 mean_model_t=0.4948 mean_corrupt_t=0.4948 wrong_frac=0.5055 init_acc_corrupt=0.4588 init_gold_top10=0.4887 init_gold_top100=0.4947
+step=107000 micro_steps=214000 elapsed=64.2s lr=3.000000e-04 loss_all=2.8151 acc_all=0.5914 loss_corrupt=2.8151 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8151 loss_recon=2.8151 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4989 init_acc_corrupt=0.4659 init_gold_top10=0.4954 init_gold_top100=0.5014
+step=107100 micro_steps=214200 elapsed=65.6s lr=3.000000e-04 loss_all=2.8139 acc_all=0.5915 loss_corrupt=2.8139 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8139 loss_recon=2.8139 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4996 init_acc_corrupt=0.4658 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=107200 micro_steps=214400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8090 acc_all=0.5924 loss_corrupt=2.8090 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8090 loss_recon=2.8090 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4986 init_acc_corrupt=0.4666 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=107300 micro_steps=214600 elapsed=63.8s lr=3.000000e-04 loss_all=2.8085 acc_all=0.5916 loss_corrupt=2.8085 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8085 loss_recon=2.8085 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5009 init_acc_corrupt=0.4649 init_gold_top10=0.4936 init_gold_top100=0.4992
+step=107400 micro_steps=214800 elapsed=63.6s lr=3.000000e-04 loss_all=2.8170 acc_all=0.5905 loss_corrupt=2.8170 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8170 loss_recon=2.8170 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5006 init_acc_corrupt=0.4645 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=107500 micro_steps=215000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8174 acc_all=0.5908 loss_corrupt=2.8174 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8174 loss_recon=2.8174 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.5002 init_acc_corrupt=0.4651 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=107600 micro_steps=215200 elapsed=63.6s lr=3.000000e-04 loss_all=2.8360 acc_all=0.5884 loss_corrupt=2.8360 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8360 loss_recon=2.8360 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5036 init_acc_corrupt=0.4620 init_gold_top10=0.4908 init_gold_top100=0.4965
+step=107700 micro_steps=215400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8439 acc_all=0.5877 loss_corrupt=2.8439 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8439 loss_recon=2.8439 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5026 init_acc_corrupt=0.4626 init_gold_top10=0.4916 init_gold_top100=0.4976
+step=107800 micro_steps=215600 elapsed=64.1s lr=3.000000e-04 loss_all=2.7928 acc_all=0.5941 loss_corrupt=2.7928 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7928 loss_recon=2.7928 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4977 init_acc_corrupt=0.4684 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=107900 micro_steps=215800 elapsed=64.2s lr=3.000000e-04 loss_all=2.8157 acc_all=0.5910 loss_corrupt=2.8157 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8157 loss_recon=2.8157 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5004 init_acc_corrupt=0.4656 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=108000 micro_steps=216000 elapsed=64.8s lr=3.000000e-04 loss_all=2.8303 acc_all=0.5893 loss_corrupt=2.8303 acc_corrupt=0.5893 corrupt_frac=1.0000 loss=2.8303 loss_recon=2.8303 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5015 init_acc_corrupt=0.4641 init_gold_top10=0.4927 init_gold_top100=0.4986
+step=108100 micro_steps=216200 elapsed=65.1s lr=3.000000e-04 loss_all=2.8257 acc_all=0.5900 loss_corrupt=2.8257 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8257 loss_recon=2.8257 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5012 init_acc_corrupt=0.4644 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=108200 micro_steps=216400 elapsed=63.6s lr=3.000000e-04 loss_all=2.8191 acc_all=0.5906 loss_corrupt=2.8191 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8191 loss_recon=2.8191 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5008 init_acc_corrupt=0.4643 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=108300 micro_steps=216600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8212 acc_all=0.5900 loss_corrupt=2.8212 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8212 loss_recon=2.8212 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5008 init_acc_corrupt=0.4645 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=108400 micro_steps=216800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7970 acc_all=0.5935 loss_corrupt=2.7970 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7970 loss_recon=2.7970 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4991 init_acc_corrupt=0.4670 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=108500 micro_steps=217000 elapsed=63.7s lr=3.000000e-04 loss_all=2.8020 acc_all=0.5933 loss_corrupt=2.8020 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.8020 loss_recon=2.8020 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4980 init_acc_corrupt=0.4668 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=108600 micro_steps=217200 elapsed=63.0s lr=3.000000e-04 loss_all=2.7775 acc_all=0.5959 loss_corrupt=2.7775 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7775 loss_recon=2.7775 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4968 init_acc_corrupt=0.4693 init_gold_top10=0.4978 init_gold_top100=0.5034
+step=108700 micro_steps=217400 elapsed=64.1s lr=3.000000e-04 loss_all=2.7698 acc_all=0.5972 loss_corrupt=2.7698 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7698 loss_recon=2.7698 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4954 init_acc_corrupt=0.4711 init_gold_top10=0.4991 init_gold_top100=0.5047
+step=108800 micro_steps=217600 elapsed=64.2s lr=3.000000e-04 loss_all=2.7819 acc_all=0.5952 loss_corrupt=2.7819 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7819 loss_recon=2.7819 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4968 init_acc_corrupt=0.4685 init_gold_top10=0.4977 init_gold_top100=0.5034
+step=108900 micro_steps=217800 elapsed=64.2s lr=3.000000e-04 loss_all=2.8276 acc_all=0.5900 loss_corrupt=2.8276 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8276 loss_recon=2.8276 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5010 init_acc_corrupt=0.4644 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=109000 micro_steps=218000 elapsed=64.0s lr=3.000000e-04 loss_all=2.8292 acc_all=0.5896 loss_corrupt=2.8292 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8292 loss_recon=2.8292 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5020 init_acc_corrupt=0.4630 init_gold_top10=0.4922 init_gold_top100=0.4981
+step=109100 micro_steps=218200 elapsed=65.4s lr=3.000000e-04 loss_all=2.8024 acc_all=0.5925 loss_corrupt=2.8024 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8024 loss_recon=2.8024 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4995 init_acc_corrupt=0.4663 init_gold_top10=0.4949 init_gold_top100=0.5006
+step=109200 micro_steps=218400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8274 acc_all=0.5895 loss_corrupt=2.8274 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8274 loss_recon=2.8274 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5017 init_acc_corrupt=0.4633 init_gold_top10=0.4927 init_gold_top100=0.4984
+step=109300 micro_steps=218600 elapsed=64.0s lr=3.000000e-04 loss_all=2.7956 acc_all=0.5939 loss_corrupt=2.7956 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7956 loss_recon=2.7956 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4971 init_acc_corrupt=0.4686 init_gold_top10=0.4974 init_gold_top100=0.5030
+step=109400 micro_steps=218800 elapsed=63.7s lr=3.000000e-04 loss_all=2.7930 acc_all=0.5945 loss_corrupt=2.7930 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7930 loss_recon=2.7930 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4975 init_acc_corrupt=0.4688 init_gold_top10=0.4971 init_gold_top100=0.5026
+step=109500 micro_steps=219000 elapsed=63.7s lr=3.000000e-04 loss_all=2.8246 acc_all=0.5903 loss_corrupt=2.8246 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8246 loss_recon=2.8246 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5006 init_acc_corrupt=0.4641 init_gold_top10=0.4937 init_gold_top100=0.4995
+step=109600 micro_steps=219200 elapsed=63.7s lr=3.000000e-04 loss_all=2.8098 acc_all=0.5918 loss_corrupt=2.8098 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8098 loss_recon=2.8098 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4996 init_acc_corrupt=0.4656 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=109700 micro_steps=219400 elapsed=63.7s lr=3.000000e-04 loss_all=2.7904 acc_all=0.5946 loss_corrupt=2.7904 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7904 loss_recon=2.7904 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4969 init_acc_corrupt=0.4688 init_gold_top10=0.4976 init_gold_top100=0.5034
+step=109800 micro_steps=219600 elapsed=63.8s lr=3.000000e-04 loss_all=2.8174 acc_all=0.5909 loss_corrupt=2.8174 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8174 loss_recon=2.8174 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4996 init_acc_corrupt=0.4654 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=109900 micro_steps=219800 elapsed=63.7s lr=3.000000e-04 loss_all=2.7805 acc_all=0.5957 loss_corrupt=2.7805 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7805 loss_recon=2.7805 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4967 init_acc_corrupt=0.4697 init_gold_top10=0.4978 init_gold_top100=0.5035
+step=110000 micro_steps=220000 elapsed=63.7s lr=3.000000e-04 loss_all=2.8042 acc_all=0.5923 loss_corrupt=2.8042 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8042 loss_recon=2.8042 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4990 init_acc_corrupt=0.4667 init_gold_top10=0.4956 init_gold_top100=0.5012
+step=110100 micro_steps=220200 elapsed=65.2s lr=3.000000e-04 loss_all=2.8085 acc_all=0.5923 loss_corrupt=2.8085 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8085 loss_recon=2.8085 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4992 init_acc_corrupt=0.4667 init_gold_top10=0.4955 init_gold_top100=0.5010
+step=110200 micro_steps=220400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8099 acc_all=0.5917 loss_corrupt=2.8099 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8099 loss_recon=2.8099 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4995 init_acc_corrupt=0.4656 init_gold_top10=0.4949 init_gold_top100=0.5006
+step=110300 micro_steps=220600 elapsed=63.9s lr=3.000000e-04 loss_all=2.8757 acc_all=0.5838 loss_corrupt=2.8757 acc_corrupt=0.5838 corrupt_frac=1.0000 loss=2.8757 loss_recon=2.8757 loss_meanflow=0.0000 mean_model_t=0.4937 mean_corrupt_t=0.4937 wrong_frac=0.5064 init_acc_corrupt=0.4586 init_gold_top10=0.4878 init_gold_top100=0.4937
+step=110400 micro_steps=220800 elapsed=63.9s lr=3.000000e-04 loss_all=2.7891 acc_all=0.5942 loss_corrupt=2.7891 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7891 loss_recon=2.7891 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4972 init_acc_corrupt=0.4689 init_gold_top10=0.4974 init_gold_top100=0.5030
+step=110500 micro_steps=221000 elapsed=63.9s lr=3.000000e-04 loss_all=2.8123 acc_all=0.5914 loss_corrupt=2.8123 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8123 loss_recon=2.8123 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4997 init_acc_corrupt=0.4658 init_gold_top10=0.4947 init_gold_top100=0.5005
+step=110600 micro_steps=221200 elapsed=64.0s lr=3.000000e-04 loss_all=2.8450 acc_all=0.5874 loss_corrupt=2.8450 acc_corrupt=0.5874 corrupt_frac=1.0000 loss=2.8450 loss_recon=2.8450 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5034 init_acc_corrupt=0.4617 init_gold_top10=0.4909 init_gold_top100=0.4966
+step=110700 micro_steps=221400 elapsed=64.2s lr=3.000000e-04 loss_all=2.8360 acc_all=0.5886 loss_corrupt=2.8360 acc_corrupt=0.5886 corrupt_frac=1.0000 loss=2.8360 loss_recon=2.8360 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5031 init_acc_corrupt=0.4625 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=110800 micro_steps=221600 elapsed=64.3s lr=3.000000e-04 loss_all=2.8136 acc_all=0.5914 loss_corrupt=2.8136 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8136 loss_recon=2.8136 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.5002 init_acc_corrupt=0.4653 init_gold_top10=0.4943 init_gold_top100=0.5000
+step=110900 micro_steps=221800 elapsed=64.3s lr=3.000000e-04 loss_all=2.8399 acc_all=0.5884 loss_corrupt=2.8399 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8399 loss_recon=2.8399 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5021 init_acc_corrupt=0.4629 init_gold_top10=0.4922 init_gold_top100=0.4980
+step=111000 micro_steps=222000 elapsed=65.1s lr=3.000000e-04 loss_all=2.8562 acc_all=0.5862 loss_corrupt=2.8562 acc_corrupt=0.5862 corrupt_frac=1.0000 loss=2.8562 loss_recon=2.8562 loss_meanflow=0.0000 mean_model_t=0.4958 mean_corrupt_t=0.4958 wrong_frac=0.5042 init_acc_corrupt=0.4602 init_gold_top10=0.4900 init_gold_top100=0.4959
+step=111100 micro_steps=222200 elapsed=66.4s lr=3.000000e-04 loss_all=2.7719 acc_all=0.5973 loss_corrupt=2.7719 acc_corrupt=0.5973 corrupt_frac=1.0000 loss=2.7719 loss_recon=2.7719 loss_meanflow=0.0000 mean_model_t=0.5061 mean_corrupt_t=0.5061 wrong_frac=0.4936 init_acc_corrupt=0.4723 init_gold_top10=0.5009 init_gold_top100=0.5066
+step=111200 micro_steps=222400 elapsed=64.0s lr=3.000000e-04 loss_all=2.7805 acc_all=0.5958 loss_corrupt=2.7805 acc_corrupt=0.5958 corrupt_frac=1.0000 loss=2.7805 loss_recon=2.7805 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4962 init_acc_corrupt=0.4690 init_gold_top10=0.4983 init_gold_top100=0.5040
+step=111300 micro_steps=222600 elapsed=64.3s lr=3.000000e-04 loss_all=2.8099 acc_all=0.5919 loss_corrupt=2.8099 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8099 loss_recon=2.8099 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4993 init_acc_corrupt=0.4659 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=111400 micro_steps=222800 elapsed=64.2s lr=3.000000e-04 loss_all=2.7973 acc_all=0.5935 loss_corrupt=2.7973 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7973 loss_recon=2.7973 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4978 init_acc_corrupt=0.4676 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=111500 micro_steps=223000 elapsed=63.8s lr=3.000000e-04 loss_all=2.7861 acc_all=0.5951 loss_corrupt=2.7861 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7861 loss_recon=2.7861 loss_meanflow=0.0000 mean_model_t=0.5039 mean_corrupt_t=0.5039 wrong_frac=0.4961 init_acc_corrupt=0.4696 init_gold_top10=0.4984 init_gold_top100=0.5041
+step=111600 micro_steps=223200 elapsed=62.8s lr=3.000000e-04 loss_all=2.8192 acc_all=0.5909 loss_corrupt=2.8192 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8192 loss_recon=2.8192 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5009 init_acc_corrupt=0.4645 init_gold_top10=0.4936 init_gold_top100=0.4992
+step=111700 micro_steps=223400 elapsed=64.6s lr=3.000000e-04 loss_all=2.8750 acc_all=0.5834 loss_corrupt=2.8750 acc_corrupt=0.5834 corrupt_frac=1.0000 loss=2.8750 loss_recon=2.8750 loss_meanflow=0.0000 mean_model_t=0.4925 mean_corrupt_t=0.4925 wrong_frac=0.5075 init_acc_corrupt=0.4575 init_gold_top10=0.4866 init_gold_top100=0.4925
+step=111800 micro_steps=223600 elapsed=63.9s lr=3.000000e-04 loss_all=2.8226 acc_all=0.5907 loss_corrupt=2.8226 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8226 loss_recon=2.8226 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5003 init_acc_corrupt=0.4661 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=111900 micro_steps=223800 elapsed=64.3s lr=3.000000e-04 loss_all=2.8054 acc_all=0.5928 loss_corrupt=2.8054 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8054 loss_recon=2.8054 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4994 init_acc_corrupt=0.4660 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=112000 micro_steps=224000 elapsed=64.6s lr=3.000000e-04 loss_all=2.8306 acc_all=0.5888 loss_corrupt=2.8306 acc_corrupt=0.5888 corrupt_frac=1.0000 loss=2.8306 loss_recon=2.8306 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5027 init_acc_corrupt=0.4622 init_gold_top10=0.4917 init_gold_top100=0.4974
+step=112100 micro_steps=224200 elapsed=65.8s lr=3.000000e-04 loss_all=2.8317 acc_all=0.5892 loss_corrupt=2.8317 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8317 loss_recon=2.8317 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=112200 micro_steps=224400 elapsed=65.0s lr=3.000000e-04 loss_all=2.8197 acc_all=0.5908 loss_corrupt=2.8197 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8197 loss_recon=2.8197 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4995 init_acc_corrupt=0.4654 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=112300 micro_steps=224600 elapsed=64.8s lr=3.000000e-04 loss_all=2.8171 acc_all=0.5911 loss_corrupt=2.8171 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8171 loss_recon=2.8171 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5003 init_acc_corrupt=0.4652 init_gold_top10=0.4939 init_gold_top100=0.4998
+step=112400 micro_steps=224800 elapsed=64.4s lr=3.000000e-04 loss_all=2.8208 acc_all=0.5905 loss_corrupt=2.8208 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8208 loss_recon=2.8208 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5011 init_acc_corrupt=0.4646 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=112500 micro_steps=225000 elapsed=63.8s lr=3.000000e-04 loss_all=2.8040 acc_all=0.5923 loss_corrupt=2.8040 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8040 loss_recon=2.8040 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4998 init_acc_corrupt=0.4660 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=112600 micro_steps=225200 elapsed=64.0s lr=3.000000e-04 loss_all=2.8176 acc_all=0.5909 loss_corrupt=2.8176 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8176 loss_recon=2.8176 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5006 init_acc_corrupt=0.4652 init_gold_top10=0.4937 init_gold_top100=0.4995
+step=112700 micro_steps=225400 elapsed=64.2s lr=3.000000e-04 loss_all=2.8240 acc_all=0.5896 loss_corrupt=2.8240 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8240 loss_recon=2.8240 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5022 init_acc_corrupt=0.4633 init_gold_top10=0.4921 init_gold_top100=0.4979
+step=112800 micro_steps=225600 elapsed=64.2s lr=3.000000e-04 loss_all=2.7961 acc_all=0.5933 loss_corrupt=2.7961 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7961 loss_recon=2.7961 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4995 init_acc_corrupt=0.4659 init_gold_top10=0.4952 init_gold_top100=0.5007
+step=112900 micro_steps=225800 elapsed=64.0s lr=3.000000e-04 loss_all=2.8381 acc_all=0.5882 loss_corrupt=2.8381 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8381 loss_recon=2.8381 loss_meanflow=0.0000 mean_model_t=0.4962 mean_corrupt_t=0.4962 wrong_frac=0.5035 init_acc_corrupt=0.4616 init_gold_top10=0.4909 init_gold_top100=0.4967
+step=113000 micro_steps=226000 elapsed=64.4s lr=3.000000e-04 loss_all=2.8067 acc_all=0.5924 loss_corrupt=2.8067 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8067 loss_recon=2.8067 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.4999 init_acc_corrupt=0.4660 init_gold_top10=0.4945 init_gold_top100=0.5003
+step=113100 micro_steps=226200 elapsed=66.3s lr=3.000000e-04 loss_all=2.8178 acc_all=0.5906 loss_corrupt=2.8178 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8178 loss_recon=2.8178 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5013 init_acc_corrupt=0.4638 init_gold_top10=0.4932 init_gold_top100=0.4989
+step=113200 micro_steps=226400 elapsed=63.7s lr=3.000000e-04 loss_all=2.8002 acc_all=0.5933 loss_corrupt=2.8002 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.8002 loss_recon=2.8002 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4974 init_acc_corrupt=0.4680 init_gold_top10=0.4971 init_gold_top100=0.5028
+step=113300 micro_steps=226600 elapsed=63.3s lr=3.000000e-04 loss_all=2.8363 acc_all=0.5882 loss_corrupt=2.8363 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8363 loss_recon=2.8363 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5032 init_acc_corrupt=0.4619 init_gold_top10=0.4910 init_gold_top100=0.4968
+step=113400 micro_steps=226800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8480 acc_all=0.5871 loss_corrupt=2.8480 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8480 loss_recon=2.8480 loss_meanflow=0.0000 mean_model_t=0.4947 mean_corrupt_t=0.4947 wrong_frac=0.5047 init_acc_corrupt=0.4601 init_gold_top10=0.4894 init_gold_top100=0.4955
+step=113500 micro_steps=227000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7805 acc_all=0.5953 loss_corrupt=2.7805 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7805 loss_recon=2.7805 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4972 init_acc_corrupt=0.4687 init_gold_top10=0.4973 init_gold_top100=0.5030
+step=113600 micro_steps=227200 elapsed=63.6s lr=3.000000e-04 loss_all=2.7957 acc_all=0.5941 loss_corrupt=2.7957 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7957 loss_recon=2.7957 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4978 init_acc_corrupt=0.4679 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=113700 micro_steps=227400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8270 acc_all=0.5895 loss_corrupt=2.8270 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8270 loss_recon=2.8270 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5021 init_acc_corrupt=0.4631 init_gold_top10=0.4921 init_gold_top100=0.4980
+step=113800 micro_steps=227600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8126 acc_all=0.5914 loss_corrupt=2.8126 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8126 loss_recon=2.8126 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5008 init_acc_corrupt=0.4646 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=113900 micro_steps=227800 elapsed=63.7s lr=3.000000e-04 loss_all=2.8138 acc_all=0.5911 loss_corrupt=2.8138 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8138 loss_recon=2.8138 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5015 init_acc_corrupt=0.4640 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=114000 micro_steps=228000 elapsed=63.6s lr=3.000000e-04 loss_all=2.8050 acc_all=0.5921 loss_corrupt=2.8050 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8050 loss_recon=2.8050 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4998 init_acc_corrupt=0.4658 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=114100 micro_steps=228200 elapsed=65.1s lr=3.000000e-04 loss_all=2.8270 acc_all=0.5892 loss_corrupt=2.8270 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8270 loss_recon=2.8270 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5023 init_acc_corrupt=0.4626 init_gold_top10=0.4922 init_gold_top100=0.4979
+step=114200 micro_steps=228400 elapsed=63.8s lr=3.000000e-04 loss_all=2.7853 acc_all=0.5949 loss_corrupt=2.7853 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7853 loss_recon=2.7853 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4974 init_acc_corrupt=0.4682 init_gold_top10=0.4971 init_gold_top100=0.5028
+step=114300 micro_steps=228600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8643 acc_all=0.5848 loss_corrupt=2.8643 acc_corrupt=0.5848 corrupt_frac=1.0000 loss=2.8643 loss_recon=2.8643 loss_meanflow=0.0000 mean_model_t=0.4938 mean_corrupt_t=0.4938 wrong_frac=0.5064 init_acc_corrupt=0.4589 init_gold_top10=0.4877 init_gold_top100=0.4937
+step=114400 micro_steps=228800 elapsed=63.6s lr=3.000000e-04 loss_all=2.8244 acc_all=0.5897 loss_corrupt=2.8244 acc_corrupt=0.5897 corrupt_frac=1.0000 loss=2.8244 loss_recon=2.8244 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5012 init_acc_corrupt=0.4637 init_gold_top10=0.4931 init_gold_top100=0.4990
+step=114500 micro_steps=229000 elapsed=62.6s lr=3.000000e-04 loss_all=2.8083 acc_all=0.5928 loss_corrupt=2.8083 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8083 loss_recon=2.8083 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4983 init_acc_corrupt=0.4673 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=114600 micro_steps=229200 elapsed=63.5s lr=3.000000e-04 loss_all=2.8151 acc_all=0.5912 loss_corrupt=2.8151 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8151 loss_recon=2.8151 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5007 init_acc_corrupt=0.4647 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=114700 micro_steps=229400 elapsed=63.7s lr=3.000000e-04 loss_all=2.8132 acc_all=0.5915 loss_corrupt=2.8132 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8132 loss_recon=2.8132 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5005 init_acc_corrupt=0.4647 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=114800 micro_steps=229600 elapsed=63.7s lr=3.000000e-04 loss_all=2.8302 acc_all=0.5894 loss_corrupt=2.8302 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8302 loss_recon=2.8302 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=114900 micro_steps=229800 elapsed=63.8s lr=3.000000e-04 loss_all=2.8074 acc_all=0.5922 loss_corrupt=2.8074 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8074 loss_recon=2.8074 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4995 init_acc_corrupt=0.4656 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=115000 micro_steps=230000 elapsed=63.6s lr=3.000000e-04 loss_all=2.8263 acc_all=0.5899 loss_corrupt=2.8263 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8263 loss_recon=2.8263 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5013 init_acc_corrupt=0.4643 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=115100 micro_steps=230200 elapsed=65.1s lr=3.000000e-04 loss_all=2.8323 acc_all=0.5890 loss_corrupt=2.8323 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8323 loss_recon=2.8323 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5029 init_acc_corrupt=0.4626 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=115200 micro_steps=230400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8154 acc_all=0.5915 loss_corrupt=2.8154 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8154 loss_recon=2.8154 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4993 init_acc_corrupt=0.4657 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=115300 micro_steps=230600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8210 acc_all=0.5907 loss_corrupt=2.8210 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8210 loss_recon=2.8210 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5010 init_acc_corrupt=0.4647 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=115400 micro_steps=230800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7992 acc_all=0.5930 loss_corrupt=2.7992 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.7992 loss_recon=2.7992 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4987 init_acc_corrupt=0.4667 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=115500 micro_steps=231000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8018 acc_all=0.5927 loss_corrupt=2.8018 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8018 loss_recon=2.8018 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5004 init_acc_corrupt=0.4652 init_gold_top10=0.4941 init_gold_top100=0.4997
+step=115600 micro_steps=231200 elapsed=63.7s lr=3.000000e-04 loss_all=2.8190 acc_all=0.5911 loss_corrupt=2.8190 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8190 loss_recon=2.8190 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5009 init_acc_corrupt=0.4647 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=115700 micro_steps=231400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8038 acc_all=0.5929 loss_corrupt=2.8038 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.8038 loss_recon=2.8038 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4981 init_acc_corrupt=0.4680 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=115800 micro_steps=231600 elapsed=63.0s lr=3.000000e-04 loss_all=2.8448 acc_all=0.5871 loss_corrupt=2.8448 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8448 loss_recon=2.8448 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5037 init_acc_corrupt=0.4612 init_gold_top10=0.4905 init_gold_top100=0.4964
+step=115900 micro_steps=231800 elapsed=66.1s lr=3.000000e-04 loss_all=2.8206 acc_all=0.5907 loss_corrupt=2.8206 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8206 loss_recon=2.8206 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5016 init_acc_corrupt=0.4639 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=116000 micro_steps=232000 elapsed=87.3s lr=3.000000e-04 loss_all=2.8092 acc_all=0.5920 loss_corrupt=2.8092 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8092 loss_recon=2.8092 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4989 init_acc_corrupt=0.4664 init_gold_top10=0.4954 init_gold_top100=0.5012
+step=116100 micro_steps=232200 elapsed=70.6s lr=3.000000e-04 loss_all=2.8272 acc_all=0.5896 loss_corrupt=2.8272 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8272 loss_recon=2.8272 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5024 init_acc_corrupt=0.4628 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=116200 micro_steps=232400 elapsed=63.3s lr=3.000000e-04 loss_all=2.8142 acc_all=0.5911 loss_corrupt=2.8142 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8142 loss_recon=2.8142 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5010 init_acc_corrupt=0.4644 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=116300 micro_steps=232600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8070 acc_all=0.5918 loss_corrupt=2.8070 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8070 loss_recon=2.8070 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5008 init_acc_corrupt=0.4653 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=116400 micro_steps=232800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8575 acc_all=0.5861 loss_corrupt=2.8575 acc_corrupt=0.5861 corrupt_frac=1.0000 loss=2.8575 loss_recon=2.8575 loss_meanflow=0.0000 mean_model_t=0.4953 mean_corrupt_t=0.4953 wrong_frac=0.5048 init_acc_corrupt=0.4603 init_gold_top10=0.4895 init_gold_top100=0.4953
+step=116500 micro_steps=233000 elapsed=63.3s lr=3.000000e-04 loss_all=2.8223 acc_all=0.5903 loss_corrupt=2.8223 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8223 loss_recon=2.8223 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5014 init_acc_corrupt=0.4639 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=116600 micro_steps=233200 elapsed=63.4s lr=3.000000e-04 loss_all=2.8081 acc_all=0.5918 loss_corrupt=2.8081 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8081 loss_recon=2.8081 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4997 init_acc_corrupt=0.4656 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=116700 micro_steps=233400 elapsed=63.2s lr=3.000000e-04 loss_all=2.8181 acc_all=0.5906 loss_corrupt=2.8181 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8181 loss_recon=2.8181 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5017 init_acc_corrupt=0.4641 init_gold_top10=0.4928 init_gold_top100=0.4984
+step=116800 micro_steps=233600 elapsed=63.0s lr=3.000000e-04 loss_all=2.8070 acc_all=0.5923 loss_corrupt=2.8070 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8070 loss_recon=2.8070 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5002 init_acc_corrupt=0.4656 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=116900 micro_steps=233800 elapsed=62.0s lr=3.000000e-04 loss_all=2.8116 acc_all=0.5923 loss_corrupt=2.8116 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8116 loss_recon=2.8116 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4992 init_acc_corrupt=0.4658 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=117000 micro_steps=234000 elapsed=62.4s lr=3.000000e-04 loss_all=2.8352 acc_all=0.5887 loss_corrupt=2.8352 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8352 loss_recon=2.8352 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5030 init_acc_corrupt=0.4623 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=117100 micro_steps=234200 elapsed=64.0s lr=3.000000e-04 loss_all=2.8355 acc_all=0.5889 loss_corrupt=2.8355 acc_corrupt=0.5889 corrupt_frac=1.0000 loss=2.8355 loss_recon=2.8355 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5023 init_acc_corrupt=0.4628 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=117200 micro_steps=234400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7774 acc_all=0.5962 loss_corrupt=2.7774 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7774 loss_recon=2.7774 loss_meanflow=0.0000 mean_model_t=0.5036 mean_corrupt_t=0.5036 wrong_frac=0.4955 init_acc_corrupt=0.4704 init_gold_top10=0.4989 init_gold_top100=0.5046
+step=117300 micro_steps=234600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8310 acc_all=0.5884 loss_corrupt=2.8310 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8310 loss_recon=2.8310 loss_meanflow=0.0000 mean_model_t=0.4954 mean_corrupt_t=0.4954 wrong_frac=0.5046 init_acc_corrupt=0.4606 init_gold_top10=0.4900 init_gold_top100=0.4956
+step=117400 micro_steps=234800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8214 acc_all=0.5903 loss_corrupt=2.8214 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8214 loss_recon=2.8214 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5011 init_acc_corrupt=0.4639 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=117500 micro_steps=235000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8136 acc_all=0.5917 loss_corrupt=2.8136 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8136 loss_recon=2.8136 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5008 init_acc_corrupt=0.4650 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=117600 micro_steps=235200 elapsed=63.4s lr=3.000000e-04 loss_all=2.7968 acc_all=0.5933 loss_corrupt=2.7968 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7968 loss_recon=2.7968 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4988 init_acc_corrupt=0.4668 init_gold_top10=0.4955 init_gold_top100=0.5013
+step=117700 micro_steps=235400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8122 acc_all=0.5915 loss_corrupt=2.8122 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8122 loss_recon=2.8122 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5004 init_acc_corrupt=0.4650 init_gold_top10=0.4940 init_gold_top100=0.4996
+step=117800 micro_steps=235600 elapsed=63.2s lr=3.000000e-04 loss_all=2.7855 acc_all=0.5948 loss_corrupt=2.7855 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7855 loss_recon=2.7855 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4976 init_acc_corrupt=0.4680 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=117900 micro_steps=235800 elapsed=62.9s lr=3.000000e-04 loss_all=2.8119 acc_all=0.5915 loss_corrupt=2.8119 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8119 loss_recon=2.8119 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5007 init_acc_corrupt=0.4647 init_gold_top10=0.4936 init_gold_top100=0.4995
+step=118000 micro_steps=236000 elapsed=62.9s lr=3.000000e-04 loss_all=2.8200 acc_all=0.5906 loss_corrupt=2.8200 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8200 loss_recon=2.8200 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5014 init_acc_corrupt=0.4642 init_gold_top10=0.4930 init_gold_top100=0.4987
+step=118100 micro_steps=236200 elapsed=64.1s lr=3.000000e-04 loss_all=2.8245 acc_all=0.5901 loss_corrupt=2.8245 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8245 loss_recon=2.8245 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=118200 micro_steps=236400 elapsed=62.8s lr=3.000000e-04 loss_all=2.8325 acc_all=0.5891 loss_corrupt=2.8325 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8325 loss_recon=2.8325 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5020 init_acc_corrupt=0.4629 init_gold_top10=0.4922 init_gold_top100=0.4981
+step=118300 micro_steps=236600 elapsed=62.8s lr=3.000000e-04 loss_all=2.8350 acc_all=0.5887 loss_corrupt=2.8350 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8350 loss_recon=2.8350 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5033 init_acc_corrupt=0.4619 init_gold_top10=0.4909 init_gold_top100=0.4968
+step=118400 micro_steps=236800 elapsed=62.9s lr=3.000000e-04 loss_all=2.8164 acc_all=0.5909 loss_corrupt=2.8164 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8164 loss_recon=2.8164 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5011 init_acc_corrupt=0.4641 init_gold_top10=0.4933 init_gold_top100=0.4989
+step=118500 micro_steps=237000 elapsed=62.8s lr=3.000000e-04 loss_all=2.7703 acc_all=0.5968 loss_corrupt=2.7703 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7703 loss_recon=2.7703 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4955 init_acc_corrupt=0.4707 init_gold_top10=0.4989 init_gold_top100=0.5046
+step=118600 micro_steps=237200 elapsed=62.7s lr=3.000000e-04 loss_all=2.7927 acc_all=0.5940 loss_corrupt=2.7927 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7927 loss_recon=2.7927 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4983 init_acc_corrupt=0.4674 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=118700 micro_steps=237400 elapsed=62.8s lr=3.000000e-04 loss_all=2.8215 acc_all=0.5906 loss_corrupt=2.8215 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8215 loss_recon=2.8215 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4996 init_acc_corrupt=0.4654 init_gold_top10=0.4947 init_gold_top100=0.5005
+step=118800 micro_steps=237600 elapsed=62.8s lr=3.000000e-04 loss_all=2.8200 acc_all=0.5903 loss_corrupt=2.8200 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8200 loss_recon=2.8200 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5021 init_acc_corrupt=0.4630 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=118900 micro_steps=237800 elapsed=62.9s lr=3.000000e-04 loss_all=2.8222 acc_all=0.5898 loss_corrupt=2.8222 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8222 loss_recon=2.8222 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5020 init_acc_corrupt=0.4631 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=119000 micro_steps=238000 elapsed=62.8s lr=3.000000e-04 loss_all=2.7670 acc_all=0.5974 loss_corrupt=2.7670 acc_corrupt=0.5974 corrupt_frac=1.0000 loss=2.7670 loss_recon=2.7670 loss_meanflow=0.0000 mean_model_t=0.5046 mean_corrupt_t=0.5046 wrong_frac=0.4956 init_acc_corrupt=0.4707 init_gold_top10=0.4990 init_gold_top100=0.5046
+step=119100 micro_steps=238200 elapsed=64.4s lr=3.000000e-04 loss_all=2.8258 acc_all=0.5897 loss_corrupt=2.8258 acc_corrupt=0.5897 corrupt_frac=1.0000 loss=2.8258 loss_recon=2.8258 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5017 init_acc_corrupt=0.4635 init_gold_top10=0.4926 init_gold_top100=0.4983
+step=119200 micro_steps=238400 elapsed=59.9s lr=3.000000e-04 loss_all=2.7900 acc_all=0.5945 loss_corrupt=2.7900 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7900 loss_recon=2.7900 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4977 init_acc_corrupt=0.4676 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=119300 micro_steps=238600 elapsed=63.6s lr=3.000000e-04 loss_all=2.7888 acc_all=0.5949 loss_corrupt=2.7888 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7888 loss_recon=2.7888 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4976 init_acc_corrupt=0.4680 init_gold_top10=0.4969 init_gold_top100=0.5024
+step=119400 micro_steps=238800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7756 acc_all=0.5964 loss_corrupt=2.7756 acc_corrupt=0.5964 corrupt_frac=1.0000 loss=2.7756 loss_recon=2.7756 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4965 init_acc_corrupt=0.4694 init_gold_top10=0.4982 init_gold_top100=0.5037
+step=119500 micro_steps=239000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8148 acc_all=0.5911 loss_corrupt=2.8148 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8148 loss_recon=2.8148 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5012 init_acc_corrupt=0.4640 init_gold_top10=0.4931 init_gold_top100=0.4990
+step=119600 micro_steps=239200 elapsed=63.5s lr=3.000000e-04 loss_all=2.8118 acc_all=0.5916 loss_corrupt=2.8118 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8118 loss_recon=2.8118 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5005 init_acc_corrupt=0.4644 init_gold_top10=0.4937 init_gold_top100=0.4996
+step=119700 micro_steps=239400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8191 acc_all=0.5907 loss_corrupt=2.8191 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8191 loss_recon=2.8191 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5008 init_acc_corrupt=0.4647 init_gold_top10=0.4933 init_gold_top100=0.4992
+step=119800 micro_steps=239600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8211 acc_all=0.5904 loss_corrupt=2.8211 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8211 loss_recon=2.8211 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5015 init_acc_corrupt=0.4635 init_gold_top10=0.4930 init_gold_top100=0.4987
+step=119900 micro_steps=239800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8015 acc_all=0.5927 loss_corrupt=2.8015 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8015 loss_recon=2.8015 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4999 init_acc_corrupt=0.4662 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=120000 micro_steps=240000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8202 acc_all=0.5908 loss_corrupt=2.8202 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8202 loss_recon=2.8202 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5005 init_acc_corrupt=0.4652 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=120100 micro_steps=240200 elapsed=66.6s lr=3.000000e-04 loss_all=2.8093 acc_all=0.5920 loss_corrupt=2.8093 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8093 loss_recon=2.8093 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5009 init_acc_corrupt=0.4647 init_gold_top10=0.4936 init_gold_top100=0.4992
+step=120200 micro_steps=240400 elapsed=63.3s lr=3.000000e-04 loss_all=2.7997 acc_all=0.5932 loss_corrupt=2.7997 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7997 loss_recon=2.7997 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4983 init_acc_corrupt=0.4673 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=120300 micro_steps=240600 elapsed=63.2s lr=3.000000e-04 loss_all=2.8198 acc_all=0.5909 loss_corrupt=2.8198 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8198 loss_recon=2.8198 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5000 init_acc_corrupt=0.4652 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=120400 micro_steps=240800 elapsed=62.8s lr=3.000000e-04 loss_all=2.8179 acc_all=0.5909 loss_corrupt=2.8179 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8179 loss_recon=2.8179 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5005 init_acc_corrupt=0.4647 init_gold_top10=0.4938 init_gold_top100=0.4996
+step=120500 micro_steps=241000 elapsed=62.7s lr=3.000000e-04 loss_all=2.8277 acc_all=0.5892 loss_corrupt=2.8277 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8277 loss_recon=2.8277 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5032 init_acc_corrupt=0.4616 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=120600 micro_steps=241200 elapsed=62.7s lr=3.000000e-04 loss_all=2.8085 acc_all=0.5920 loss_corrupt=2.8085 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8085 loss_recon=2.8085 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4996 init_acc_corrupt=0.4659 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=120700 micro_steps=241400 elapsed=63.1s lr=3.000000e-04 loss_all=2.8027 acc_all=0.5928 loss_corrupt=2.8027 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8027 loss_recon=2.8027 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4988 init_acc_corrupt=0.4665 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=120800 micro_steps=241600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7724 acc_all=0.5971 loss_corrupt=2.7724 acc_corrupt=0.5971 corrupt_frac=1.0000 loss=2.7724 loss_recon=2.7724 loss_meanflow=0.0000 mean_model_t=0.5053 mean_corrupt_t=0.5053 wrong_frac=0.4949 init_acc_corrupt=0.4710 init_gold_top10=0.4996 init_gold_top100=0.5053
+step=120900 micro_steps=241800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8065 acc_all=0.5922 loss_corrupt=2.8065 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8065 loss_recon=2.8065 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4995 init_acc_corrupt=0.4657 init_gold_top10=0.4949 init_gold_top100=0.5007
+step=121000 micro_steps=242000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7955 acc_all=0.5940 loss_corrupt=2.7955 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7955 loss_recon=2.7955 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4985 init_acc_corrupt=0.4672 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=121100 micro_steps=242200 elapsed=65.0s lr=3.000000e-04 loss_all=2.8354 acc_all=0.5881 loss_corrupt=2.8354 acc_corrupt=0.5881 corrupt_frac=1.0000 loss=2.8354 loss_recon=2.8354 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5035 init_acc_corrupt=0.4618 init_gold_top10=0.4907 init_gold_top100=0.4966
+step=121200 micro_steps=242400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8014 acc_all=0.5929 loss_corrupt=2.8014 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.8014 loss_recon=2.8014 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4999 init_acc_corrupt=0.4658 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=121300 micro_steps=242600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8196 acc_all=0.5900 loss_corrupt=2.8196 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8196 loss_recon=2.8196 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5025 init_acc_corrupt=0.4626 init_gold_top10=0.4919 init_gold_top100=0.4976
+step=121400 micro_steps=242800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8169 acc_all=0.5910 loss_corrupt=2.8169 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8169 loss_recon=2.8169 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5014 init_acc_corrupt=0.4638 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=121500 micro_steps=243000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8377 acc_all=0.5878 loss_corrupt=2.8377 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8377 loss_recon=2.8377 loss_meanflow=0.0000 mean_model_t=0.4957 mean_corrupt_t=0.4957 wrong_frac=0.5043 init_acc_corrupt=0.4607 init_gold_top10=0.4899 init_gold_top100=0.4959
+step=121600 micro_steps=243200 elapsed=63.4s lr=3.000000e-04 loss_all=2.8283 acc_all=0.5892 loss_corrupt=2.8283 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8283 loss_recon=2.8283 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5025 init_acc_corrupt=0.4632 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=121700 micro_steps=243400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8044 acc_all=0.5921 loss_corrupt=2.8044 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8044 loss_recon=2.8044 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5004 init_acc_corrupt=0.4654 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=121800 micro_steps=243600 elapsed=63.2s lr=3.000000e-04 loss_all=2.8137 acc_all=0.5911 loss_corrupt=2.8137 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8137 loss_recon=2.8137 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4646 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=121900 micro_steps=243800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8044 acc_all=0.5924 loss_corrupt=2.8044 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8044 loss_recon=2.8044 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5003 init_acc_corrupt=0.4656 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=122000 micro_steps=244000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8044 acc_all=0.5926 loss_corrupt=2.8044 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8044 loss_recon=2.8044 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4996 init_acc_corrupt=0.4664 init_gold_top10=0.4949 init_gold_top100=0.5005
+step=122100 micro_steps=244200 elapsed=65.0s lr=3.000000e-04 loss_all=2.8228 acc_all=0.5891 loss_corrupt=2.8228 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8228 loss_recon=2.8228 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5033 init_acc_corrupt=0.4619 init_gold_top10=0.4912 init_gold_top100=0.4969
+step=122200 micro_steps=244400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8320 acc_all=0.5889 loss_corrupt=2.8320 acc_corrupt=0.5889 corrupt_frac=1.0000 loss=2.8320 loss_recon=2.8320 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5027 init_acc_corrupt=0.4622 init_gold_top10=0.4916 init_gold_top100=0.4975
+step=122300 micro_steps=244600 elapsed=59.9s lr=3.000000e-04 loss_all=2.8024 acc_all=0.5926 loss_corrupt=2.8024 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8024 loss_recon=2.8024 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4997 init_acc_corrupt=0.4654 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=122400 micro_steps=244800 elapsed=62.8s lr=3.000000e-04 loss_all=2.8134 acc_all=0.5918 loss_corrupt=2.8134 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8134 loss_recon=2.8134 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5007 init_acc_corrupt=0.4650 init_gold_top10=0.4938 init_gold_top100=0.4995
+step=122500 micro_steps=245000 elapsed=62.6s lr=3.000000e-04 loss_all=2.8121 acc_all=0.5916 loss_corrupt=2.8121 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8121 loss_recon=2.8121 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5002 init_acc_corrupt=0.4646 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=122600 micro_steps=245200 elapsed=62.7s lr=3.000000e-04 loss_all=2.8435 acc_all=0.5875 loss_corrupt=2.8435 acc_corrupt=0.5875 corrupt_frac=1.0000 loss=2.8435 loss_recon=2.8435 loss_meanflow=0.0000 mean_model_t=0.4960 mean_corrupt_t=0.4960 wrong_frac=0.5038 init_acc_corrupt=0.4609 init_gold_top10=0.4905 init_gold_top100=0.4963
+step=122700 micro_steps=245400 elapsed=63.0s lr=3.000000e-04 loss_all=2.7622 acc_all=0.5979 loss_corrupt=2.7622 acc_corrupt=0.5979 corrupt_frac=1.0000 loss=2.7622 loss_recon=2.7622 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5045 wrong_frac=0.4956 init_acc_corrupt=0.4703 init_gold_top10=0.4991 init_gold_top100=0.5046
+step=122800 micro_steps=245600 elapsed=63.0s lr=3.000000e-04 loss_all=2.8174 acc_all=0.5907 loss_corrupt=2.8174 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8174 loss_recon=2.8174 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5017 init_acc_corrupt=0.4631 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=122900 micro_steps=245800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7844 acc_all=0.5951 loss_corrupt=2.7844 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7844 loss_recon=2.7844 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4977 init_acc_corrupt=0.4679 init_gold_top10=0.4968 init_gold_top100=0.5024
+step=123000 micro_steps=246000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8124 acc_all=0.5912 loss_corrupt=2.8124 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8124 loss_recon=2.8124 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5018 init_acc_corrupt=0.4641 init_gold_top10=0.4928 init_gold_top100=0.4984
+step=123100 micro_steps=246200 elapsed=64.3s lr=3.000000e-04 loss_all=2.8376 acc_all=0.5882 loss_corrupt=2.8376 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8376 loss_recon=2.8376 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5033 init_acc_corrupt=0.4618 init_gold_top10=0.4910 init_gold_top100=0.4968
+step=123200 micro_steps=246400 elapsed=62.8s lr=3.000000e-04 loss_all=2.8064 acc_all=0.5920 loss_corrupt=2.8064 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8064 loss_recon=2.8064 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5003 init_acc_corrupt=0.4653 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=123300 micro_steps=246600 elapsed=63.0s lr=3.000000e-04 loss_all=2.8042 acc_all=0.5924 loss_corrupt=2.8042 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8042 loss_recon=2.8042 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4991 init_acc_corrupt=0.4662 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=123400 micro_steps=246800 elapsed=62.8s lr=3.000000e-04 loss_all=2.8171 acc_all=0.5909 loss_corrupt=2.8171 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8171 loss_recon=2.8171 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5014 init_acc_corrupt=0.4643 init_gold_top10=0.4930 init_gold_top100=0.4987
+step=123500 micro_steps=247000 elapsed=63.3s lr=3.000000e-04 loss_all=2.8235 acc_all=0.5901 loss_corrupt=2.8235 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8235 loss_recon=2.8235 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5018 init_acc_corrupt=0.4632 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=123600 micro_steps=247200 elapsed=63.4s lr=3.000000e-04 loss_all=2.8098 acc_all=0.5922 loss_corrupt=2.8098 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8098 loss_recon=2.8098 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4989 init_acc_corrupt=0.4664 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=123700 micro_steps=247400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8175 acc_all=0.5912 loss_corrupt=2.8175 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8175 loss_recon=2.8175 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5004 init_acc_corrupt=0.4651 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=123800 micro_steps=247600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7900 acc_all=0.5941 loss_corrupt=2.7900 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7900 loss_recon=2.7900 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4984 init_acc_corrupt=0.4674 init_gold_top10=0.4960 init_gold_top100=0.5018
+step=123900 micro_steps=247800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8243 acc_all=0.5898 loss_corrupt=2.8243 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8243 loss_recon=2.8243 loss_meanflow=0.0000 mean_model_t=0.4966 mean_corrupt_t=0.4966 wrong_frac=0.5030 init_acc_corrupt=0.4628 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=124000 micro_steps=248000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7936 acc_all=0.5940 loss_corrupt=2.7936 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7936 loss_recon=2.7936 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4982 init_acc_corrupt=0.4671 init_gold_top10=0.4963 init_gold_top100=0.5019
+step=124100 micro_steps=248200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7971 acc_all=0.5935 loss_corrupt=2.7971 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7971 loss_recon=2.7971 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4994 init_acc_corrupt=0.4666 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=124200 micro_steps=248400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7918 acc_all=0.5941 loss_corrupt=2.7918 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7918 loss_recon=2.7918 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4994 init_acc_corrupt=0.4669 init_gold_top10=0.4951 init_gold_top100=0.5006
+step=124300 micro_steps=248600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7699 acc_all=0.5970 loss_corrupt=2.7699 acc_corrupt=0.5970 corrupt_frac=1.0000 loss=2.7699 loss_recon=2.7699 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5045 wrong_frac=0.4956 init_acc_corrupt=0.4701 init_gold_top10=0.4988 init_gold_top100=0.5046
+step=124400 micro_steps=248800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8186 acc_all=0.5906 loss_corrupt=2.8186 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8186 loss_recon=2.8186 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5021 init_acc_corrupt=0.4635 init_gold_top10=0.4922 init_gold_top100=0.4980
+step=124500 micro_steps=249000 elapsed=63.3s lr=3.000000e-04 loss_all=2.8340 acc_all=0.5886 loss_corrupt=2.8340 acc_corrupt=0.5886 corrupt_frac=1.0000 loss=2.8340 loss_recon=2.8340 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5029 init_acc_corrupt=0.4617 init_gold_top10=0.4913 init_gold_top100=0.4972
+step=124600 micro_steps=249200 elapsed=63.3s lr=3.000000e-04 loss_all=2.8445 acc_all=0.5876 loss_corrupt=2.8445 acc_corrupt=0.5876 corrupt_frac=1.0000 loss=2.8445 loss_recon=2.8445 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5038 init_acc_corrupt=0.4612 init_gold_top10=0.4905 init_gold_top100=0.4964
+step=124700 micro_steps=249400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7990 acc_all=0.5932 loss_corrupt=2.7990 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7990 loss_recon=2.7990 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4999 init_acc_corrupt=0.4660 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=124800 micro_steps=249600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8257 acc_all=0.5898 loss_corrupt=2.8257 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8257 loss_recon=2.8257 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5023 init_acc_corrupt=0.4624 init_gold_top10=0.4920 init_gold_top100=0.4979
+step=124900 micro_steps=249800 elapsed=65.1s lr=3.000000e-04 loss_all=2.8011 acc_all=0.5928 loss_corrupt=2.8011 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8011 loss_recon=2.8011 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4989 init_acc_corrupt=0.4662 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=125000 micro_steps=250000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8371 acc_all=0.5881 loss_corrupt=2.8371 acc_corrupt=0.5881 corrupt_frac=1.0000 loss=2.8371 loss_recon=2.8371 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5040 init_acc_corrupt=0.4609 init_gold_top10=0.4904 init_gold_top100=0.4961
+step=125100 micro_steps=250200 elapsed=65.1s lr=3.000000e-04 loss_all=2.8014 acc_all=0.5927 loss_corrupt=2.8014 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8014 loss_recon=2.8014 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4993 init_acc_corrupt=0.4659 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=125200 micro_steps=250400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7998 acc_all=0.5938 loss_corrupt=2.7998 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7998 loss_recon=2.7998 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4987 init_acc_corrupt=0.4671 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=125300 micro_steps=250600 elapsed=60.0s lr=3.000000e-04 loss_all=2.7987 acc_all=0.5931 loss_corrupt=2.7987 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.7987 loss_recon=2.7987 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4994 init_acc_corrupt=0.4661 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=125400 micro_steps=250800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8229 acc_all=0.5901 loss_corrupt=2.8229 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8229 loss_recon=2.8229 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5015 init_acc_corrupt=0.4635 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=125500 micro_steps=251000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8491 acc_all=0.5868 loss_corrupt=2.8491 acc_corrupt=0.5868 corrupt_frac=1.0000 loss=2.8491 loss_recon=2.8491 loss_meanflow=0.0000 mean_model_t=0.4948 mean_corrupt_t=0.4948 wrong_frac=0.5053 init_acc_corrupt=0.4595 init_gold_top10=0.4888 init_gold_top100=0.4948
+step=125600 micro_steps=251200 elapsed=63.3s lr=3.000000e-04 loss_all=2.8134 acc_all=0.5913 loss_corrupt=2.8134 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8134 loss_recon=2.8134 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5008 init_acc_corrupt=0.4646 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=125700 micro_steps=251400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8000 acc_all=0.5925 loss_corrupt=2.8000 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8000 loss_recon=2.8000 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4994 init_acc_corrupt=0.4662 init_gold_top10=0.4951 init_gold_top100=0.5007
+step=125800 micro_steps=251600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7965 acc_all=0.5933 loss_corrupt=2.7965 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7965 loss_recon=2.7965 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4991 init_acc_corrupt=0.4665 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=125900 micro_steps=251800 elapsed=63.3s lr=3.000000e-04 loss_all=2.8152 acc_all=0.5917 loss_corrupt=2.8152 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8152 loss_recon=2.8152 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4997 init_acc_corrupt=0.4656 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=126000 micro_steps=252000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7526 acc_all=0.5992 loss_corrupt=2.7526 acc_corrupt=0.5992 corrupt_frac=1.0000 loss=2.7526 loss_recon=2.7526 loss_meanflow=0.0000 mean_model_t=0.5071 mean_corrupt_t=0.5071 wrong_frac=0.4932 init_acc_corrupt=0.4731 init_gold_top10=0.5015 init_gold_top100=0.5070
+step=126100 micro_steps=252200 elapsed=65.0s lr=3.000000e-04 loss_all=2.8269 acc_all=0.5895 loss_corrupt=2.8269 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8269 loss_recon=2.8269 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5031 init_acc_corrupt=0.4623 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=126200 micro_steps=252400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8082 acc_all=0.5921 loss_corrupt=2.8082 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8082 loss_recon=2.8082 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5002 init_acc_corrupt=0.4651 init_gold_top10=0.4943 init_gold_top100=0.4999
+step=126300 micro_steps=252600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7949 acc_all=0.5944 loss_corrupt=2.7949 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7949 loss_recon=2.7949 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4979 init_acc_corrupt=0.4675 init_gold_top10=0.4964 init_gold_top100=0.5023
+step=126400 micro_steps=252800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7693 acc_all=0.5971 loss_corrupt=2.7693 acc_corrupt=0.5971 corrupt_frac=1.0000 loss=2.7693 loss_recon=2.7693 loss_meanflow=0.0000 mean_model_t=0.5052 mean_corrupt_t=0.5052 wrong_frac=0.4952 init_acc_corrupt=0.4711 init_gold_top10=0.4993 init_gold_top100=0.5049
+step=126500 micro_steps=253000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7986 acc_all=0.5927 loss_corrupt=2.7986 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.7986 loss_recon=2.7986 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4998 init_acc_corrupt=0.4655 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=126600 micro_steps=253200 elapsed=63.4s lr=3.000000e-04 loss_all=2.8136 acc_all=0.5912 loss_corrupt=2.8136 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8136 loss_recon=2.8136 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5017 init_acc_corrupt=0.4642 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=126700 micro_steps=253400 elapsed=62.9s lr=3.000000e-04 loss_all=2.7953 acc_all=0.5936 loss_corrupt=2.7953 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7953 loss_recon=2.7953 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4987 init_acc_corrupt=0.4667 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=126800 micro_steps=253600 elapsed=62.9s lr=3.000000e-04 loss_all=2.7782 acc_all=0.5957 loss_corrupt=2.7782 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7782 loss_recon=2.7782 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4971 init_acc_corrupt=0.4693 init_gold_top10=0.4974 init_gold_top100=0.5030
+step=126900 micro_steps=253800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8127 acc_all=0.5915 loss_corrupt=2.8127 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8127 loss_recon=2.8127 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4644 init_gold_top10=0.4937 init_gold_top100=0.4995
+step=127000 micro_steps=254000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7658 acc_all=0.5974 loss_corrupt=2.7658 acc_corrupt=0.5974 corrupt_frac=1.0000 loss=2.7658 loss_recon=2.7658 loss_meanflow=0.0000 mean_model_t=0.5046 mean_corrupt_t=0.5046 wrong_frac=0.4958 init_acc_corrupt=0.4707 init_gold_top10=0.4988 init_gold_top100=0.5043
+step=127100 micro_steps=254200 elapsed=65.2s lr=3.000000e-04 loss_all=2.8251 acc_all=0.5897 loss_corrupt=2.8251 acc_corrupt=0.5897 corrupt_frac=1.0000 loss=2.8251 loss_recon=2.8251 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5023 init_acc_corrupt=0.4628 init_gold_top10=0.4920 init_gold_top100=0.4978
+step=127200 micro_steps=254400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7801 acc_all=0.5956 loss_corrupt=2.7801 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7801 loss_recon=2.7801 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4975 init_acc_corrupt=0.4683 init_gold_top10=0.4971 init_gold_top100=0.5026
+step=127300 micro_steps=254600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8147 acc_all=0.5911 loss_corrupt=2.8147 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8147 loss_recon=2.8147 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5017 init_acc_corrupt=0.4633 init_gold_top10=0.4925 init_gold_top100=0.4983
+step=127400 micro_steps=254800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8216 acc_all=0.5903 loss_corrupt=2.8216 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8216 loss_recon=2.8216 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5019 init_acc_corrupt=0.4632 init_gold_top10=0.4923 init_gold_top100=0.4982
+step=127500 micro_steps=255000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8196 acc_all=0.5901 loss_corrupt=2.8196 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8196 loss_recon=2.8196 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5027 init_acc_corrupt=0.4619 init_gold_top10=0.4917 init_gold_top100=0.4975
+step=127600 micro_steps=255200 elapsed=63.3s lr=3.000000e-04 loss_all=2.7693 acc_all=0.5973 loss_corrupt=2.7693 acc_corrupt=0.5973 corrupt_frac=1.0000 loss=2.7693 loss_recon=2.7693 loss_meanflow=0.0000 mean_model_t=0.5052 mean_corrupt_t=0.5052 wrong_frac=0.4950 init_acc_corrupt=0.4713 init_gold_top10=0.4995 init_gold_top100=0.5051
+step=127700 micro_steps=255400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8095 acc_all=0.5914 loss_corrupt=2.8095 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8095 loss_recon=2.8095 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5020 init_acc_corrupt=0.4634 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=127800 micro_steps=255600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7839 acc_all=0.5947 loss_corrupt=2.7839 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7839 loss_recon=2.7839 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4982 init_acc_corrupt=0.4670 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=127900 micro_steps=255800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8187 acc_all=0.5906 loss_corrupt=2.8187 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8187 loss_recon=2.8187 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5011 init_acc_corrupt=0.4641 init_gold_top10=0.4930 init_gold_top100=0.4990
+step=128000 micro_steps=256000 elapsed=63.0s lr=3.000000e-04 loss_all=2.8110 acc_all=0.5918 loss_corrupt=2.8110 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8110 loss_recon=2.8110 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.4998 init_acc_corrupt=0.4653 init_gold_top10=0.4945 init_gold_top100=0.5004
+step=128100 micro_steps=256200 elapsed=64.2s lr=3.000000e-04 loss_all=2.8043 acc_all=0.5928 loss_corrupt=2.8043 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8043 loss_recon=2.8043 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4993 init_acc_corrupt=0.4668 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=128200 micro_steps=256400 elapsed=62.7s lr=3.000000e-04 loss_all=2.7883 acc_all=0.5949 loss_corrupt=2.7883 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7883 loss_recon=2.7883 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4976 init_acc_corrupt=0.4684 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=128300 micro_steps=256600 elapsed=63.3s lr=3.000000e-04 loss_all=2.8238 acc_all=0.5902 loss_corrupt=2.8238 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8238 loss_recon=2.8238 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5021 init_acc_corrupt=0.4631 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=128400 micro_steps=256800 elapsed=59.8s lr=3.000000e-04 loss_all=2.7955 acc_all=0.5937 loss_corrupt=2.7955 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7955 loss_recon=2.7955 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4993 init_acc_corrupt=0.4663 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=128500 micro_steps=257000 elapsed=62.7s lr=3.000000e-04 loss_all=2.8263 acc_all=0.5896 loss_corrupt=2.8263 acc_corrupt=0.5896 corrupt_frac=1.0000 loss=2.8263 loss_recon=2.8263 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5023 init_acc_corrupt=0.4629 init_gold_top10=0.4920 init_gold_top100=0.4978
+step=128600 micro_steps=257200 elapsed=63.4s lr=3.000000e-04 loss_all=2.7672 acc_all=0.5974 loss_corrupt=2.7672 acc_corrupt=0.5974 corrupt_frac=1.0000 loss=2.7672 loss_recon=2.7672 loss_meanflow=0.0000 mean_model_t=0.5049 mean_corrupt_t=0.5049 wrong_frac=0.4950 init_acc_corrupt=0.4712 init_gold_top10=0.4997 init_gold_top100=0.5052
+step=128700 micro_steps=257400 elapsed=63.3s lr=3.000000e-04 loss_all=2.8138 acc_all=0.5915 loss_corrupt=2.8138 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8138 loss_recon=2.8138 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5000 init_acc_corrupt=0.4652 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=128800 micro_steps=257600 elapsed=63.3s lr=3.000000e-04 loss_all=2.7960 acc_all=0.5932 loss_corrupt=2.7960 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7960 loss_recon=2.7960 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5000 init_acc_corrupt=0.4651 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=128900 micro_steps=257800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7981 acc_all=0.5934 loss_corrupt=2.7981 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7981 loss_recon=2.7981 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4985 init_acc_corrupt=0.4668 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=129000 micro_steps=258000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7997 acc_all=0.5934 loss_corrupt=2.7997 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7997 loss_recon=2.7997 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4994 init_acc_corrupt=0.4663 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=129100 micro_steps=258200 elapsed=65.0s lr=3.000000e-04 loss_all=2.8111 acc_all=0.5918 loss_corrupt=2.8111 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8111 loss_recon=2.8111 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5009 init_acc_corrupt=0.4644 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=129200 micro_steps=258400 elapsed=63.6s lr=3.000000e-04 loss_all=2.7914 acc_all=0.5938 loss_corrupt=2.7914 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7914 loss_recon=2.7914 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4998 init_acc_corrupt=0.4656 init_gold_top10=0.4947 init_gold_top100=0.5003
+step=129300 micro_steps=258600 elapsed=63.3s lr=3.000000e-04 loss_all=2.8082 acc_all=0.5921 loss_corrupt=2.8082 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8082 loss_recon=2.8082 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4993 init_acc_corrupt=0.4657 init_gold_top10=0.4948 init_gold_top100=0.5008
+step=129400 micro_steps=258800 elapsed=63.0s lr=3.000000e-04 loss_all=2.7921 acc_all=0.5940 loss_corrupt=2.7921 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7921 loss_recon=2.7921 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4990 init_acc_corrupt=0.4664 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=129500 micro_steps=259000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8056 acc_all=0.5928 loss_corrupt=2.8056 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8056 loss_recon=2.8056 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4987 init_acc_corrupt=0.4665 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=129600 micro_steps=259200 elapsed=64.2s lr=3.000000e-04 loss_all=2.7899 acc_all=0.5944 loss_corrupt=2.7899 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7899 loss_recon=2.7899 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4981 init_acc_corrupt=0.4676 init_gold_top10=0.4964 init_gold_top100=0.5020
+step=129700 micro_steps=259400 elapsed=64.1s lr=3.000000e-04 loss_all=2.7876 acc_all=0.5943 loss_corrupt=2.7876 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7876 loss_recon=2.7876 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4998 init_acc_corrupt=0.4658 init_gold_top10=0.4948 init_gold_top100=0.5004
+step=129800 micro_steps=259600 elapsed=63.9s lr=3.000000e-04 loss_all=2.7967 acc_all=0.5940 loss_corrupt=2.7967 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7967 loss_recon=2.7967 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4975 init_acc_corrupt=0.4676 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=129900 micro_steps=259800 elapsed=63.8s lr=3.000000e-04 loss_all=2.8034 acc_all=0.5926 loss_corrupt=2.8034 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8034 loss_recon=2.8034 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5004 init_acc_corrupt=0.4655 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=130000 micro_steps=260000 elapsed=64.3s lr=3.000000e-04 loss_all=2.7922 acc_all=0.5945 loss_corrupt=2.7922 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7922 loss_recon=2.7922 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4976 init_acc_corrupt=0.4684 init_gold_top10=0.4967 init_gold_top100=0.5025
+step=130100 micro_steps=260200 elapsed=65.8s lr=3.000000e-04 loss_all=2.8155 acc_all=0.5908 loss_corrupt=2.8155 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8155 loss_recon=2.8155 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5024 init_acc_corrupt=0.4627 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=130200 micro_steps=260400 elapsed=64.2s lr=3.000000e-04 loss_all=2.8203 acc_all=0.5908 loss_corrupt=2.8203 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8203 loss_recon=2.8203 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5012 init_acc_corrupt=0.4641 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=130300 micro_steps=260600 elapsed=64.2s lr=3.000000e-04 loss_all=2.8127 acc_all=0.5920 loss_corrupt=2.8127 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8127 loss_recon=2.8127 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5000 init_acc_corrupt=0.4653 init_gold_top10=0.4942 init_gold_top100=0.5002
+step=130400 micro_steps=260800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8052 acc_all=0.5921 loss_corrupt=2.8052 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8052 loss_recon=2.8052 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5002 init_acc_corrupt=0.4650 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=130500 micro_steps=261000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7837 acc_all=0.5955 loss_corrupt=2.7837 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7837 loss_recon=2.7837 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4961 init_acc_corrupt=0.4698 init_gold_top10=0.4982 init_gold_top100=0.5039
+step=130600 micro_steps=261200 elapsed=63.4s lr=3.000000e-04 loss_all=2.7830 acc_all=0.5956 loss_corrupt=2.7830 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7830 loss_recon=2.7830 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4970 init_acc_corrupt=0.4687 init_gold_top10=0.4974 init_gold_top100=0.5031
+step=130700 micro_steps=261400 elapsed=63.8s lr=3.000000e-04 loss_all=2.8206 acc_all=0.5906 loss_corrupt=2.8206 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8206 loss_recon=2.8206 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5014 init_acc_corrupt=0.4631 init_gold_top10=0.4929 init_gold_top100=0.4988
+step=130800 micro_steps=261600 elapsed=64.2s lr=3.000000e-04 loss_all=2.7946 acc_all=0.5939 loss_corrupt=2.7946 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7946 loss_recon=2.7946 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4986 init_acc_corrupt=0.4663 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=130900 micro_steps=261800 elapsed=63.9s lr=3.000000e-04 loss_all=2.7854 acc_all=0.5952 loss_corrupt=2.7854 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7854 loss_recon=2.7854 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4970 init_acc_corrupt=0.4691 init_gold_top10=0.4974 init_gold_top100=0.5031
+step=131000 micro_steps=262000 elapsed=63.9s lr=3.000000e-04 loss_all=2.8069 acc_all=0.5921 loss_corrupt=2.8069 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8069 loss_recon=2.8069 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5008 init_acc_corrupt=0.4644 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=131100 micro_steps=262200 elapsed=65.4s lr=3.000000e-04 loss_all=2.8129 acc_all=0.5914 loss_corrupt=2.8129 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8129 loss_recon=2.8129 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5004 init_acc_corrupt=0.4645 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=131200 micro_steps=262400 elapsed=64.3s lr=3.000000e-04 loss_all=2.7886 acc_all=0.5944 loss_corrupt=2.7886 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7886 loss_recon=2.7886 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4987 init_acc_corrupt=0.4670 init_gold_top10=0.4958 init_gold_top100=0.5014
+step=131300 micro_steps=262600 elapsed=63.9s lr=3.000000e-04 loss_all=2.7954 acc_all=0.5936 loss_corrupt=2.7954 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7954 loss_recon=2.7954 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4983 init_acc_corrupt=0.4670 init_gold_top10=0.4958 init_gold_top100=0.5018
+step=131400 micro_steps=262800 elapsed=60.2s lr=3.000000e-04 loss_all=2.7856 acc_all=0.5948 loss_corrupt=2.7856 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7856 loss_recon=2.7856 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4982 init_acc_corrupt=0.4677 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=131500 micro_steps=263000 elapsed=63.1s lr=3.000000e-04 loss_all=2.8072 acc_all=0.5917 loss_corrupt=2.8072 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8072 loss_recon=2.8072 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5024 init_acc_corrupt=0.4634 init_gold_top10=0.4921 init_gold_top100=0.4977
+step=131600 micro_steps=263200 elapsed=64.0s lr=3.000000e-04 loss_all=2.8414 acc_all=0.5877 loss_corrupt=2.8414 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8414 loss_recon=2.8414 loss_meanflow=0.0000 mean_model_t=0.4952 mean_corrupt_t=0.4952 wrong_frac=0.5044 init_acc_corrupt=0.4602 init_gold_top10=0.4898 init_gold_top100=0.4957
+step=131700 micro_steps=263400 elapsed=64.0s lr=3.000000e-04 loss_all=2.8112 acc_all=0.5919 loss_corrupt=2.8112 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8112 loss_recon=2.8112 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4997 init_acc_corrupt=0.4652 init_gold_top10=0.4945 init_gold_top100=0.5004
+step=131800 micro_steps=263600 elapsed=63.9s lr=3.000000e-04 loss_all=2.8161 acc_all=0.5902 loss_corrupt=2.8161 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8161 loss_recon=2.8161 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5036 init_acc_corrupt=0.4614 init_gold_top10=0.4909 init_gold_top100=0.4965
+step=131900 micro_steps=263800 elapsed=64.2s lr=3.000000e-04 loss_all=2.7825 acc_all=0.5957 loss_corrupt=2.7825 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7825 loss_recon=2.7825 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4969 init_acc_corrupt=0.4690 init_gold_top10=0.4976 init_gold_top100=0.5032
+step=132000 micro_steps=264000 elapsed=64.3s lr=3.000000e-04 loss_all=2.8039 acc_all=0.5925 loss_corrupt=2.8039 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8039 loss_recon=2.8039 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.5002 init_acc_corrupt=0.4658 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=132100 micro_steps=264200 elapsed=64.7s lr=3.000000e-04 loss_all=2.8167 acc_all=0.5911 loss_corrupt=2.8167 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8167 loss_recon=2.8167 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5010 init_acc_corrupt=0.4639 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=132200 micro_steps=264400 elapsed=63.3s lr=3.000000e-04 loss_all=2.8112 acc_all=0.5919 loss_corrupt=2.8112 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8112 loss_recon=2.8112 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5008 init_acc_corrupt=0.4652 init_gold_top10=0.4937 init_gold_top100=0.4993
+step=132300 micro_steps=264600 elapsed=63.3s lr=3.000000e-04 loss_all=2.7863 acc_all=0.5947 loss_corrupt=2.7863 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7863 loss_recon=2.7863 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4983 init_acc_corrupt=0.4665 init_gold_top10=0.4964 init_gold_top100=0.5019
+step=132400 micro_steps=264800 elapsed=63.3s lr=3.000000e-04 loss_all=2.7975 acc_all=0.5934 loss_corrupt=2.7975 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7975 loss_recon=2.7975 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4993 init_acc_corrupt=0.4657 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=132500 micro_steps=265000 elapsed=63.2s lr=3.000000e-04 loss_all=2.7949 acc_all=0.5938 loss_corrupt=2.7949 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7949 loss_recon=2.7949 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4986 init_acc_corrupt=0.4670 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=132600 micro_steps=265200 elapsed=63.2s lr=3.000000e-04 loss_all=2.7660 acc_all=0.5972 loss_corrupt=2.7660 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7660 loss_recon=2.7660 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4961 init_acc_corrupt=0.4699 init_gold_top10=0.4984 init_gold_top100=0.5041
+step=132700 micro_steps=265400 elapsed=63.1s lr=3.000000e-04 loss_all=2.8100 acc_all=0.5921 loss_corrupt=2.8100 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8100 loss_recon=2.8100 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4998 init_acc_corrupt=0.4652 init_gold_top10=0.4945 init_gold_top100=0.5004
+step=132800 micro_steps=265600 elapsed=63.1s lr=3.000000e-04 loss_all=2.8309 acc_all=0.5890 loss_corrupt=2.8309 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8309 loss_recon=2.8309 loss_meanflow=0.0000 mean_model_t=0.4966 mean_corrupt_t=0.4966 wrong_frac=0.5033 init_acc_corrupt=0.4612 init_gold_top10=0.4909 init_gold_top100=0.4969
+step=132900 micro_steps=265800 elapsed=63.3s lr=3.000000e-04 loss_all=2.7880 acc_all=0.5948 loss_corrupt=2.7880 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7880 loss_recon=2.7880 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4978 init_acc_corrupt=0.4678 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=133000 micro_steps=266000 elapsed=63.3s lr=3.000000e-04 loss_all=2.7739 acc_all=0.5962 loss_corrupt=2.7739 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7739 loss_recon=2.7739 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4973 init_acc_corrupt=0.4687 init_gold_top10=0.4973 init_gold_top100=0.5028
+step=133100 micro_steps=266200 elapsed=64.9s lr=3.000000e-04 loss_all=2.8034 acc_all=0.5926 loss_corrupt=2.8034 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8034 loss_recon=2.8034 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5001 init_acc_corrupt=0.4654 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=133200 micro_steps=266400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8102 acc_all=0.5919 loss_corrupt=2.8102 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8102 loss_recon=2.8102 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5010 init_acc_corrupt=0.4647 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=133300 micro_steps=266600 elapsed=63.7s lr=3.000000e-04 loss_all=2.7955 acc_all=0.5928 loss_corrupt=2.7955 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.7955 loss_recon=2.7955 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5010 init_acc_corrupt=0.4645 init_gold_top10=0.4935 init_gold_top100=0.4991
+step=133400 micro_steps=266800 elapsed=63.7s lr=3.000000e-04 loss_all=2.8304 acc_all=0.5890 loss_corrupt=2.8304 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8304 loss_recon=2.8304 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5041 init_acc_corrupt=0.4614 init_gold_top10=0.4903 init_gold_top100=0.4960
+step=133500 micro_steps=267000 elapsed=63.8s lr=3.000000e-04 loss_all=2.7901 acc_all=0.5941 loss_corrupt=2.7901 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7901 loss_recon=2.7901 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4985 init_acc_corrupt=0.4673 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=133600 micro_steps=267200 elapsed=63.7s lr=3.000000e-04 loss_all=2.8133 acc_all=0.5917 loss_corrupt=2.8133 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8133 loss_recon=2.8133 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5013 init_acc_corrupt=0.4640 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=133700 micro_steps=267400 elapsed=63.3s lr=3.000000e-04 loss_all=2.8189 acc_all=0.5906 loss_corrupt=2.8189 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8189 loss_recon=2.8189 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5011 init_acc_corrupt=0.4642 init_gold_top10=0.4932 init_gold_top100=0.4991
+step=133800 micro_steps=267600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8077 acc_all=0.5918 loss_corrupt=2.8077 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8077 loss_recon=2.8077 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5012 init_acc_corrupt=0.4643 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=133900 micro_steps=267800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7937 acc_all=0.5941 loss_corrupt=2.7937 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7937 loss_recon=2.7937 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4982 init_acc_corrupt=0.4677 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=134000 micro_steps=268000 elapsed=63.2s lr=3.000000e-04 loss_all=2.8329 acc_all=0.5886 loss_corrupt=2.8329 acc_corrupt=0.5886 corrupt_frac=1.0000 loss=2.8329 loss_recon=2.8329 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5037 init_acc_corrupt=0.4618 init_gold_top10=0.4907 init_gold_top100=0.4964
+step=134100 micro_steps=268200 elapsed=65.1s lr=3.000000e-04 loss_all=2.8093 acc_all=0.5916 loss_corrupt=2.8093 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8093 loss_recon=2.8093 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5015 init_acc_corrupt=0.4640 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=134200 micro_steps=268400 elapsed=63.6s lr=3.000000e-04 loss_all=2.8139 acc_all=0.5913 loss_corrupt=2.8139 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8139 loss_recon=2.8139 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5005 init_acc_corrupt=0.4649 init_gold_top10=0.4937 init_gold_top100=0.4995
+step=134300 micro_steps=268600 elapsed=63.1s lr=3.000000e-04 loss_all=2.8034 acc_all=0.5930 loss_corrupt=2.8034 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.8034 loss_recon=2.8034 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4998 init_acc_corrupt=0.4656 init_gold_top10=0.4945 init_gold_top100=0.5003
+step=134400 micro_steps=268800 elapsed=61.3s lr=3.000000e-04 loss_all=2.7926 acc_all=0.5939 loss_corrupt=2.7926 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7926 loss_recon=2.7926 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4990 init_acc_corrupt=0.4665 init_gold_top10=0.4954 init_gold_top100=0.5012
+step=134500 micro_steps=269000 elapsed=62.0s lr=3.000000e-04 loss_all=2.8098 acc_all=0.5913 loss_corrupt=2.8098 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8098 loss_recon=2.8098 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5019 init_acc_corrupt=0.4637 init_gold_top10=0.4926 init_gold_top100=0.4983
+step=134600 micro_steps=269200 elapsed=63.5s lr=3.000000e-04 loss_all=2.8184 acc_all=0.5911 loss_corrupt=2.8184 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8184 loss_recon=2.8184 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5010 init_acc_corrupt=0.4647 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=134700 micro_steps=269400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8083 acc_all=0.5924 loss_corrupt=2.8083 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8083 loss_recon=2.8083 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5003 init_acc_corrupt=0.4649 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=134800 micro_steps=269600 elapsed=64.2s lr=3.000000e-04 loss_all=2.7856 acc_all=0.5951 loss_corrupt=2.7856 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7856 loss_recon=2.7856 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4984 init_acc_corrupt=0.4669 init_gold_top10=0.4960 init_gold_top100=0.5018
+step=134900 micro_steps=269800 elapsed=64.0s lr=3.000000e-04 loss_all=2.8439 acc_all=0.5873 loss_corrupt=2.8439 acc_corrupt=0.5873 corrupt_frac=1.0000 loss=2.8439 loss_recon=2.8439 loss_meanflow=0.0000 mean_model_t=0.4953 mean_corrupt_t=0.4953 wrong_frac=0.5046 init_acc_corrupt=0.4602 init_gold_top10=0.4897 init_gold_top100=0.4955
+step=135000 micro_steps=270000 elapsed=64.1s lr=3.000000e-04 loss_all=2.7920 acc_all=0.5943 loss_corrupt=2.7920 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7920 loss_recon=2.7920 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4981 init_acc_corrupt=0.4673 init_gold_top10=0.4964 init_gold_top100=0.5020
+step=135100 micro_steps=270200 elapsed=65.1s lr=3.000000e-04 loss_all=2.8333 acc_all=0.5887 loss_corrupt=2.8333 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8333 loss_recon=2.8333 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5034 init_acc_corrupt=0.4614 init_gold_top10=0.4909 init_gold_top100=0.4967
+step=135200 micro_steps=270400 elapsed=63.2s lr=3.000000e-04 loss_all=2.7951 acc_all=0.5942 loss_corrupt=2.7951 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7951 loss_recon=2.7951 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4987 init_acc_corrupt=0.4663 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=135300 micro_steps=270600 elapsed=63.8s lr=3.000000e-04 loss_all=2.7768 acc_all=0.5963 loss_corrupt=2.7768 acc_corrupt=0.5963 corrupt_frac=1.0000 loss=2.7768 loss_recon=2.7768 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4966 init_acc_corrupt=0.4698 init_gold_top10=0.4979 init_gold_top100=0.5035
+step=135400 micro_steps=270800 elapsed=63.9s lr=3.000000e-04 loss_all=2.7843 acc_all=0.5951 loss_corrupt=2.7843 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7843 loss_recon=2.7843 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4975 init_acc_corrupt=0.4681 init_gold_top10=0.4968 init_gold_top100=0.5026
+step=135500 micro_steps=271000 elapsed=64.1s lr=3.000000e-04 loss_all=2.8138 acc_all=0.5912 loss_corrupt=2.8138 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8138 loss_recon=2.8138 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5018 init_acc_corrupt=0.4637 init_gold_top10=0.4926 init_gold_top100=0.4984
+step=135600 micro_steps=271200 elapsed=64.1s lr=3.000000e-04 loss_all=2.8182 acc_all=0.5904 loss_corrupt=2.8182 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8182 loss_recon=2.8182 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5024 init_acc_corrupt=0.4626 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=135700 micro_steps=271400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8050 acc_all=0.5927 loss_corrupt=2.8050 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8050 loss_recon=2.8050 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5005 init_acc_corrupt=0.4648 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=135800 micro_steps=271600 elapsed=63.3s lr=3.000000e-04 loss_all=2.7741 acc_all=0.5962 loss_corrupt=2.7741 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7741 loss_recon=2.7741 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4976 init_acc_corrupt=0.4683 init_gold_top10=0.4968 init_gold_top100=0.5026
+step=135900 micro_steps=271800 elapsed=64.1s lr=3.000000e-04 loss_all=2.7888 acc_all=0.5947 loss_corrupt=2.7888 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7888 loss_recon=2.7888 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4991 init_acc_corrupt=0.4670 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=136000 micro_steps=272000 elapsed=64.1s lr=3.000000e-04 loss_all=2.8160 acc_all=0.5913 loss_corrupt=2.8160 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8160 loss_recon=2.8160 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4998 init_acc_corrupt=0.4652 init_gold_top10=0.4943 init_gold_top100=0.5003
+step=136100 micro_steps=272200 elapsed=64.6s lr=3.000000e-04 loss_all=2.8428 acc_all=0.5877 loss_corrupt=2.8428 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8428 loss_recon=2.8428 loss_meanflow=0.0000 mean_model_t=0.4958 mean_corrupt_t=0.4958 wrong_frac=0.5045 init_acc_corrupt=0.4609 init_gold_top10=0.4899 init_gold_top100=0.4956
+step=136200 micro_steps=272400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7983 acc_all=0.5939 loss_corrupt=2.7983 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7983 loss_recon=2.7983 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4990 init_acc_corrupt=0.4662 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=136300 micro_steps=272600 elapsed=63.8s lr=3.000000e-04 loss_all=2.8205 acc_all=0.5905 loss_corrupt=2.8205 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8205 loss_recon=2.8205 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5016 init_acc_corrupt=0.4635 init_gold_top10=0.4927 init_gold_top100=0.4986
+step=136400 micro_steps=272800 elapsed=63.7s lr=3.000000e-04 loss_all=2.8328 acc_all=0.5891 loss_corrupt=2.8328 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8328 loss_recon=2.8328 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5039 init_acc_corrupt=0.4609 init_gold_top10=0.4904 init_gold_top100=0.4962
+step=136500 micro_steps=273000 elapsed=63.7s lr=3.000000e-04 loss_all=2.8322 acc_all=0.5888 loss_corrupt=2.8322 acc_corrupt=0.5888 corrupt_frac=1.0000 loss=2.8322 loss_recon=2.8322 loss_meanflow=0.0000 mean_model_t=0.4962 mean_corrupt_t=0.4962 wrong_frac=0.5035 init_acc_corrupt=0.4612 init_gold_top10=0.4907 init_gold_top100=0.4967
+step=136600 micro_steps=273200 elapsed=63.9s lr=3.000000e-04 loss_all=2.8200 acc_all=0.5908 loss_corrupt=2.8200 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8200 loss_recon=2.8200 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5019 init_acc_corrupt=0.4629 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=136700 micro_steps=273400 elapsed=63.7s lr=3.000000e-04 loss_all=2.7801 acc_all=0.5960 loss_corrupt=2.7801 acc_corrupt=0.5960 corrupt_frac=1.0000 loss=2.7801 loss_recon=2.7801 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4966 init_acc_corrupt=0.4688 init_gold_top10=0.4978 init_gold_top100=0.5035
+step=136800 micro_steps=273600 elapsed=63.8s lr=3.000000e-04 loss_all=2.7832 acc_all=0.5955 loss_corrupt=2.7832 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7832 loss_recon=2.7832 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4979 init_acc_corrupt=0.4675 init_gold_top10=0.4967 init_gold_top100=0.5023
+step=136900 micro_steps=273800 elapsed=63.8s lr=3.000000e-04 loss_all=2.7923 acc_all=0.5938 loss_corrupt=2.7923 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7923 loss_recon=2.7923 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4989 init_acc_corrupt=0.4659 init_gold_top10=0.4954 init_gold_top100=0.5012
+step=137000 micro_steps=274000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7933 acc_all=0.5940 loss_corrupt=2.7933 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7933 loss_recon=2.7933 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4986 init_acc_corrupt=0.4671 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=137100 micro_steps=274200 elapsed=64.8s lr=3.000000e-04 loss_all=2.8122 acc_all=0.5916 loss_corrupt=2.8122 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8122 loss_recon=2.8122 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5009 init_acc_corrupt=0.4649 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=137200 micro_steps=274400 elapsed=63.3s lr=3.000000e-04 loss_all=2.7887 acc_all=0.5949 loss_corrupt=2.7887 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7887 loss_recon=2.7887 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4980 init_acc_corrupt=0.4680 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=137300 micro_steps=274600 elapsed=63.2s lr=3.000000e-04 loss_all=2.8010 acc_all=0.5929 loss_corrupt=2.8010 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.8010 loss_recon=2.8010 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.5003 init_acc_corrupt=0.4655 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=137400 micro_steps=274800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8115 acc_all=0.5918 loss_corrupt=2.8115 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8115 loss_recon=2.8115 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4644 init_gold_top10=0.4937 init_gold_top100=0.4995
+step=137500 micro_steps=275000 elapsed=60.3s lr=3.000000e-04 loss_all=2.8196 acc_all=0.5912 loss_corrupt=2.8196 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8196 loss_recon=2.8196 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5013 init_acc_corrupt=0.4633 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=137600 micro_steps=275200 elapsed=63.9s lr=3.000000e-04 loss_all=2.7939 acc_all=0.5944 loss_corrupt=2.7939 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7939 loss_recon=2.7939 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4978 init_acc_corrupt=0.4677 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=137700 micro_steps=275400 elapsed=64.0s lr=3.000000e-04 loss_all=2.7820 acc_all=0.5954 loss_corrupt=2.7820 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7820 loss_recon=2.7820 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4977 init_acc_corrupt=0.4682 init_gold_top10=0.4967 init_gold_top100=0.5025
+step=137800 micro_steps=275600 elapsed=63.9s lr=3.000000e-04 loss_all=2.7975 acc_all=0.5932 loss_corrupt=2.7975 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7975 loss_recon=2.7975 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5005 init_acc_corrupt=0.4650 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=137900 micro_steps=275800 elapsed=63.9s lr=3.000000e-04 loss_all=2.7911 acc_all=0.5945 loss_corrupt=2.7911 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7911 loss_recon=2.7911 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4986 init_acc_corrupt=0.4670 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=138000 micro_steps=276000 elapsed=64.0s lr=3.000000e-04 loss_all=2.8007 acc_all=0.5933 loss_corrupt=2.8007 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.8007 loss_recon=2.8007 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4988 init_acc_corrupt=0.4669 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=138100 micro_steps=276200 elapsed=65.5s lr=3.000000e-04 loss_all=2.7817 acc_all=0.5953 loss_corrupt=2.7817 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7817 loss_recon=2.7817 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4979 init_acc_corrupt=0.4680 init_gold_top10=0.4966 init_gold_top100=0.5023
+step=138200 micro_steps=276400 elapsed=64.0s lr=3.000000e-04 loss_all=2.8042 acc_all=0.5920 loss_corrupt=2.8042 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8042 loss_recon=2.8042 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5015 init_acc_corrupt=0.4640 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=138300 micro_steps=276600 elapsed=64.0s lr=3.000000e-04 loss_all=2.7929 acc_all=0.5942 loss_corrupt=2.7929 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7929 loss_recon=2.7929 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4986 init_acc_corrupt=0.4669 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=138400 micro_steps=276800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7826 acc_all=0.5953 loss_corrupt=2.7826 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7826 loss_recon=2.7826 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4978 init_acc_corrupt=0.4679 init_gold_top10=0.4966 init_gold_top100=0.5023
+step=138500 micro_steps=277000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7846 acc_all=0.5952 loss_corrupt=2.7846 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7846 loss_recon=2.7846 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4980 init_acc_corrupt=0.4673 init_gold_top10=0.4963 init_gold_top100=0.5021
+step=138600 micro_steps=277200 elapsed=64.2s lr=3.000000e-04 loss_all=2.8045 acc_all=0.5925 loss_corrupt=2.8045 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8045 loss_recon=2.8045 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4996 init_acc_corrupt=0.4662 init_gold_top10=0.4947 init_gold_top100=0.5006
+step=138700 micro_steps=277400 elapsed=64.0s lr=3.000000e-04 loss_all=2.8162 acc_all=0.5909 loss_corrupt=2.8162 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8162 loss_recon=2.8162 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5031 init_acc_corrupt=0.4620 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=138800 micro_steps=277600 elapsed=65.8s lr=3.000000e-04 loss_all=2.7841 acc_all=0.5950 loss_corrupt=2.7841 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7841 loss_recon=2.7841 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4976 init_acc_corrupt=0.4687 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=138900 micro_steps=277800 elapsed=63.7s lr=3.000000e-04 loss_all=2.8083 acc_all=0.5915 loss_corrupt=2.8083 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8083 loss_recon=2.8083 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5016 init_acc_corrupt=0.4637 init_gold_top10=0.4930 init_gold_top100=0.4986
+step=139000 micro_steps=278000 elapsed=64.0s lr=3.000000e-04 loss_all=2.8200 acc_all=0.5904 loss_corrupt=2.8200 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8200 loss_recon=2.8200 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5031 init_acc_corrupt=0.4623 init_gold_top10=0.4912 init_gold_top100=0.4970
+step=139100 micro_steps=278200 elapsed=65.4s lr=3.000000e-04 loss_all=2.8222 acc_all=0.5907 loss_corrupt=2.8222 acc_corrupt=0.5907 corrupt_frac=1.0000 loss=2.8222 loss_recon=2.8222 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5012 init_acc_corrupt=0.4637 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=139200 micro_steps=278400 elapsed=64.0s lr=3.000000e-04 loss_all=2.7922 acc_all=0.5940 loss_corrupt=2.7922 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7922 loss_recon=2.7922 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4993 init_acc_corrupt=0.4660 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=139300 micro_steps=278600 elapsed=64.0s lr=3.000000e-04 loss_all=2.7782 acc_all=0.5963 loss_corrupt=2.7782 acc_corrupt=0.5963 corrupt_frac=1.0000 loss=2.7782 loss_recon=2.7782 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4977 init_acc_corrupt=0.4678 init_gold_top10=0.4968 init_gold_top100=0.5025
+step=139400 micro_steps=278800 elapsed=64.0s lr=3.000000e-04 loss_all=2.8060 acc_all=0.5926 loss_corrupt=2.8060 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8060 loss_recon=2.8060 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4991 init_acc_corrupt=0.4664 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=139500 micro_steps=279000 elapsed=63.9s lr=3.000000e-04 loss_all=2.8064 acc_all=0.5926 loss_corrupt=2.8064 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8064 loss_recon=2.8064 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.4999 init_acc_corrupt=0.4655 init_gold_top10=0.4943 init_gold_top100=0.5002
+step=139600 micro_steps=279200 elapsed=64.0s lr=3.000000e-04 loss_all=2.8155 acc_all=0.5909 loss_corrupt=2.8155 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8155 loss_recon=2.8155 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5019 init_acc_corrupt=0.4630 init_gold_top10=0.4925 init_gold_top100=0.4983
+step=139700 micro_steps=279400 elapsed=64.1s lr=3.000000e-04 loss_all=2.7932 acc_all=0.5940 loss_corrupt=2.7932 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7932 loss_recon=2.7932 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4989 init_acc_corrupt=0.4671 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=139800 micro_steps=279600 elapsed=64.0s lr=3.000000e-04 loss_all=2.8450 acc_all=0.5871 loss_corrupt=2.8450 acc_corrupt=0.5871 corrupt_frac=1.0000 loss=2.8450 loss_recon=2.8450 loss_meanflow=0.0000 mean_model_t=0.4949 mean_corrupt_t=0.4949 wrong_frac=0.5054 init_acc_corrupt=0.4594 init_gold_top10=0.4888 init_gold_top100=0.4947
+step=139900 micro_steps=279800 elapsed=64.0s lr=3.000000e-04 loss_all=2.7743 acc_all=0.5964 loss_corrupt=2.7743 acc_corrupt=0.5964 corrupt_frac=1.0000 loss=2.7743 loss_recon=2.7743 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4963 init_acc_corrupt=0.4695 init_gold_top10=0.4981 init_gold_top100=0.5038
+step=140000 micro_steps=280000 elapsed=63.9s lr=3.000000e-04 loss_all=2.8332 acc_all=0.5887 loss_corrupt=2.8332 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8332 loss_recon=2.8332 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5034 init_acc_corrupt=0.4610 init_gold_top10=0.4908 init_gold_top100=0.4968
+step=140100 micro_steps=280200 elapsed=66.8s lr=3.000000e-04 loss_all=2.7872 acc_all=0.5944 loss_corrupt=2.7872 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7872 loss_recon=2.7872 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4984 init_acc_corrupt=0.4664 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=140200 micro_steps=280400 elapsed=64.0s lr=3.000000e-04 loss_all=2.8428 acc_all=0.5878 loss_corrupt=2.8428 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8428 loss_recon=2.8428 loss_meanflow=0.0000 mean_model_t=0.4951 mean_corrupt_t=0.4951 wrong_frac=0.5044 init_acc_corrupt=0.4605 init_gold_top10=0.4899 init_gold_top100=0.4958
+step=140300 micro_steps=280600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8085 acc_all=0.5924 loss_corrupt=2.8085 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8085 loss_recon=2.8085 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5009 init_acc_corrupt=0.4651 init_gold_top10=0.4934 init_gold_top100=0.4993
+step=140400 micro_steps=280800 elapsed=63.7s lr=3.000000e-04 loss_all=2.7706 acc_all=0.5964 loss_corrupt=2.7706 acc_corrupt=0.5964 corrupt_frac=1.0000 loss=2.7706 loss_recon=2.7706 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4980 init_acc_corrupt=0.4681 init_gold_top10=0.4965 init_gold_top100=0.5021
+step=140500 micro_steps=281000 elapsed=59.9s lr=3.000000e-04 loss_all=2.7722 acc_all=0.5967 loss_corrupt=2.7722 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7722 loss_recon=2.7722 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4972 init_acc_corrupt=0.4686 init_gold_top10=0.4973 init_gold_top100=0.5029
+step=140600 micro_steps=281200 elapsed=63.3s lr=3.000000e-04 loss_all=2.7460 acc_all=0.6000 loss_corrupt=2.7460 acc_corrupt=0.6000 corrupt_frac=1.0000 loss=2.7460 loss_recon=2.7460 loss_meanflow=0.0000 mean_model_t=0.5053 mean_corrupt_t=0.5053 wrong_frac=0.4942 init_acc_corrupt=0.4722 init_gold_top10=0.5004 init_gold_top100=0.5059
+step=140700 micro_steps=281400 elapsed=63.1s lr=3.000000e-04 loss_all=2.8173 acc_all=0.5908 loss_corrupt=2.8173 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8173 loss_recon=2.8173 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5024 init_acc_corrupt=0.4632 init_gold_top10=0.4920 init_gold_top100=0.4978
+step=140800 micro_steps=281600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7851 acc_all=0.5951 loss_corrupt=2.7851 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7851 loss_recon=2.7851 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4974 init_acc_corrupt=0.4682 init_gold_top10=0.4969 init_gold_top100=0.5028
+step=140900 micro_steps=281800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8073 acc_all=0.5921 loss_corrupt=2.8073 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8073 loss_recon=2.8073 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5019 init_acc_corrupt=0.4634 init_gold_top10=0.4926 init_gold_top100=0.4983
+step=141000 micro_steps=282000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7853 acc_all=0.5954 loss_corrupt=2.7853 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7853 loss_recon=2.7853 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4975 init_acc_corrupt=0.4685 init_gold_top10=0.4968 init_gold_top100=0.5026
+step=141100 micro_steps=282200 elapsed=64.8s lr=3.000000e-04 loss_all=2.7812 acc_all=0.5956 loss_corrupt=2.7812 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7812 loss_recon=2.7812 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4978 init_acc_corrupt=0.4684 init_gold_top10=0.4966 init_gold_top100=0.5023
+step=141200 micro_steps=282400 elapsed=64.0s lr=3.000000e-04 loss_all=2.7931 acc_all=0.5936 loss_corrupt=2.7931 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7931 loss_recon=2.7931 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5011 init_acc_corrupt=0.4651 init_gold_top10=0.4936 init_gold_top100=0.4991
+step=141300 micro_steps=282600 elapsed=64.1s lr=3.000000e-04 loss_all=2.7998 acc_all=0.5933 loss_corrupt=2.7998 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7998 loss_recon=2.7998 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5008 init_acc_corrupt=0.4652 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=141400 micro_steps=282800 elapsed=64.0s lr=3.000000e-04 loss_all=2.7908 acc_all=0.5938 loss_corrupt=2.7908 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7908 loss_recon=2.7908 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4994 init_acc_corrupt=0.4666 init_gold_top10=0.4951 init_gold_top100=0.5007
+step=141500 micro_steps=283000 elapsed=63.8s lr=3.000000e-04 loss_all=2.7900 acc_all=0.5943 loss_corrupt=2.7900 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7900 loss_recon=2.7900 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4994 init_acc_corrupt=0.4660 init_gold_top10=0.4952 init_gold_top100=0.5008
+step=141600 micro_steps=283200 elapsed=63.9s lr=3.000000e-04 loss_all=2.8298 acc_all=0.5892 loss_corrupt=2.8298 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8298 loss_recon=2.8298 loss_meanflow=0.0000 mean_model_t=0.4955 mean_corrupt_t=0.4955 wrong_frac=0.5045 init_acc_corrupt=0.4604 init_gold_top10=0.4898 init_gold_top100=0.4957
+step=141700 micro_steps=283400 elapsed=63.9s lr=3.000000e-04 loss_all=2.8026 acc_all=0.5934 loss_corrupt=2.8026 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.8026 loss_recon=2.8026 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4994 init_acc_corrupt=0.4657 init_gold_top10=0.4949 init_gold_top100=0.5007
+step=141800 micro_steps=283600 elapsed=64.2s lr=3.000000e-04 loss_all=2.7959 acc_all=0.5929 loss_corrupt=2.7959 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.7959 loss_recon=2.7959 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5003 init_acc_corrupt=0.4650 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=141900 micro_steps=283800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7933 acc_all=0.5939 loss_corrupt=2.7933 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7933 loss_recon=2.7933 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4988 init_acc_corrupt=0.4672 init_gold_top10=0.4956 init_gold_top100=0.5012
+step=142000 micro_steps=284000 elapsed=63.9s lr=3.000000e-04 loss_all=2.7979 acc_all=0.5929 loss_corrupt=2.7979 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.7979 loss_recon=2.7979 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5010 init_acc_corrupt=0.4646 init_gold_top10=0.4936 init_gold_top100=0.4992
+step=142100 micro_steps=284200 elapsed=65.3s lr=3.000000e-04 loss_all=2.7646 acc_all=0.5978 loss_corrupt=2.7646 acc_corrupt=0.5978 corrupt_frac=1.0000 loss=2.7646 loss_recon=2.7646 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4964 init_acc_corrupt=0.4696 init_gold_top10=0.4981 init_gold_top100=0.5038
+step=142200 micro_steps=284400 elapsed=63.9s lr=3.000000e-04 loss_all=2.7974 acc_all=0.5935 loss_corrupt=2.7974 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7974 loss_recon=2.7974 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4994 init_acc_corrupt=0.4659 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=142300 micro_steps=284600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7919 acc_all=0.5937 loss_corrupt=2.7919 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7919 loss_recon=2.7919 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.5002 init_acc_corrupt=0.4654 init_gold_top10=0.4943 init_gold_top100=0.5000
+step=142400 micro_steps=284800 elapsed=63.3s lr=3.000000e-04 loss_all=2.7802 acc_all=0.5955 loss_corrupt=2.7802 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7802 loss_recon=2.7802 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4990 init_acc_corrupt=0.4672 init_gold_top10=0.4956 init_gold_top100=0.5012
+step=142500 micro_steps=285000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8209 acc_all=0.5906 loss_corrupt=2.8209 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8209 loss_recon=2.8209 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5023 init_acc_corrupt=0.4627 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=142600 micro_steps=285200 elapsed=63.9s lr=3.000000e-04 loss_all=2.7708 acc_all=0.5968 loss_corrupt=2.7708 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7708 loss_recon=2.7708 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4967 init_acc_corrupt=0.4693 init_gold_top10=0.4978 init_gold_top100=0.5035
+step=142700 micro_steps=285400 elapsed=63.8s lr=3.000000e-04 loss_all=2.8283 acc_all=0.5895 loss_corrupt=2.8283 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8283 loss_recon=2.8283 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5036 init_acc_corrupt=0.4611 init_gold_top10=0.4907 init_gold_top100=0.4965
+step=142800 micro_steps=285600 elapsed=64.1s lr=3.000000e-04 loss_all=2.7456 acc_all=0.6002 loss_corrupt=2.7456 acc_corrupt=0.6002 corrupt_frac=1.0000 loss=2.7456 loss_recon=2.7456 loss_meanflow=0.0000 mean_model_t=0.5067 mean_corrupt_t=0.5067 wrong_frac=0.4931 init_acc_corrupt=0.4733 init_gold_top10=0.5014 init_gold_top100=0.5071
+step=142900 micro_steps=285800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8379 acc_all=0.5880 loss_corrupt=2.8379 acc_corrupt=0.5880 corrupt_frac=1.0000 loss=2.8379 loss_recon=2.8379 loss_meanflow=0.0000 mean_model_t=0.4960 mean_corrupt_t=0.4960 wrong_frac=0.5043 init_acc_corrupt=0.4606 init_gold_top10=0.4900 init_gold_top100=0.4958
+step=143000 micro_steps=286000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8073 acc_all=0.5925 loss_corrupt=2.8073 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8073 loss_recon=2.8073 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5001 init_acc_corrupt=0.4651 init_gold_top10=0.4942 init_gold_top100=0.5001
+step=143100 micro_steps=286200 elapsed=65.2s lr=3.000000e-04 loss_all=2.7955 acc_all=0.5934 loss_corrupt=2.7955 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7955 loss_recon=2.7955 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4990 init_acc_corrupt=0.4662 init_gold_top10=0.4955 init_gold_top100=0.5011
+step=143200 micro_steps=286400 elapsed=63.2s lr=3.000000e-04 loss_all=2.7833 acc_all=0.5953 loss_corrupt=2.7833 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7833 loss_recon=2.7833 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4979 init_acc_corrupt=0.4681 init_gold_top10=0.4966 init_gold_top100=0.5022
+step=143300 micro_steps=286600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8459 acc_all=0.5866 loss_corrupt=2.8459 acc_corrupt=0.5866 corrupt_frac=1.0000 loss=2.8459 loss_recon=2.8459 loss_meanflow=0.0000 mean_model_t=0.4923 mean_corrupt_t=0.4923 wrong_frac=0.5072 init_acc_corrupt=0.4579 init_gold_top10=0.4871 init_gold_top100=0.4930
+step=143400 micro_steps=286800 elapsed=63.6s lr=3.000000e-04 loss_all=2.8289 acc_all=0.5894 loss_corrupt=2.8289 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8289 loss_recon=2.8289 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5042 init_acc_corrupt=0.4610 init_gold_top10=0.4902 init_gold_top100=0.4959
+step=143500 micro_steps=287000 elapsed=60.1s lr=3.000000e-04 loss_all=2.7930 acc_all=0.5947 loss_corrupt=2.7930 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7930 loss_recon=2.7930 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4980 init_acc_corrupt=0.4672 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=143600 micro_steps=287200 elapsed=63.5s lr=3.000000e-04 loss_all=2.8027 acc_all=0.5926 loss_corrupt=2.8027 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8027 loss_recon=2.8027 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5012 init_acc_corrupt=0.4638 init_gold_top10=0.4932 init_gold_top100=0.4989
+step=143700 micro_steps=287400 elapsed=63.7s lr=3.000000e-04 loss_all=2.7744 acc_all=0.5967 loss_corrupt=2.7744 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7744 loss_recon=2.7744 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4973 init_acc_corrupt=0.4687 init_gold_top10=0.4972 init_gold_top100=0.5028
+step=143800 micro_steps=287600 elapsed=63.8s lr=3.000000e-04 loss_all=2.7676 acc_all=0.5969 loss_corrupt=2.7676 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7676 loss_recon=2.7676 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4967 init_acc_corrupt=0.4688 init_gold_top10=0.4979 init_gold_top100=0.5035
+step=143900 micro_steps=287800 elapsed=63.8s lr=3.000000e-04 loss_all=2.7565 acc_all=0.5989 loss_corrupt=2.7565 acc_corrupt=0.5989 corrupt_frac=1.0000 loss=2.7565 loss_recon=2.7565 loss_meanflow=0.0000 mean_model_t=0.5057 mean_corrupt_t=0.5057 wrong_frac=0.4949 init_acc_corrupt=0.4717 init_gold_top10=0.4998 init_gold_top100=0.5053
+step=144000 micro_steps=288000 elapsed=63.1s lr=3.000000e-04 loss_all=2.8153 acc_all=0.5908 loss_corrupt=2.8153 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8153 loss_recon=2.8153 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5024 init_acc_corrupt=0.4624 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=144100 micro_steps=288200 elapsed=64.4s lr=3.000000e-04 loss_all=2.7532 acc_all=0.5993 loss_corrupt=2.7532 acc_corrupt=0.5993 corrupt_frac=1.0000 loss=2.7532 loss_recon=2.7532 loss_meanflow=0.0000 mean_model_t=0.5060 mean_corrupt_t=0.5060 wrong_frac=0.4943 init_acc_corrupt=0.4720 init_gold_top10=0.5003 init_gold_top100=0.5059
+step=144200 micro_steps=288400 elapsed=64.0s lr=3.000000e-04 loss_all=2.7706 acc_all=0.5972 loss_corrupt=2.7706 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7706 loss_recon=2.7706 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4964 init_acc_corrupt=0.4693 init_gold_top10=0.4979 init_gold_top100=0.5037
+step=144300 micro_steps=288600 elapsed=64.1s lr=3.000000e-04 loss_all=2.7829 acc_all=0.5951 loss_corrupt=2.7829 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7829 loss_recon=2.7829 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4985 init_acc_corrupt=0.4671 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=144400 micro_steps=288800 elapsed=64.0s lr=3.000000e-04 loss_all=2.7986 acc_all=0.5934 loss_corrupt=2.7986 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7986 loss_recon=2.7986 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4990 init_acc_corrupt=0.4660 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=144500 micro_steps=289000 elapsed=63.6s lr=3.000000e-04 loss_all=2.8131 acc_all=0.5912 loss_corrupt=2.8131 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8131 loss_recon=2.8131 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5021 init_acc_corrupt=0.4631 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=144600 micro_steps=289200 elapsed=63.0s lr=3.000000e-04 loss_all=2.8107 acc_all=0.5910 loss_corrupt=2.8107 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8107 loss_recon=2.8107 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5016 init_acc_corrupt=0.4629 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=144700 micro_steps=289400 elapsed=63.1s lr=3.000000e-04 loss_all=2.8410 acc_all=0.5882 loss_corrupt=2.8410 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8410 loss_recon=2.8410 loss_meanflow=0.0000 mean_model_t=0.4950 mean_corrupt_t=0.4950 wrong_frac=0.5049 init_acc_corrupt=0.4597 init_gold_top10=0.4892 init_gold_top100=0.4952
+step=144800 micro_steps=289600 elapsed=63.2s lr=3.000000e-04 loss_all=2.7879 acc_all=0.5946 loss_corrupt=2.7879 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7879 loss_recon=2.7879 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4994 init_acc_corrupt=0.4662 init_gold_top10=0.4951 init_gold_top100=0.5007
+step=144900 micro_steps=289800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8478 acc_all=0.5863 loss_corrupt=2.8478 acc_corrupt=0.5863 corrupt_frac=1.0000 loss=2.8478 loss_recon=2.8478 loss_meanflow=0.0000 mean_model_t=0.4935 mean_corrupt_t=0.4935 wrong_frac=0.5067 init_acc_corrupt=0.4578 init_gold_top10=0.4877 init_gold_top100=0.4934
+step=145000 micro_steps=290000 elapsed=63.3s lr=3.000000e-04 loss_all=2.7879 acc_all=0.5945 loss_corrupt=2.7879 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7879 loss_recon=2.7879 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4981 init_acc_corrupt=0.4672 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=145100 micro_steps=290200 elapsed=64.9s lr=3.000000e-04 loss_all=2.7771 acc_all=0.5957 loss_corrupt=2.7771 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7771 loss_recon=2.7771 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4989 init_acc_corrupt=0.4665 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=145200 micro_steps=290400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8343 acc_all=0.5885 loss_corrupt=2.8343 acc_corrupt=0.5885 corrupt_frac=1.0000 loss=2.8343 loss_recon=2.8343 loss_meanflow=0.0000 mean_model_t=0.4955 mean_corrupt_t=0.4955 wrong_frac=0.5046 init_acc_corrupt=0.4601 init_gold_top10=0.4896 init_gold_top100=0.4955
+step=145300 micro_steps=290600 elapsed=63.3s lr=3.000000e-04 loss_all=2.7777 acc_all=0.5962 loss_corrupt=2.7777 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7777 loss_recon=2.7777 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4975 init_acc_corrupt=0.4685 init_gold_top10=0.4971 init_gold_top100=0.5027
+step=145400 micro_steps=290800 elapsed=63.1s lr=3.000000e-04 loss_all=2.8157 acc_all=0.5909 loss_corrupt=2.8157 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8157 loss_recon=2.8157 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5022 init_acc_corrupt=0.4623 init_gold_top10=0.4921 init_gold_top100=0.4979
+step=145500 micro_steps=291000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8235 acc_all=0.5899 loss_corrupt=2.8235 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8235 loss_recon=2.8235 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5030 init_acc_corrupt=0.4620 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=145600 micro_steps=291200 elapsed=63.6s lr=3.000000e-04 loss_all=2.7982 acc_all=0.5933 loss_corrupt=2.7982 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7982 loss_recon=2.7982 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5003 init_acc_corrupt=0.4655 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=145700 micro_steps=291400 elapsed=63.3s lr=3.000000e-04 loss_all=2.8086 acc_all=0.5922 loss_corrupt=2.8086 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8086 loss_recon=2.8086 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5010 init_acc_corrupt=0.4644 init_gold_top10=0.4934 init_gold_top100=0.4990
+step=145800 micro_steps=291600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8435 acc_all=0.5872 loss_corrupt=2.8435 acc_corrupt=0.5872 corrupt_frac=1.0000 loss=2.8435 loss_recon=2.8435 loss_meanflow=0.0000 mean_model_t=0.4944 mean_corrupt_t=0.4944 wrong_frac=0.5059 init_acc_corrupt=0.4585 init_gold_top10=0.4883 init_gold_top100=0.4942
+step=145900 micro_steps=291800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8361 acc_all=0.5882 loss_corrupt=2.8361 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8361 loss_recon=2.8361 loss_meanflow=0.0000 mean_model_t=0.4944 mean_corrupt_t=0.4944 wrong_frac=0.5054 init_acc_corrupt=0.4599 init_gold_top10=0.4889 init_gold_top100=0.4948
+step=146000 micro_steps=292000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8057 acc_all=0.5920 loss_corrupt=2.8057 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8057 loss_recon=2.8057 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5011 init_acc_corrupt=0.4645 init_gold_top10=0.4933 init_gold_top100=0.4991
+step=146100 micro_steps=292200 elapsed=64.8s lr=3.000000e-04 loss_all=2.7855 acc_all=0.5950 loss_corrupt=2.7855 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7855 loss_recon=2.7855 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4984 init_acc_corrupt=0.4669 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=146200 micro_steps=292400 elapsed=62.8s lr=3.000000e-04 loss_all=2.8122 acc_all=0.5918 loss_corrupt=2.8122 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8122 loss_recon=2.8122 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5009 init_acc_corrupt=0.4642 init_gold_top10=0.4933 init_gold_top100=0.4992
+step=146300 micro_steps=292600 elapsed=62.8s lr=3.000000e-04 loss_all=2.7873 acc_all=0.5947 loss_corrupt=2.7873 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7873 loss_recon=2.7873 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4989 init_acc_corrupt=0.4672 init_gold_top10=0.4957 init_gold_top100=0.5012
+step=146400 micro_steps=292800 elapsed=63.1s lr=3.000000e-04 loss_all=2.7941 acc_all=0.5937 loss_corrupt=2.7941 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7941 loss_recon=2.7941 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5000 init_acc_corrupt=0.4657 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=146500 micro_steps=293000 elapsed=62.6s lr=3.000000e-04 loss_all=2.7922 acc_all=0.5940 loss_corrupt=2.7922 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7922 loss_recon=2.7922 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4988 init_acc_corrupt=0.4659 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=146600 micro_steps=293200 elapsed=59.6s lr=3.000000e-04 loss_all=2.7645 acc_all=0.5975 loss_corrupt=2.7645 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7645 loss_recon=2.7645 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4967 init_acc_corrupt=0.4695 init_gold_top10=0.4979 init_gold_top100=0.5034
+step=146700 micro_steps=293400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8210 acc_all=0.5903 loss_corrupt=2.8210 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8210 loss_recon=2.8210 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5029 init_acc_corrupt=0.4624 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=146800 micro_steps=293600 elapsed=62.8s lr=3.000000e-04 loss_all=2.7775 acc_all=0.5960 loss_corrupt=2.7775 acc_corrupt=0.5960 corrupt_frac=1.0000 loss=2.7775 loss_recon=2.7775 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4970 init_acc_corrupt=0.4687 init_gold_top10=0.4975 init_gold_top100=0.5032
+step=146900 micro_steps=293800 elapsed=62.8s lr=3.000000e-04 loss_all=2.7882 acc_all=0.5943 loss_corrupt=2.7882 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7882 loss_recon=2.7882 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4992 init_acc_corrupt=0.4663 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=147000 micro_steps=294000 elapsed=62.5s lr=3.000000e-04 loss_all=2.8271 acc_all=0.5895 loss_corrupt=2.8271 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8271 loss_recon=2.8271 loss_meanflow=0.0000 mean_model_t=0.4959 mean_corrupt_t=0.4959 wrong_frac=0.5039 init_acc_corrupt=0.4615 init_gold_top10=0.4904 init_gold_top100=0.4963
+step=147100 micro_steps=294200 elapsed=64.1s lr=3.000000e-04 loss_all=2.7931 acc_all=0.5936 loss_corrupt=2.7931 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7931 loss_recon=2.7931 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5001 init_acc_corrupt=0.4652 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=147200 micro_steps=294400 elapsed=62.6s lr=3.000000e-04 loss_all=2.8044 acc_all=0.5927 loss_corrupt=2.8044 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8044 loss_recon=2.8044 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5001 init_acc_corrupt=0.4656 init_gold_top10=0.4940 init_gold_top100=0.4999
+step=147300 micro_steps=294600 elapsed=62.7s lr=3.000000e-04 loss_all=2.7776 acc_all=0.5958 loss_corrupt=2.7776 acc_corrupt=0.5958 corrupt_frac=1.0000 loss=2.7776 loss_recon=2.7776 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4979 init_acc_corrupt=0.4676 init_gold_top10=0.4965 init_gold_top100=0.5023
+step=147400 micro_steps=294800 elapsed=62.9s lr=3.000000e-04 loss_all=2.7516 acc_all=0.5995 loss_corrupt=2.7516 acc_corrupt=0.5995 corrupt_frac=1.0000 loss=2.7516 loss_recon=2.7516 loss_meanflow=0.0000 mean_model_t=0.5057 mean_corrupt_t=0.5057 wrong_frac=0.4943 init_acc_corrupt=0.4719 init_gold_top10=0.5002 init_gold_top100=0.5058
+step=147500 micro_steps=295000 elapsed=62.7s lr=3.000000e-04 loss_all=2.7850 acc_all=0.5951 loss_corrupt=2.7850 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7850 loss_recon=2.7850 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4985 init_acc_corrupt=0.4675 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=147600 micro_steps=295200 elapsed=63.2s lr=3.000000e-04 loss_all=2.7940 acc_all=0.5934 loss_corrupt=2.7940 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7940 loss_recon=2.7940 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5003 init_acc_corrupt=0.4655 init_gold_top10=0.4940 init_gold_top100=0.4999
+step=147700 micro_steps=295400 elapsed=63.6s lr=3.000000e-04 loss_all=2.7921 acc_all=0.5940 loss_corrupt=2.7921 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7921 loss_recon=2.7921 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4999 init_acc_corrupt=0.4663 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=147800 micro_steps=295600 elapsed=63.3s lr=3.000000e-04 loss_all=2.7927 acc_all=0.5941 loss_corrupt=2.7927 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7927 loss_recon=2.7927 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4996 init_acc_corrupt=0.4659 init_gold_top10=0.4947 init_gold_top100=0.5006
+step=147900 micro_steps=295800 elapsed=63.1s lr=3.000000e-04 loss_all=2.7601 acc_all=0.5982 loss_corrupt=2.7601 acc_corrupt=0.5982 corrupt_frac=1.0000 loss=2.7601 loss_recon=2.7601 loss_meanflow=0.0000 mean_model_t=0.5036 mean_corrupt_t=0.5036 wrong_frac=0.4965 init_acc_corrupt=0.4690 init_gold_top10=0.4980 init_gold_top100=0.5037
+step=148000 micro_steps=296000 elapsed=63.1s lr=3.000000e-04 loss_all=2.7696 acc_all=0.5969 loss_corrupt=2.7696 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7696 loss_recon=2.7696 loss_meanflow=0.0000 mean_model_t=0.5038 mean_corrupt_t=0.5038 wrong_frac=0.4964 init_acc_corrupt=0.4695 init_gold_top10=0.4982 init_gold_top100=0.5037
+step=148100 micro_steps=296200 elapsed=64.7s lr=3.000000e-04 loss_all=2.7793 acc_all=0.5955 loss_corrupt=2.7793 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7793 loss_recon=2.7793 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4983 init_acc_corrupt=0.4675 init_gold_top10=0.4963 init_gold_top100=0.5019
+step=148200 micro_steps=296400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7673 acc_all=0.5966 loss_corrupt=2.7673 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7673 loss_recon=2.7673 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4980 init_acc_corrupt=0.4678 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=148300 micro_steps=296600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7616 acc_all=0.5981 loss_corrupt=2.7616 acc_corrupt=0.5981 corrupt_frac=1.0000 loss=2.7616 loss_recon=2.7616 loss_meanflow=0.0000 mean_model_t=0.5044 mean_corrupt_t=0.5044 wrong_frac=0.4957 init_acc_corrupt=0.4700 init_gold_top10=0.4988 init_gold_top100=0.5044
+step=148400 micro_steps=296800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7755 acc_all=0.5960 loss_corrupt=2.7755 acc_corrupt=0.5960 corrupt_frac=1.0000 loss=2.7755 loss_recon=2.7755 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4968 init_acc_corrupt=0.4683 init_gold_top10=0.4977 init_gold_top100=0.5034
+step=148500 micro_steps=297000 elapsed=63.1s lr=3.000000e-04 loss_all=2.7773 acc_all=0.5962 loss_corrupt=2.7773 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7773 loss_recon=2.7773 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4970 init_acc_corrupt=0.4683 init_gold_top10=0.4975 init_gold_top100=0.5031
+step=148600 micro_steps=297200 elapsed=63.2s lr=3.000000e-04 loss_all=2.7956 acc_all=0.5936 loss_corrupt=2.7956 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7956 loss_recon=2.7956 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4999 init_acc_corrupt=0.4654 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=148700 micro_steps=297400 elapsed=62.7s lr=3.000000e-04 loss_all=2.7934 acc_all=0.5937 loss_corrupt=2.7934 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7934 loss_recon=2.7934 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4998 init_acc_corrupt=0.4657 init_gold_top10=0.4945 init_gold_top100=0.5003
+step=148800 micro_steps=297600 elapsed=63.3s lr=3.000000e-04 loss_all=2.7813 acc_all=0.5954 loss_corrupt=2.7813 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7813 loss_recon=2.7813 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4981 init_acc_corrupt=0.4674 init_gold_top10=0.4965 init_gold_top100=0.5021
+step=148900 micro_steps=297800 elapsed=63.3s lr=3.000000e-04 loss_all=2.7789 acc_all=0.5959 loss_corrupt=2.7789 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7789 loss_recon=2.7789 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4978 init_acc_corrupt=0.4677 init_gold_top10=0.4968 init_gold_top100=0.5024
+step=149000 micro_steps=298000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7972 acc_all=0.5935 loss_corrupt=2.7972 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7972 loss_recon=2.7972 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.4997 init_acc_corrupt=0.4653 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=149100 micro_steps=298200 elapsed=64.8s lr=3.000000e-04 loss_all=2.7466 acc_all=0.5998 loss_corrupt=2.7466 acc_corrupt=0.5998 corrupt_frac=1.0000 loss=2.7466 loss_recon=2.7466 loss_meanflow=0.0000 mean_model_t=0.5055 mean_corrupt_t=0.5055 wrong_frac=0.4948 init_acc_corrupt=0.4715 init_gold_top10=0.4999 init_gold_top100=0.5054
+step=149200 micro_steps=298400 elapsed=63.1s lr=3.000000e-04 loss_all=2.8062 acc_all=0.5914 loss_corrupt=2.8062 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8062 loss_recon=2.8062 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5025 init_acc_corrupt=0.4631 init_gold_top10=0.4920 init_gold_top100=0.4976
+step=149300 micro_steps=298600 elapsed=63.2s lr=3.000000e-04 loss_all=2.8136 acc_all=0.5909 loss_corrupt=2.8136 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8136 loss_recon=2.8136 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5028 init_acc_corrupt=0.4627 init_gold_top10=0.4917 init_gold_top100=0.4974
+step=149400 micro_steps=298800 elapsed=63.2s lr=3.000000e-04 loss_all=2.7744 acc_all=0.5962 loss_corrupt=2.7744 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7744 loss_recon=2.7744 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4977 init_acc_corrupt=0.4680 init_gold_top10=0.4967 init_gold_top100=0.5025
+step=149500 micro_steps=299000 elapsed=63.2s lr=3.000000e-04 loss_all=2.8076 acc_all=0.5918 loss_corrupt=2.8076 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8076 loss_recon=2.8076 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5026 init_acc_corrupt=0.4628 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=149600 micro_steps=299200 elapsed=60.0s lr=3.000000e-04 loss_all=2.8065 acc_all=0.5920 loss_corrupt=2.8065 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8065 loss_recon=2.8065 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5017 init_acc_corrupt=0.4637 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=149700 micro_steps=299400 elapsed=63.3s lr=3.000000e-04 loss_all=2.8371 acc_all=0.5878 loss_corrupt=2.8371 acc_corrupt=0.5878 corrupt_frac=1.0000 loss=2.8371 loss_recon=2.8371 loss_meanflow=0.0000 mean_model_t=0.4945 mean_corrupt_t=0.4945 wrong_frac=0.5057 init_acc_corrupt=0.4589 init_gold_top10=0.4886 init_gold_top100=0.4944
+step=149800 micro_steps=299600 elapsed=63.2s lr=3.000000e-04 loss_all=2.7917 acc_all=0.5940 loss_corrupt=2.7917 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7917 loss_recon=2.7917 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4997 init_acc_corrupt=0.4654 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=149900 micro_steps=299800 elapsed=63.3s lr=3.000000e-04 loss_all=2.7927 acc_all=0.5937 loss_corrupt=2.7927 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7927 loss_recon=2.7927 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4995 init_acc_corrupt=0.4657 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=150000 micro_steps=300000 elapsed=63.1s lr=3.000000e-04 loss_all=2.7998 acc_all=0.5930 loss_corrupt=2.7998 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.7998 loss_recon=2.7998 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5006 init_acc_corrupt=0.4649 init_gold_top10=0.4936 init_gold_top100=0.4995
+step=150100 micro_steps=300200 elapsed=64.6s lr=3.000000e-04 loss_all=2.8097 acc_all=0.5918 loss_corrupt=2.8097 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8097 loss_recon=2.8097 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5016 init_acc_corrupt=0.4638 init_gold_top10=0.4927 init_gold_top100=0.4984
+step=150200 micro_steps=300400 elapsed=63.3s lr=3.000000e-04 loss_all=2.7989 acc_all=0.5931 loss_corrupt=2.7989 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.7989 loss_recon=2.7989 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4993 init_acc_corrupt=0.4661 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=150300 micro_steps=300600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8064 acc_all=0.5921 loss_corrupt=2.8064 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8064 loss_recon=2.8064 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5013 init_acc_corrupt=0.4643 init_gold_top10=0.4932 init_gold_top100=0.4989
+step=150400 micro_steps=300800 elapsed=63.2s lr=3.000000e-04 loss_all=2.8182 acc_all=0.5905 loss_corrupt=2.8182 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8182 loss_recon=2.8182 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5033 init_acc_corrupt=0.4615 init_gold_top10=0.4912 init_gold_top100=0.4968
+step=150500 micro_steps=301000 elapsed=62.8s lr=3.000000e-04 loss_all=2.8026 acc_all=0.5929 loss_corrupt=2.8026 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.8026 loss_recon=2.8026 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4994 init_acc_corrupt=0.4659 init_gold_top10=0.4949 init_gold_top100=0.5007
+step=150600 micro_steps=301200 elapsed=63.0s lr=3.000000e-04 loss_all=2.8239 acc_all=0.5895 loss_corrupt=2.8239 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8239 loss_recon=2.8239 loss_meanflow=0.0000 mean_model_t=0.4962 mean_corrupt_t=0.4962 wrong_frac=0.5042 init_acc_corrupt=0.4601 init_gold_top10=0.4901 init_gold_top100=0.4959
+step=150700 micro_steps=301400 elapsed=63.0s lr=3.000000e-04 loss_all=2.7920 acc_all=0.5939 loss_corrupt=2.7920 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7920 loss_recon=2.7920 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5002 init_acc_corrupt=0.4654 init_gold_top10=0.4943 init_gold_top100=0.5000
+step=150800 micro_steps=301600 elapsed=63.1s lr=3.000000e-04 loss_all=2.8064 acc_all=0.5924 loss_corrupt=2.8064 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8064 loss_recon=2.8064 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5016 init_acc_corrupt=0.4642 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=150900 micro_steps=301800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7790 acc_all=0.5957 loss_corrupt=2.7790 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7790 loss_recon=2.7790 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4972 init_acc_corrupt=0.4684 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=151000 micro_steps=302000 elapsed=63.1s lr=3.000000e-04 loss_all=2.7697 acc_all=0.5967 loss_corrupt=2.7697 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7697 loss_recon=2.7697 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4966 init_acc_corrupt=0.4693 init_gold_top10=0.4979 init_gold_top100=0.5036
+step=151100 micro_steps=302200 elapsed=64.2s lr=3.000000e-04 loss_all=2.7727 acc_all=0.5966 loss_corrupt=2.7727 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7727 loss_recon=2.7727 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4977 init_acc_corrupt=0.4681 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=151200 micro_steps=302400 elapsed=62.8s lr=3.000000e-04 loss_all=2.7964 acc_all=0.5933 loss_corrupt=2.7964 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7964 loss_recon=2.7964 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.4999 init_acc_corrupt=0.4653 init_gold_top10=0.4944 init_gold_top100=0.5003
+step=151300 micro_steps=302600 elapsed=62.6s lr=3.000000e-04 loss_all=2.8400 acc_all=0.5876 loss_corrupt=2.8400 acc_corrupt=0.5876 corrupt_frac=1.0000 loss=2.8400 loss_recon=2.8400 loss_meanflow=0.0000 mean_model_t=0.4944 mean_corrupt_t=0.4944 wrong_frac=0.5057 init_acc_corrupt=0.4595 init_gold_top10=0.4886 init_gold_top100=0.4944
+step=151400 micro_steps=302800 elapsed=63.1s lr=3.000000e-04 loss_all=2.8092 acc_all=0.5916 loss_corrupt=2.8092 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8092 loss_recon=2.8092 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5006 init_acc_corrupt=0.4643 init_gold_top10=0.4937 init_gold_top100=0.4996
+step=151500 micro_steps=303000 elapsed=63.3s lr=3.000000e-04 loss_all=2.8010 acc_all=0.5928 loss_corrupt=2.8010 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8010 loss_recon=2.8010 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5007 init_acc_corrupt=0.4645 init_gold_top10=0.4937 init_gold_top100=0.4995
+step=151600 micro_steps=303200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7676 acc_all=0.5964 loss_corrupt=2.7676 acc_corrupt=0.5964 corrupt_frac=1.0000 loss=2.7676 loss_recon=2.7676 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4977 init_acc_corrupt=0.4680 init_gold_top10=0.4968 init_gold_top100=0.5024
+step=151700 micro_steps=303400 elapsed=63.8s lr=3.000000e-04 loss_all=2.7931 acc_all=0.5940 loss_corrupt=2.7931 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7931 loss_recon=2.7931 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4999 init_acc_corrupt=0.4661 init_gold_top10=0.4944 init_gold_top100=0.5003
+step=151800 micro_steps=303600 elapsed=64.0s lr=3.000000e-04 loss_all=2.7785 acc_all=0.5959 loss_corrupt=2.7785 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7785 loss_recon=2.7785 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4979 init_acc_corrupt=0.4684 init_gold_top10=0.4966 init_gold_top100=0.5022
+step=151900 micro_steps=303800 elapsed=63.9s lr=3.000000e-04 loss_all=2.8169 acc_all=0.5909 loss_corrupt=2.8169 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8169 loss_recon=2.8169 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5025 init_acc_corrupt=0.4627 init_gold_top10=0.4918 init_gold_top100=0.4976
+step=152000 micro_steps=304000 elapsed=64.1s lr=3.000000e-04 loss_all=2.7727 acc_all=0.5967 loss_corrupt=2.7727 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7727 loss_recon=2.7727 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4975 init_acc_corrupt=0.4682 init_gold_top10=0.4970 init_gold_top100=0.5026
+step=152100 micro_steps=304200 elapsed=65.4s lr=3.000000e-04 loss_all=2.8193 acc_all=0.5906 loss_corrupt=2.8193 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8193 loss_recon=2.8193 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5027 init_acc_corrupt=0.4627 init_gold_top10=0.4915 init_gold_top100=0.4974
+step=152200 micro_steps=304400 elapsed=64.3s lr=3.000000e-04 loss_all=2.7884 acc_all=0.5942 loss_corrupt=2.7884 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7884 loss_recon=2.7884 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5000 init_acc_corrupt=0.4654 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=152300 micro_steps=304600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8228 acc_all=0.5902 loss_corrupt=2.8228 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8228 loss_recon=2.8228 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5027 init_acc_corrupt=0.4618 init_gold_top10=0.4914 init_gold_top100=0.4974
+step=152400 micro_steps=304800 elapsed=63.7s lr=3.000000e-04 loss_all=2.8219 acc_all=0.5905 loss_corrupt=2.8219 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8219 loss_recon=2.8219 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5025 init_acc_corrupt=0.4625 init_gold_top10=0.4917 init_gold_top100=0.4976
+step=152500 micro_steps=305000 elapsed=63.2s lr=3.000000e-04 loss_all=2.7984 acc_all=0.5931 loss_corrupt=2.7984 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.7984 loss_recon=2.7984 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5008 init_acc_corrupt=0.4647 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=152600 micro_steps=305200 elapsed=63.4s lr=3.000000e-04 loss_all=2.7763 acc_all=0.5963 loss_corrupt=2.7763 acc_corrupt=0.5963 corrupt_frac=1.0000 loss=2.7763 loss_recon=2.7763 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4977 init_acc_corrupt=0.4678 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=152700 micro_steps=305400 elapsed=61.8s lr=3.000000e-04 loss_all=2.7821 acc_all=0.5953 loss_corrupt=2.7821 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7821 loss_recon=2.7821 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4987 init_acc_corrupt=0.4670 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=152800 micro_steps=305600 elapsed=62.9s lr=3.000000e-04 loss_all=2.7866 acc_all=0.5947 loss_corrupt=2.7866 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7866 loss_recon=2.7866 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4990 init_acc_corrupt=0.4666 init_gold_top10=0.4954 init_gold_top100=0.5012
+step=152900 micro_steps=305800 elapsed=62.8s lr=3.000000e-04 loss_all=2.7937 acc_all=0.5937 loss_corrupt=2.7937 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7937 loss_recon=2.7937 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4997 init_acc_corrupt=0.4656 init_gold_top10=0.4948 init_gold_top100=0.5004
+step=153000 micro_steps=306000 elapsed=62.9s lr=3.000000e-04 loss_all=2.8148 acc_all=0.5912 loss_corrupt=2.8148 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8148 loss_recon=2.8148 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5018 init_acc_corrupt=0.4634 init_gold_top10=0.4926 init_gold_top100=0.4984
+step=153100 micro_steps=306200 elapsed=64.4s lr=3.000000e-04 loss_all=2.7777 acc_all=0.5960 loss_corrupt=2.7777 acc_corrupt=0.5960 corrupt_frac=1.0000 loss=2.7777 loss_recon=2.7777 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4977 init_acc_corrupt=0.4677 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=153200 micro_steps=306400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8179 acc_all=0.5908 loss_corrupt=2.8179 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8179 loss_recon=2.8179 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5024 init_acc_corrupt=0.4623 init_gold_top10=0.4918 init_gold_top100=0.4976
+step=153300 micro_steps=306600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7906 acc_all=0.5941 loss_corrupt=2.7906 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7906 loss_recon=2.7906 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5010 init_acc_corrupt=0.4648 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=153400 micro_steps=306800 elapsed=64.0s lr=3.000000e-04 loss_all=2.8046 acc_all=0.5926 loss_corrupt=2.8046 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8046 loss_recon=2.8046 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5011 init_acc_corrupt=0.4642 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=153500 micro_steps=307000 elapsed=63.8s lr=3.000000e-04 loss_all=2.8051 acc_all=0.5923 loss_corrupt=2.8051 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8051 loss_recon=2.8051 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5012 init_acc_corrupt=0.4648 init_gold_top10=0.4932 init_gold_top100=0.4989
+step=153600 micro_steps=307200 elapsed=63.6s lr=3.000000e-04 loss_all=2.7814 acc_all=0.5953 loss_corrupt=2.7814 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7814 loss_recon=2.7814 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4993 init_acc_corrupt=0.4672 init_gold_top10=0.4952 init_gold_top100=0.5008
+step=153700 micro_steps=307400 elapsed=63.1s lr=3.000000e-04 loss_all=2.7727 acc_all=0.5968 loss_corrupt=2.7727 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7727 loss_recon=2.7727 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4968 init_acc_corrupt=0.4694 init_gold_top10=0.4977 init_gold_top100=0.5033
+step=153800 micro_steps=307600 elapsed=62.7s lr=3.000000e-04 loss_all=2.7680 acc_all=0.5973 loss_corrupt=2.7680 acc_corrupt=0.5973 corrupt_frac=1.0000 loss=2.7680 loss_recon=2.7680 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4964 init_acc_corrupt=0.4697 init_gold_top10=0.4982 init_gold_top100=0.5038
+step=153900 micro_steps=307800 elapsed=62.8s lr=3.000000e-04 loss_all=2.8115 acc_all=0.5915 loss_corrupt=2.8115 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8115 loss_recon=2.8115 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5014 init_acc_corrupt=0.4642 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=154000 micro_steps=308000 elapsed=63.1s lr=3.000000e-04 loss_all=2.8027 acc_all=0.5926 loss_corrupt=2.8027 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8027 loss_recon=2.8027 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5012 init_acc_corrupt=0.4641 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=154100 micro_steps=308200 elapsed=64.2s lr=3.000000e-04 loss_all=2.8122 acc_all=0.5918 loss_corrupt=2.8122 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8122 loss_recon=2.8122 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5012 init_acc_corrupt=0.4640 init_gold_top10=0.4930 init_gold_top100=0.4989
+step=154200 micro_steps=308400 elapsed=62.8s lr=3.000000e-04 loss_all=2.8126 acc_all=0.5914 loss_corrupt=2.8126 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8126 loss_recon=2.8126 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5028 init_acc_corrupt=0.4621 init_gold_top10=0.4917 init_gold_top100=0.4974
+step=154300 micro_steps=308600 elapsed=63.6s lr=3.000000e-04 loss_all=2.7585 acc_all=0.5981 loss_corrupt=2.7585 acc_corrupt=0.5981 corrupt_frac=1.0000 loss=2.7585 loss_recon=2.7585 loss_meanflow=0.0000 mean_model_t=0.5038 mean_corrupt_t=0.5038 wrong_frac=0.4957 init_acc_corrupt=0.4702 init_gold_top10=0.4987 init_gold_top100=0.5044
+step=154400 micro_steps=308800 elapsed=63.9s lr=3.000000e-04 loss_all=2.7800 acc_all=0.5956 loss_corrupt=2.7800 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7800 loss_recon=2.7800 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4984 init_acc_corrupt=0.4668 init_gold_top10=0.4960 init_gold_top100=0.5018
+step=154500 micro_steps=309000 elapsed=65.1s lr=3.000000e-04 loss_all=2.8068 acc_all=0.5919 loss_corrupt=2.8068 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8068 loss_recon=2.8068 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5020 init_acc_corrupt=0.4633 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=154600 micro_steps=309200 elapsed=98.3s lr=3.000000e-04 loss_all=2.8084 acc_all=0.5916 loss_corrupt=2.8084 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8084 loss_recon=2.8084 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5029 init_acc_corrupt=0.4623 init_gold_top10=0.4915 init_gold_top100=0.4972
+step=154700 micro_steps=309400 elapsed=68.9s lr=3.000000e-04 loss_all=2.8344 acc_all=0.5884 loss_corrupt=2.8344 acc_corrupt=0.5884 corrupt_frac=1.0000 loss=2.8344 loss_recon=2.8344 loss_meanflow=0.0000 mean_model_t=0.4941 mean_corrupt_t=0.4941 wrong_frac=0.5056 init_acc_corrupt=0.4593 init_gold_top10=0.4886 init_gold_top100=0.4945
+step=154800 micro_steps=309600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8268 acc_all=0.5898 loss_corrupt=2.8268 acc_corrupt=0.5898 corrupt_frac=1.0000 loss=2.8268 loss_recon=2.8268 loss_meanflow=0.0000 mean_model_t=0.4960 mean_corrupt_t=0.4960 wrong_frac=0.5039 init_acc_corrupt=0.4615 init_gold_top10=0.4904 init_gold_top100=0.4963
+step=154900 micro_steps=309800 elapsed=63.3s lr=3.000000e-04 loss_all=2.8013 acc_all=0.5925 loss_corrupt=2.8013 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8013 loss_recon=2.8013 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5019 init_acc_corrupt=0.4636 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=155000 micro_steps=310000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7529 acc_all=0.5993 loss_corrupt=2.7529 acc_corrupt=0.5993 corrupt_frac=1.0000 loss=2.7529 loss_recon=2.7529 loss_meanflow=0.0000 mean_model_t=0.5056 mean_corrupt_t=0.5056 wrong_frac=0.4944 init_acc_corrupt=0.4717 init_gold_top10=0.5001 init_gold_top100=0.5058
+step=155100 micro_steps=310200 elapsed=65.1s lr=3.000000e-04 loss_all=2.8007 acc_all=0.5929 loss_corrupt=2.8007 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.8007 loss_recon=2.8007 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5011 init_acc_corrupt=0.4643 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=155200 micro_steps=310400 elapsed=63.1s lr=3.000000e-04 loss_all=2.7696 acc_all=0.5968 loss_corrupt=2.7696 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7696 loss_recon=2.7696 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4982 init_acc_corrupt=0.4679 init_gold_top10=0.4964 init_gold_top100=0.5020
+step=155300 micro_steps=310600 elapsed=62.9s lr=3.000000e-04 loss_all=2.7861 acc_all=0.5949 loss_corrupt=2.7861 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7861 loss_recon=2.7861 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4984 init_acc_corrupt=0.4673 init_gold_top10=0.4961 init_gold_top100=0.5017
+step=155400 micro_steps=310800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8309 acc_all=0.5887 loss_corrupt=2.8309 acc_corrupt=0.5887 corrupt_frac=1.0000 loss=2.8309 loss_recon=2.8309 loss_meanflow=0.0000 mean_model_t=0.4957 mean_corrupt_t=0.4957 wrong_frac=0.5045 init_acc_corrupt=0.4599 init_gold_top10=0.4898 init_gold_top100=0.4956
+step=155500 micro_steps=311000 elapsed=63.7s lr=3.000000e-04 loss_all=2.7582 acc_all=0.5979 loss_corrupt=2.7582 acc_corrupt=0.5979 corrupt_frac=1.0000 loss=2.7582 loss_recon=2.7582 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4968 init_acc_corrupt=0.4694 init_gold_top10=0.4976 init_gold_top100=0.5033
+step=155600 micro_steps=311200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7887 acc_all=0.5947 loss_corrupt=2.7887 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7887 loss_recon=2.7887 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4993 init_acc_corrupt=0.4661 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=155700 micro_steps=311400 elapsed=60.1s lr=3.000000e-04 loss_all=2.8142 acc_all=0.5909 loss_corrupt=2.8142 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8142 loss_recon=2.8142 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5030 init_acc_corrupt=0.4624 init_gold_top10=0.4913 init_gold_top100=0.4972
+step=155800 micro_steps=311600 elapsed=63.6s lr=3.000000e-04 loss_all=2.7989 acc_all=0.5932 loss_corrupt=2.7989 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7989 loss_recon=2.7989 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5003 init_acc_corrupt=0.4656 init_gold_top10=0.4939 init_gold_top100=0.4998
+step=155900 micro_steps=311800 elapsed=63.8s lr=3.000000e-04 loss_all=2.8074 acc_all=0.5916 loss_corrupt=2.8074 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8074 loss_recon=2.8074 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5025 init_acc_corrupt=0.4627 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=156000 micro_steps=312000 elapsed=63.2s lr=3.000000e-04 loss_all=2.7902 acc_all=0.5944 loss_corrupt=2.7902 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7902 loss_recon=2.7902 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4995 init_acc_corrupt=0.4664 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=156100 micro_steps=312200 elapsed=65.1s lr=3.000000e-04 loss_all=2.7930 acc_all=0.5935 loss_corrupt=2.7930 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7930 loss_recon=2.7930 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5010 init_acc_corrupt=0.4637 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=156200 micro_steps=312400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7990 acc_all=0.5934 loss_corrupt=2.7990 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7990 loss_recon=2.7990 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4998 init_acc_corrupt=0.4654 init_gold_top10=0.4945 init_gold_top100=0.5003
+step=156300 micro_steps=312600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8005 acc_all=0.5934 loss_corrupt=2.8005 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.8005 loss_recon=2.8005 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4997 init_acc_corrupt=0.4658 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=156400 micro_steps=312800 elapsed=63.8s lr=3.000000e-04 loss_all=2.7845 acc_all=0.5946 loss_corrupt=2.7845 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7845 loss_recon=2.7845 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.4994 init_acc_corrupt=0.4657 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=156500 micro_steps=313000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7678 acc_all=0.5970 loss_corrupt=2.7678 acc_corrupt=0.5970 corrupt_frac=1.0000 loss=2.7678 loss_recon=2.7678 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4973 init_acc_corrupt=0.4685 init_gold_top10=0.4973 init_gold_top100=0.5029
+step=156600 micro_steps=313200 elapsed=63.7s lr=3.000000e-04 loss_all=2.7822 acc_all=0.5954 loss_corrupt=2.7822 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7822 loss_recon=2.7822 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4983 init_acc_corrupt=0.4670 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=156700 micro_steps=313400 elapsed=64.2s lr=3.000000e-04 loss_all=2.8112 acc_all=0.5913 loss_corrupt=2.8112 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8112 loss_recon=2.8112 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5024 init_acc_corrupt=0.4625 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=156800 micro_steps=313600 elapsed=64.3s lr=3.000000e-04 loss_all=2.7658 acc_all=0.5977 loss_corrupt=2.7658 acc_corrupt=0.5977 corrupt_frac=1.0000 loss=2.7658 loss_recon=2.7658 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4963 init_acc_corrupt=0.4696 init_gold_top10=0.4984 init_gold_top100=0.5039
+step=156900 micro_steps=313800 elapsed=64.4s lr=3.000000e-04 loss_all=2.7722 acc_all=0.5969 loss_corrupt=2.7722 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7722 loss_recon=2.7722 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4972 init_acc_corrupt=0.4681 init_gold_top10=0.4971 init_gold_top100=0.5030
+step=157000 micro_steps=314000 elapsed=64.6s lr=3.000000e-04 loss_all=2.7743 acc_all=0.5959 loss_corrupt=2.7743 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7743 loss_recon=2.7743 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4988 init_acc_corrupt=0.4668 init_gold_top10=0.4959 init_gold_top100=0.5014
+step=157100 micro_steps=314200 elapsed=66.3s lr=3.000000e-04 loss_all=2.7537 acc_all=0.5989 loss_corrupt=2.7537 acc_corrupt=0.5989 corrupt_frac=1.0000 loss=2.7537 loss_recon=2.7537 loss_meanflow=0.0000 mean_model_t=0.5036 mean_corrupt_t=0.5036 wrong_frac=0.4966 init_acc_corrupt=0.4693 init_gold_top10=0.4981 init_gold_top100=0.5036
+step=157200 micro_steps=314400 elapsed=64.5s lr=3.000000e-04 loss_all=2.7879 acc_all=0.5943 loss_corrupt=2.7879 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7879 loss_recon=2.7879 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4993 init_acc_corrupt=0.4658 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=157300 micro_steps=314600 elapsed=64.7s lr=3.000000e-04 loss_all=2.7822 acc_all=0.5955 loss_corrupt=2.7822 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7822 loss_recon=2.7822 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4986 init_acc_corrupt=0.4670 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=157400 micro_steps=314800 elapsed=64.7s lr=3.000000e-04 loss_all=2.8109 acc_all=0.5913 loss_corrupt=2.8109 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8109 loss_recon=2.8109 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5028 init_acc_corrupt=0.4626 init_gold_top10=0.4916 init_gold_top100=0.4972
+step=157500 micro_steps=315000 elapsed=64.7s lr=3.000000e-04 loss_all=2.8074 acc_all=0.5919 loss_corrupt=2.8074 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8074 loss_recon=2.8074 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5012 init_acc_corrupt=0.4642 init_gold_top10=0.4929 init_gold_top100=0.4988
+step=157600 micro_steps=315200 elapsed=64.6s lr=3.000000e-04 loss_all=2.7926 acc_all=0.5937 loss_corrupt=2.7926 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7926 loss_recon=2.7926 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4999 init_acc_corrupt=0.4657 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=157700 micro_steps=315400 elapsed=71.0s lr=3.000000e-04 loss_all=2.7822 acc_all=0.5950 loss_corrupt=2.7822 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7822 loss_recon=2.7822 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4991 init_acc_corrupt=0.4667 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=157800 micro_steps=315600 elapsed=72.3s lr=3.000000e-04 loss_all=2.7774 acc_all=0.5956 loss_corrupt=2.7774 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7774 loss_recon=2.7774 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.5000 init_acc_corrupt=0.4664 init_gold_top10=0.4946 init_gold_top100=0.5002
+step=157900 micro_steps=315800 elapsed=75.6s lr=3.000000e-04 loss_all=2.7941 acc_all=0.5934 loss_corrupt=2.7941 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7941 loss_recon=2.7941 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5004 init_acc_corrupt=0.4649 init_gold_top10=0.4939 init_gold_top100=0.4998
+step=158000 micro_steps=316000 elapsed=74.7s lr=3.000000e-04 loss_all=2.7481 acc_all=0.5996 loss_corrupt=2.7481 acc_corrupt=0.5996 corrupt_frac=1.0000 loss=2.7481 loss_recon=2.7481 loss_meanflow=0.0000 mean_model_t=0.5060 mean_corrupt_t=0.5060 wrong_frac=0.4941 init_acc_corrupt=0.4718 init_gold_top10=0.5005 init_gold_top100=0.5060
+step=158100 micro_steps=316200 elapsed=74.2s lr=3.000000e-04 loss_all=2.7904 acc_all=0.5946 loss_corrupt=2.7904 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7904 loss_recon=2.7904 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4989 init_acc_corrupt=0.4667 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=158200 micro_steps=316400 elapsed=69.3s lr=3.000000e-04 loss_all=2.8108 acc_all=0.5915 loss_corrupt=2.8108 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8108 loss_recon=2.8108 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5022 init_acc_corrupt=0.4631 init_gold_top10=0.4921 init_gold_top100=0.4980
+step=158300 micro_steps=316600 elapsed=72.3s lr=3.000000e-04 loss_all=2.8305 acc_all=0.5891 loss_corrupt=2.8305 acc_corrupt=0.5891 corrupt_frac=1.0000 loss=2.8305 loss_recon=2.8305 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5048 init_acc_corrupt=0.4601 init_gold_top10=0.4893 init_gold_top100=0.4953
+step=158400 micro_steps=316800 elapsed=69.7s lr=3.000000e-04 loss_all=2.7902 acc_all=0.5938 loss_corrupt=2.7902 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7902 loss_recon=2.7902 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5012 init_acc_corrupt=0.4648 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=158500 micro_steps=317000 elapsed=71.4s lr=3.000000e-04 loss_all=2.7846 acc_all=0.5954 loss_corrupt=2.7846 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7846 loss_recon=2.7846 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4984 init_acc_corrupt=0.4675 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=158600 micro_steps=317200 elapsed=71.1s lr=3.000000e-04 loss_all=2.7682 acc_all=0.5968 loss_corrupt=2.7682 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7682 loss_recon=2.7682 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4978 init_acc_corrupt=0.4682 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=158700 micro_steps=317400 elapsed=70.4s lr=3.000000e-04 loss_all=2.7814 acc_all=0.5950 loss_corrupt=2.7814 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7814 loss_recon=2.7814 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4994 init_acc_corrupt=0.4658 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=158800 micro_steps=317600 elapsed=71.4s lr=3.000000e-04 loss_all=2.7980 acc_all=0.5928 loss_corrupt=2.7980 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.7980 loss_recon=2.7980 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5012 init_acc_corrupt=0.4638 init_gold_top10=0.4932 init_gold_top100=0.4990
+step=158900 micro_steps=317800 elapsed=63.0s lr=3.000000e-04 loss_all=2.7803 acc_all=0.5953 loss_corrupt=2.7803 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7803 loss_recon=2.7803 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4998 init_acc_corrupt=0.4662 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=159000 micro_steps=318000 elapsed=64.4s lr=3.000000e-04 loss_all=2.7817 acc_all=0.5951 loss_corrupt=2.7817 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7817 loss_recon=2.7817 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5003 init_acc_corrupt=0.4646 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=159100 micro_steps=318200 elapsed=63.0s lr=3.000000e-04 loss_all=2.7833 acc_all=0.5948 loss_corrupt=2.7833 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7833 loss_recon=2.7833 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5000 init_acc_corrupt=0.4658 init_gold_top10=0.4944 init_gold_top100=0.5002
+step=159200 micro_steps=318400 elapsed=64.6s lr=3.000000e-04 loss_all=2.7884 acc_all=0.5941 loss_corrupt=2.7884 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7884 loss_recon=2.7884 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4993 init_acc_corrupt=0.4662 init_gold_top10=0.4949 init_gold_top100=0.5008
+step=159300 micro_steps=318600 elapsed=64.4s lr=3.000000e-04 loss_all=2.7855 acc_all=0.5943 loss_corrupt=2.7855 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7855 loss_recon=2.7855 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5002 init_acc_corrupt=0.4652 init_gold_top10=0.4944 init_gold_top100=0.5000
+step=159400 micro_steps=318800 elapsed=64.0s lr=3.000000e-04 loss_all=2.7811 acc_all=0.5957 loss_corrupt=2.7811 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7811 loss_recon=2.7811 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4985 init_acc_corrupt=0.4673 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=159500 micro_steps=319000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8015 acc_all=0.5929 loss_corrupt=2.8015 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.8015 loss_recon=2.8015 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5008 init_acc_corrupt=0.4648 init_gold_top10=0.4935 init_gold_top100=0.4994
+step=159600 micro_steps=319200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7881 acc_all=0.5942 loss_corrupt=2.7881 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7881 loss_recon=2.7881 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5003 init_acc_corrupt=0.4654 init_gold_top10=0.4943 init_gold_top100=0.4999
+step=159700 micro_steps=319400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7942 acc_all=0.5937 loss_corrupt=2.7942 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7942 loss_recon=2.7942 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5000 init_acc_corrupt=0.4655 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=159800 micro_steps=319600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7506 acc_all=0.5994 loss_corrupt=2.7506 acc_corrupt=0.5994 corrupt_frac=1.0000 loss=2.7506 loss_recon=2.7506 loss_meanflow=0.0000 mean_model_t=0.5052 mean_corrupt_t=0.5052 wrong_frac=0.4944 init_acc_corrupt=0.4719 init_gold_top10=0.5001 init_gold_top100=0.5058
+step=159900 micro_steps=319800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8194 acc_all=0.5900 loss_corrupt=2.8194 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8194 loss_recon=2.8194 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5038 init_acc_corrupt=0.4616 init_gold_top10=0.4905 init_gold_top100=0.4964
+step=160000 micro_steps=320000 elapsed=63.3s lr=3.000000e-04 loss_all=2.7546 acc_all=0.5993 loss_corrupt=2.7546 acc_corrupt=0.5993 corrupt_frac=1.0000 loss=2.7546 loss_recon=2.7546 loss_meanflow=0.0000 mean_model_t=0.5051 mean_corrupt_t=0.5051 wrong_frac=0.4949 init_acc_corrupt=0.4715 init_gold_top10=0.4997 init_gold_top100=0.5053
+step=160100 micro_steps=320200 elapsed=66.7s lr=3.000000e-04 loss_all=2.8027 acc_all=0.5927 loss_corrupt=2.8027 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8027 loss_recon=2.8027 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5007 init_acc_corrupt=0.4644 init_gold_top10=0.4936 init_gold_top100=0.4994
+step=160200 micro_steps=320400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8000 acc_all=0.5927 loss_corrupt=2.8000 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8000 loss_recon=2.8000 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5012 init_acc_corrupt=0.4641 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=160300 micro_steps=320600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7797 acc_all=0.5956 loss_corrupt=2.7797 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7797 loss_recon=2.7797 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4977 init_acc_corrupt=0.4679 init_gold_top10=0.4968 init_gold_top100=0.5025
+step=160400 micro_steps=320800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7538 acc_all=0.5994 loss_corrupt=2.7538 acc_corrupt=0.5994 corrupt_frac=1.0000 loss=2.7538 loss_recon=2.7538 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4946 init_acc_corrupt=0.4711 init_gold_top10=0.4997 init_gold_top100=0.5055
+step=160500 micro_steps=321000 elapsed=63.3s lr=3.000000e-04 loss_all=2.7948 acc_all=0.5936 loss_corrupt=2.7948 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7948 loss_recon=2.7948 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5007 init_acc_corrupt=0.4647 init_gold_top10=0.4938 init_gold_top100=0.4994
+step=160600 micro_steps=321200 elapsed=63.4s lr=3.000000e-04 loss_all=2.7874 acc_all=0.5943 loss_corrupt=2.7874 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7874 loss_recon=2.7874 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4997 init_acc_corrupt=0.4658 init_gold_top10=0.4946 init_gold_top100=0.5005
+step=160700 micro_steps=321400 elapsed=63.6s lr=3.000000e-04 loss_all=2.7978 acc_all=0.5932 loss_corrupt=2.7978 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7978 loss_recon=2.7978 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5011 init_acc_corrupt=0.4642 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=160800 micro_steps=321600 elapsed=63.6s lr=3.000000e-04 loss_all=2.7827 acc_all=0.5949 loss_corrupt=2.7827 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7827 loss_recon=2.7827 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4989 init_acc_corrupt=0.4666 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=160900 micro_steps=321800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7645 acc_all=0.5974 loss_corrupt=2.7645 acc_corrupt=0.5974 corrupt_frac=1.0000 loss=2.7645 loss_recon=2.7645 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4968 init_acc_corrupt=0.4692 init_gold_top10=0.4977 init_gold_top100=0.5034
+step=161000 micro_steps=322000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7965 acc_all=0.5928 loss_corrupt=2.7965 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.7965 loss_recon=2.7965 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5020 init_acc_corrupt=0.4636 init_gold_top10=0.4924 init_gold_top100=0.4981
+step=161100 micro_steps=322200 elapsed=65.2s lr=3.000000e-04 loss_all=2.8008 acc_all=0.5930 loss_corrupt=2.8008 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.8008 loss_recon=2.8008 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5010 init_acc_corrupt=0.4642 init_gold_top10=0.4931 init_gold_top100=0.4990
+step=161200 micro_steps=322400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7909 acc_all=0.5944 loss_corrupt=2.7909 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7909 loss_recon=2.7909 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4993 init_acc_corrupt=0.4659 init_gold_top10=0.4949 init_gold_top100=0.5008
+step=161300 micro_steps=322600 elapsed=63.2s lr=3.000000e-04 loss_all=2.8017 acc_all=0.5930 loss_corrupt=2.8017 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.8017 loss_recon=2.8017 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5001 init_acc_corrupt=0.4649 init_gold_top10=0.4941 init_gold_top100=0.5000
+step=161400 micro_steps=322800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7837 acc_all=0.5949 loss_corrupt=2.7837 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7837 loss_recon=2.7837 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4989 init_acc_corrupt=0.4663 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=161500 micro_steps=323000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7771 acc_all=0.5959 loss_corrupt=2.7771 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7771 loss_recon=2.7771 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4982 init_acc_corrupt=0.4673 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=161600 micro_steps=323200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7832 acc_all=0.5954 loss_corrupt=2.7832 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7832 loss_recon=2.7832 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4987 init_acc_corrupt=0.4664 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=161700 micro_steps=323400 elapsed=63.6s lr=3.000000e-04 loss_all=2.8003 acc_all=0.5932 loss_corrupt=2.8003 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.8003 loss_recon=2.8003 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5011 init_acc_corrupt=0.4646 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=161800 micro_steps=323600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7794 acc_all=0.5957 loss_corrupt=2.7794 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7794 loss_recon=2.7794 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4984 init_acc_corrupt=0.4674 init_gold_top10=0.4959 init_gold_top100=0.5017
+step=161900 micro_steps=323800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7475 acc_all=0.6001 loss_corrupt=2.7475 acc_corrupt=0.6001 corrupt_frac=1.0000 loss=2.7475 loss_recon=2.7475 loss_meanflow=0.0000 mean_model_t=0.5058 mean_corrupt_t=0.5058 wrong_frac=0.4943 init_acc_corrupt=0.4726 init_gold_top10=0.5003 init_gold_top100=0.5059
+step=162000 micro_steps=324000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7756 acc_all=0.5961 loss_corrupt=2.7756 acc_corrupt=0.5961 corrupt_frac=1.0000 loss=2.7756 loss_recon=2.7756 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4975 init_acc_corrupt=0.4683 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=162100 micro_steps=324200 elapsed=64.8s lr=3.000000e-04 loss_all=2.7580 acc_all=0.5986 loss_corrupt=2.7580 acc_corrupt=0.5986 corrupt_frac=1.0000 loss=2.7580 loss_recon=2.7580 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4968 init_acc_corrupt=0.4690 init_gold_top10=0.4976 init_gold_top100=0.5033
+step=162200 micro_steps=324400 elapsed=60.0s lr=3.000000e-04 loss_all=2.7881 acc_all=0.5945 loss_corrupt=2.7881 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7881 loss_recon=2.7881 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4994 init_acc_corrupt=0.4662 init_gold_top10=0.4949 init_gold_top100=0.5008
+step=162300 micro_steps=324600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8006 acc_all=0.5923 loss_corrupt=2.8006 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8006 loss_recon=2.8006 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5027 init_acc_corrupt=0.4634 init_gold_top10=0.4918 init_gold_top100=0.4973
+step=162400 micro_steps=324800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7997 acc_all=0.5929 loss_corrupt=2.7997 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.7997 loss_recon=2.7997 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5014 init_acc_corrupt=0.4637 init_gold_top10=0.4929 init_gold_top100=0.4988
+step=162500 micro_steps=325000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7967 acc_all=0.5931 loss_corrupt=2.7967 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.7967 loss_recon=2.7967 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5014 init_acc_corrupt=0.4647 init_gold_top10=0.4931 init_gold_top100=0.4987
+step=162600 micro_steps=325200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7742 acc_all=0.5963 loss_corrupt=2.7742 acc_corrupt=0.5963 corrupt_frac=1.0000 loss=2.7742 loss_recon=2.7742 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4979 init_acc_corrupt=0.4672 init_gold_top10=0.4966 init_gold_top100=0.5022
+step=162700 micro_steps=325400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8109 acc_all=0.5912 loss_corrupt=2.8109 acc_corrupt=0.5912 corrupt_frac=1.0000 loss=2.8109 loss_recon=2.8109 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5020 init_acc_corrupt=0.4634 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=162800 micro_steps=325600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7730 acc_all=0.5962 loss_corrupt=2.7730 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7730 loss_recon=2.7730 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4988 init_acc_corrupt=0.4670 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=162900 micro_steps=325800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7934 acc_all=0.5943 loss_corrupt=2.7934 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7934 loss_recon=2.7934 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4993 init_acc_corrupt=0.4662 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=163000 micro_steps=326000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8123 acc_all=0.5915 loss_corrupt=2.8123 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8123 loss_recon=2.8123 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5015 init_acc_corrupt=0.4631 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=163100 micro_steps=326200 elapsed=65.2s lr=3.000000e-04 loss_all=2.8006 acc_all=0.5924 loss_corrupt=2.8006 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8006 loss_recon=2.8006 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5024 init_acc_corrupt=0.4630 init_gold_top10=0.4919 init_gold_top100=0.4977
+step=163200 micro_steps=326400 elapsed=63.6s lr=3.000000e-04 loss_all=2.7989 acc_all=0.5931 loss_corrupt=2.7989 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.7989 loss_recon=2.7989 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5004 init_acc_corrupt=0.4649 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=163300 micro_steps=326600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7975 acc_all=0.5928 loss_corrupt=2.7975 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.7975 loss_recon=2.7975 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5010 init_acc_corrupt=0.4638 init_gold_top10=0.4933 init_gold_top100=0.4992
+step=163400 micro_steps=326800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7553 acc_all=0.5987 loss_corrupt=2.7553 acc_corrupt=0.5987 corrupt_frac=1.0000 loss=2.7553 loss_recon=2.7553 loss_meanflow=0.0000 mean_model_t=0.5049 mean_corrupt_t=0.5049 wrong_frac=0.4958 init_acc_corrupt=0.4706 init_gold_top10=0.4987 init_gold_top100=0.5044
+step=163500 micro_steps=327000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7858 acc_all=0.5951 loss_corrupt=2.7858 acc_corrupt=0.5951 corrupt_frac=1.0000 loss=2.7858 loss_recon=2.7858 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4983 init_acc_corrupt=0.4675 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=163600 micro_steps=327200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7677 acc_all=0.5968 loss_corrupt=2.7677 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7677 loss_recon=2.7677 loss_meanflow=0.0000 mean_model_t=0.5033 mean_corrupt_t=0.5033 wrong_frac=0.4973 init_acc_corrupt=0.4691 init_gold_top10=0.4971 init_gold_top100=0.5028
+step=163700 micro_steps=327400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8208 acc_all=0.5901 loss_corrupt=2.8208 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8208 loss_recon=2.8208 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5029 init_acc_corrupt=0.4622 init_gold_top10=0.4913 init_gold_top100=0.4972
+step=163800 micro_steps=327600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7916 acc_all=0.5936 loss_corrupt=2.7916 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7916 loss_recon=2.7916 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5008 init_acc_corrupt=0.4652 init_gold_top10=0.4937 init_gold_top100=0.4993
+step=163900 micro_steps=327800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7882 acc_all=0.5944 loss_corrupt=2.7882 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7882 loss_recon=2.7882 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4996 init_acc_corrupt=0.4660 init_gold_top10=0.4949 init_gold_top100=0.5005
+step=164000 micro_steps=328000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7902 acc_all=0.5940 loss_corrupt=2.7902 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7902 loss_recon=2.7902 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4999 init_acc_corrupt=0.4654 init_gold_top10=0.4945 init_gold_top100=0.5001
+step=164100 micro_steps=328200 elapsed=64.7s lr=3.000000e-04 loss_all=2.7906 acc_all=0.5939 loss_corrupt=2.7906 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7906 loss_recon=2.7906 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5006 init_acc_corrupt=0.4649 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=164200 micro_steps=328400 elapsed=63.0s lr=3.000000e-04 loss_all=2.7939 acc_all=0.5939 loss_corrupt=2.7939 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7939 loss_recon=2.7939 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5013 init_acc_corrupt=0.4646 init_gold_top10=0.4931 init_gold_top100=0.4988
+step=164300 micro_steps=328600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7678 acc_all=0.5970 loss_corrupt=2.7678 acc_corrupt=0.5970 corrupt_frac=1.0000 loss=2.7678 loss_recon=2.7678 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4982 init_acc_corrupt=0.4674 init_gold_top10=0.4962 init_gold_top100=0.5018
+step=164400 micro_steps=328800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7950 acc_all=0.5939 loss_corrupt=2.7950 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7950 loss_recon=2.7950 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5003 init_acc_corrupt=0.4652 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=164500 micro_steps=329000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7921 acc_all=0.5943 loss_corrupt=2.7921 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7921 loss_recon=2.7921 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4997 init_acc_corrupt=0.4660 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=164600 micro_steps=329200 elapsed=63.4s lr=3.000000e-04 loss_all=2.7962 acc_all=0.5934 loss_corrupt=2.7962 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7962 loss_recon=2.7962 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5014 init_acc_corrupt=0.4641 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=164700 micro_steps=329400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8141 acc_all=0.5910 loss_corrupt=2.8141 acc_corrupt=0.5910 corrupt_frac=1.0000 loss=2.8141 loss_recon=2.8141 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5031 init_acc_corrupt=0.4620 init_gold_top10=0.4914 init_gold_top100=0.4970
+step=164800 micro_steps=329600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7957 acc_all=0.5931 loss_corrupt=2.7957 acc_corrupt=0.5931 corrupt_frac=1.0000 loss=2.7957 loss_recon=2.7957 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5015 init_acc_corrupt=0.4642 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=164900 micro_steps=329800 elapsed=63.3s lr=3.000000e-04 loss_all=2.8173 acc_all=0.5909 loss_corrupt=2.8173 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8173 loss_recon=2.8173 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5029 init_acc_corrupt=0.4622 init_gold_top10=0.4914 init_gold_top100=0.4973
+step=165000 micro_steps=330000 elapsed=63.3s lr=3.000000e-04 loss_all=2.7781 acc_all=0.5961 loss_corrupt=2.7781 acc_corrupt=0.5961 corrupt_frac=1.0000 loss=2.7781 loss_recon=2.7781 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4979 init_acc_corrupt=0.4676 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=165100 micro_steps=330200 elapsed=64.8s lr=3.000000e-04 loss_all=2.7893 acc_all=0.5942 loss_corrupt=2.7893 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7893 loss_recon=2.7893 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5006 init_acc_corrupt=0.4649 init_gold_top10=0.4938 init_gold_top100=0.4995
+step=165200 micro_steps=330400 elapsed=59.7s lr=3.000000e-04 loss_all=2.8242 acc_all=0.5899 loss_corrupt=2.8242 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8242 loss_recon=2.8242 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5040 init_acc_corrupt=0.4606 init_gold_top10=0.4902 init_gold_top100=0.4962
+step=165300 micro_steps=330600 elapsed=63.3s lr=3.000000e-04 loss_all=2.7993 acc_all=0.5932 loss_corrupt=2.7993 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7993 loss_recon=2.7993 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5006 init_acc_corrupt=0.4650 init_gold_top10=0.4938 init_gold_top100=0.4995
+step=165400 micro_steps=330800 elapsed=63.3s lr=3.000000e-04 loss_all=2.7641 acc_all=0.5975 loss_corrupt=2.7641 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7641 loss_recon=2.7641 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4974 init_acc_corrupt=0.4686 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=165500 micro_steps=331000 elapsed=63.3s lr=3.000000e-04 loss_all=2.7710 acc_all=0.5969 loss_corrupt=2.7710 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7710 loss_recon=2.7710 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4982 init_acc_corrupt=0.4678 init_gold_top10=0.4963 init_gold_top100=0.5019
+step=165600 micro_steps=331200 elapsed=63.3s lr=3.000000e-04 loss_all=2.7729 acc_all=0.5964 loss_corrupt=2.7729 acc_corrupt=0.5964 corrupt_frac=1.0000 loss=2.7729 loss_recon=2.7729 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4975 init_acc_corrupt=0.4680 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=165700 micro_steps=331400 elapsed=63.3s lr=3.000000e-04 loss_all=2.7699 acc_all=0.5962 loss_corrupt=2.7699 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7699 loss_recon=2.7699 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4981 init_acc_corrupt=0.4672 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=165800 micro_steps=331600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8263 acc_all=0.5899 loss_corrupt=2.8263 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8263 loss_recon=2.8263 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5034 init_acc_corrupt=0.4613 init_gold_top10=0.4907 init_gold_top100=0.4967
+step=165900 micro_steps=331800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7789 acc_all=0.5954 loss_corrupt=2.7789 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7789 loss_recon=2.7789 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4997 init_acc_corrupt=0.4653 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=166000 micro_steps=332000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7931 acc_all=0.5937 loss_corrupt=2.7931 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7931 loss_recon=2.7931 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5006 init_acc_corrupt=0.4646 init_gold_top10=0.4938 init_gold_top100=0.4996
+step=166100 micro_steps=332200 elapsed=65.1s lr=3.000000e-04 loss_all=2.7782 acc_all=0.5959 loss_corrupt=2.7782 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7782 loss_recon=2.7782 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4985 init_acc_corrupt=0.4671 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=166200 micro_steps=332400 elapsed=63.6s lr=3.000000e-04 loss_all=2.7762 acc_all=0.5959 loss_corrupt=2.7762 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7762 loss_recon=2.7762 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4990 init_acc_corrupt=0.4665 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=166300 micro_steps=332600 elapsed=63.6s lr=3.000000e-04 loss_all=2.7709 acc_all=0.5967 loss_corrupt=2.7709 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7709 loss_recon=2.7709 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4982 init_acc_corrupt=0.4674 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=166400 micro_steps=332800 elapsed=63.7s lr=3.000000e-04 loss_all=2.7799 acc_all=0.5957 loss_corrupt=2.7799 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7799 loss_recon=2.7799 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4978 init_acc_corrupt=0.4677 init_gold_top10=0.4965 init_gold_top100=0.5023
+step=166500 micro_steps=333000 elapsed=64.5s lr=3.000000e-04 loss_all=2.8174 acc_all=0.5909 loss_corrupt=2.8174 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8174 loss_recon=2.8174 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5020 init_acc_corrupt=0.4631 init_gold_top10=0.4922 init_gold_top100=0.4980
+step=166600 micro_steps=333200 elapsed=64.4s lr=3.000000e-04 loss_all=2.7830 acc_all=0.5950 loss_corrupt=2.7830 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7830 loss_recon=2.7830 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4995 init_acc_corrupt=0.4663 init_gold_top10=0.4949 init_gold_top100=0.5006
+step=166700 micro_steps=333400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8108 acc_all=0.5917 loss_corrupt=2.8108 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8108 loss_recon=2.8108 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5023 init_acc_corrupt=0.4622 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=166800 micro_steps=333600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7979 acc_all=0.5932 loss_corrupt=2.7979 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7979 loss_recon=2.7979 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5004 init_acc_corrupt=0.4645 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=166900 micro_steps=333800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7921 acc_all=0.5940 loss_corrupt=2.7921 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7921 loss_recon=2.7921 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.5001 init_acc_corrupt=0.4653 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=167000 micro_steps=334000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7814 acc_all=0.5955 loss_corrupt=2.7814 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7814 loss_recon=2.7814 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4989 init_acc_corrupt=0.4672 init_gold_top10=0.4955 init_gold_top100=0.5012
+step=167100 micro_steps=334200 elapsed=65.0s lr=3.000000e-04 loss_all=2.8001 acc_all=0.5930 loss_corrupt=2.8001 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.8001 loss_recon=2.8001 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4643 init_gold_top10=0.4936 init_gold_top100=0.4994
+step=167200 micro_steps=334400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7601 acc_all=0.5978 loss_corrupt=2.7601 acc_corrupt=0.5978 corrupt_frac=1.0000 loss=2.7601 loss_recon=2.7601 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4976 init_acc_corrupt=0.4684 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=167300 micro_steps=334600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7632 acc_all=0.5976 loss_corrupt=2.7632 acc_corrupt=0.5976 corrupt_frac=1.0000 loss=2.7632 loss_recon=2.7632 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4980 init_acc_corrupt=0.4682 init_gold_top10=0.4967 init_gold_top100=0.5022
+step=167400 micro_steps=334800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7811 acc_all=0.5954 loss_corrupt=2.7811 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7811 loss_recon=2.7811 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4995 init_acc_corrupt=0.4667 init_gold_top10=0.4949 init_gold_top100=0.5006
+step=167500 micro_steps=335000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7740 acc_all=0.5963 loss_corrupt=2.7740 acc_corrupt=0.5963 corrupt_frac=1.0000 loss=2.7740 loss_recon=2.7740 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4979 init_acc_corrupt=0.4684 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=167600 micro_steps=335200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7741 acc_all=0.5965 loss_corrupt=2.7741 acc_corrupt=0.5965 corrupt_frac=1.0000 loss=2.7741 loss_recon=2.7741 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4977 init_acc_corrupt=0.4680 init_gold_top10=0.4967 init_gold_top100=0.5025
+step=167700 micro_steps=335400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7815 acc_all=0.5958 loss_corrupt=2.7815 acc_corrupt=0.5958 corrupt_frac=1.0000 loss=2.7815 loss_recon=2.7815 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4983 init_acc_corrupt=0.4669 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=167800 micro_steps=335600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8038 acc_all=0.5923 loss_corrupt=2.8038 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8038 loss_recon=2.8038 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5015 init_acc_corrupt=0.4635 init_gold_top10=0.4928 init_gold_top100=0.4986
+step=167900 micro_steps=335800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8243 acc_all=0.5901 loss_corrupt=2.8243 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8243 loss_recon=2.8243 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5035 init_acc_corrupt=0.4612 init_gold_top10=0.4908 init_gold_top100=0.4966
+step=168000 micro_steps=336000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7683 acc_all=0.5970 loss_corrupt=2.7683 acc_corrupt=0.5970 corrupt_frac=1.0000 loss=2.7683 loss_recon=2.7683 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4975 init_acc_corrupt=0.4686 init_gold_top10=0.4970 init_gold_top100=0.5026
+step=168100 micro_steps=336200 elapsed=65.0s lr=3.000000e-04 loss_all=2.8204 acc_all=0.5900 loss_corrupt=2.8204 acc_corrupt=0.5900 corrupt_frac=1.0000 loss=2.8204 loss_recon=2.8204 loss_meanflow=0.0000 mean_model_t=0.4951 mean_corrupt_t=0.4951 wrong_frac=0.5050 init_acc_corrupt=0.4603 init_gold_top10=0.4893 init_gold_top100=0.4951
+step=168200 micro_steps=336400 elapsed=60.0s lr=3.000000e-04 loss_all=2.7941 acc_all=0.5938 loss_corrupt=2.7941 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7941 loss_recon=2.7941 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4997 init_acc_corrupt=0.4654 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=168300 micro_steps=336600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8133 acc_all=0.5909 loss_corrupt=2.8133 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8133 loss_recon=2.8133 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5043 init_acc_corrupt=0.4613 init_gold_top10=0.4900 init_gold_top100=0.4957
+step=168400 micro_steps=336800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7945 acc_all=0.5933 loss_corrupt=2.7945 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7945 loss_recon=2.7945 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5012 init_acc_corrupt=0.4643 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=168500 micro_steps=337000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7777 acc_all=0.5966 loss_corrupt=2.7777 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7777 loss_recon=2.7777 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4980 init_acc_corrupt=0.4679 init_gold_top10=0.4964 init_gold_top100=0.5022
+step=168600 micro_steps=337200 elapsed=63.3s lr=3.000000e-04 loss_all=2.7688 acc_all=0.5965 loss_corrupt=2.7688 acc_corrupt=0.5965 corrupt_frac=1.0000 loss=2.7688 loss_recon=2.7688 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4980 init_acc_corrupt=0.4680 init_gold_top10=0.4964 init_gold_top100=0.5022
+step=168700 micro_steps=337400 elapsed=62.7s lr=3.000000e-04 loss_all=2.8424 acc_all=0.5877 loss_corrupt=2.8424 acc_corrupt=0.5877 corrupt_frac=1.0000 loss=2.8424 loss_recon=2.8424 loss_meanflow=0.0000 mean_model_t=0.4939 mean_corrupt_t=0.4939 wrong_frac=0.5058 init_acc_corrupt=0.4585 init_gold_top10=0.4883 init_gold_top100=0.4943
+step=168800 micro_steps=337600 elapsed=62.6s lr=3.000000e-04 loss_all=2.7957 acc_all=0.5935 loss_corrupt=2.7957 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7957 loss_recon=2.7957 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5014 init_acc_corrupt=0.4640 init_gold_top10=0.4931 init_gold_top100=0.4988
+step=168900 micro_steps=337800 elapsed=62.4s lr=3.000000e-04 loss_all=2.7690 acc_all=0.5969 loss_corrupt=2.7690 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7690 loss_recon=2.7690 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4975 init_acc_corrupt=0.4678 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=169000 micro_steps=338000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7674 acc_all=0.5969 loss_corrupt=2.7674 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7674 loss_recon=2.7674 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4983 init_acc_corrupt=0.4682 init_gold_top10=0.4964 init_gold_top100=0.5019
+step=169100 micro_steps=338200 elapsed=64.9s lr=3.000000e-04 loss_all=2.8073 acc_all=0.5915 loss_corrupt=2.8073 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8073 loss_recon=2.8073 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5030 init_acc_corrupt=0.4621 init_gold_top10=0.4914 init_gold_top100=0.4972
+step=169200 micro_steps=338400 elapsed=63.7s lr=3.000000e-04 loss_all=2.7663 acc_all=0.5971 loss_corrupt=2.7663 acc_corrupt=0.5971 corrupt_frac=1.0000 loss=2.7663 loss_recon=2.7663 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4972 init_acc_corrupt=0.4686 init_gold_top10=0.4972 init_gold_top100=0.5030
+step=169300 micro_steps=338600 elapsed=63.7s lr=3.000000e-04 loss_all=2.8116 acc_all=0.5914 loss_corrupt=2.8116 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8116 loss_recon=2.8116 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5031 init_acc_corrupt=0.4618 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=169400 micro_steps=338800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7773 acc_all=0.5958 loss_corrupt=2.7773 acc_corrupt=0.5958 corrupt_frac=1.0000 loss=2.7773 loss_recon=2.7773 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4992 init_acc_corrupt=0.4662 init_gold_top10=0.4951 init_gold_top100=0.5009
+step=169500 micro_steps=339000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7689 acc_all=0.5970 loss_corrupt=2.7689 acc_corrupt=0.5970 corrupt_frac=1.0000 loss=2.7689 loss_recon=2.7689 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4979 init_acc_corrupt=0.4683 init_gold_top10=0.4967 init_gold_top100=0.5022
+step=169600 micro_steps=339200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7950 acc_all=0.5937 loss_corrupt=2.7950 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7950 loss_recon=2.7950 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.5003 init_acc_corrupt=0.4651 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=169700 micro_steps=339400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8348 acc_all=0.5879 loss_corrupt=2.8348 acc_corrupt=0.5879 corrupt_frac=1.0000 loss=2.8348 loss_recon=2.8348 loss_meanflow=0.0000 mean_model_t=0.4947 mean_corrupt_t=0.4947 wrong_frac=0.5055 init_acc_corrupt=0.4593 init_gold_top10=0.4888 init_gold_top100=0.4946
+step=169800 micro_steps=339600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7546 acc_all=0.5989 loss_corrupt=2.7546 acc_corrupt=0.5989 corrupt_frac=1.0000 loss=2.7546 loss_recon=2.7546 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4956 init_acc_corrupt=0.4699 init_gold_top10=0.4988 init_gold_top100=0.5045
+step=169900 micro_steps=339800 elapsed=63.7s lr=3.000000e-04 loss_all=2.8095 acc_all=0.5920 loss_corrupt=2.8095 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8095 loss_recon=2.8095 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5030 init_acc_corrupt=0.4622 init_gold_top10=0.4914 init_gold_top100=0.4970
+step=170000 micro_steps=340000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7784 acc_all=0.5961 loss_corrupt=2.7784 acc_corrupt=0.5961 corrupt_frac=1.0000 loss=2.7784 loss_recon=2.7784 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4987 init_acc_corrupt=0.4673 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=170100 micro_steps=340200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7921 acc_all=0.5940 loss_corrupt=2.7921 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7921 loss_recon=2.7921 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5013 init_acc_corrupt=0.4640 init_gold_top10=0.4933 init_gold_top100=0.4989
+step=170200 micro_steps=340400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7800 acc_all=0.5956 loss_corrupt=2.7800 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7800 loss_recon=2.7800 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4996 init_acc_corrupt=0.4663 init_gold_top10=0.4949 init_gold_top100=0.5005
+step=170300 micro_steps=340600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7923 acc_all=0.5938 loss_corrupt=2.7923 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7923 loss_recon=2.7923 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5012 init_acc_corrupt=0.4643 init_gold_top10=0.4930 init_gold_top100=0.4990
+step=170400 micro_steps=340800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7921 acc_all=0.5937 loss_corrupt=2.7921 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7921 loss_recon=2.7921 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5011 init_acc_corrupt=0.4644 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=170500 micro_steps=341000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8053 acc_all=0.5916 loss_corrupt=2.8053 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8053 loss_recon=2.8053 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5027 init_acc_corrupt=0.4621 init_gold_top10=0.4915 init_gold_top100=0.4974
+step=170600 micro_steps=341200 elapsed=63.2s lr=3.000000e-04 loss_all=2.8094 acc_all=0.5922 loss_corrupt=2.8094 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8094 loss_recon=2.8094 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5012 init_acc_corrupt=0.4640 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=170700 micro_steps=341400 elapsed=62.4s lr=3.000000e-04 loss_all=2.8194 acc_all=0.5904 loss_corrupt=2.8194 acc_corrupt=0.5904 corrupt_frac=1.0000 loss=2.8194 loss_recon=2.8194 loss_meanflow=0.0000 mean_model_t=0.4971 mean_corrupt_t=0.4971 wrong_frac=0.5033 init_acc_corrupt=0.4619 init_gold_top10=0.4911 init_gold_top100=0.4968
+step=170800 micro_steps=341600 elapsed=62.8s lr=3.000000e-04 loss_all=2.7705 acc_all=0.5960 loss_corrupt=2.7705 acc_corrupt=0.5960 corrupt_frac=1.0000 loss=2.7705 loss_recon=2.7705 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4994 init_acc_corrupt=0.4665 init_gold_top10=0.4951 init_gold_top100=0.5008
+step=170900 micro_steps=341800 elapsed=62.6s lr=3.000000e-04 loss_all=2.8054 acc_all=0.5920 loss_corrupt=2.8054 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8054 loss_recon=2.8054 loss_meanflow=0.0000 mean_model_t=0.4966 mean_corrupt_t=0.4966 wrong_frac=0.5035 init_acc_corrupt=0.4620 init_gold_top10=0.4910 init_gold_top100=0.4966
+step=171000 micro_steps=342000 elapsed=62.8s lr=3.000000e-04 loss_all=2.7565 acc_all=0.5986 loss_corrupt=2.7565 acc_corrupt=0.5986 corrupt_frac=1.0000 loss=2.7565 loss_recon=2.7565 loss_meanflow=0.0000 mean_model_t=0.5035 mean_corrupt_t=0.5035 wrong_frac=0.4965 init_acc_corrupt=0.4693 init_gold_top10=0.4979 init_gold_top100=0.5036
+step=171100 micro_steps=342200 elapsed=64.2s lr=3.000000e-04 loss_all=2.7540 acc_all=0.5984 loss_corrupt=2.7540 acc_corrupt=0.5984 corrupt_frac=1.0000 loss=2.7540 loss_recon=2.7540 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4970 init_acc_corrupt=0.4688 init_gold_top10=0.4975 init_gold_top100=0.5032
+step=171200 micro_steps=342400 elapsed=62.9s lr=3.000000e-04 loss_all=2.7933 acc_all=0.5939 loss_corrupt=2.7933 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7933 loss_recon=2.7933 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5002 init_acc_corrupt=0.4649 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=171300 micro_steps=342600 elapsed=60.1s lr=3.000000e-04 loss_all=2.7630 acc_all=0.5975 loss_corrupt=2.7630 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7630 loss_recon=2.7630 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4968 init_acc_corrupt=0.4690 init_gold_top10=0.4976 init_gold_top100=0.5034
+step=171400 micro_steps=342800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8165 acc_all=0.5909 loss_corrupt=2.8165 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8165 loss_recon=2.8165 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5023 init_acc_corrupt=0.4625 init_gold_top10=0.4918 init_gold_top100=0.4978
+step=171500 micro_steps=343000 elapsed=63.2s lr=3.000000e-04 loss_all=2.7776 acc_all=0.5962 loss_corrupt=2.7776 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7776 loss_recon=2.7776 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4978 init_acc_corrupt=0.4679 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=171600 micro_steps=343200 elapsed=64.4s lr=3.000000e-04 loss_all=2.7866 acc_all=0.5949 loss_corrupt=2.7866 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7866 loss_recon=2.7866 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4992 init_acc_corrupt=0.4664 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=171700 micro_steps=343400 elapsed=66.0s lr=3.000000e-04 loss_all=2.7757 acc_all=0.5962 loss_corrupt=2.7757 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7757 loss_recon=2.7757 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4974 init_acc_corrupt=0.4683 init_gold_top10=0.4969 init_gold_top100=0.5028
+step=171800 micro_steps=343600 elapsed=66.7s lr=3.000000e-04 loss_all=2.7719 acc_all=0.5966 loss_corrupt=2.7719 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7719 loss_recon=2.7719 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4986 init_acc_corrupt=0.4675 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=171900 micro_steps=343800 elapsed=66.7s lr=3.000000e-04 loss_all=2.7936 acc_all=0.5939 loss_corrupt=2.7936 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7936 loss_recon=2.7936 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5001 init_acc_corrupt=0.4647 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=172000 micro_steps=344000 elapsed=66.7s lr=3.000000e-04 loss_all=2.7890 acc_all=0.5944 loss_corrupt=2.7890 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7890 loss_recon=2.7890 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4990 init_acc_corrupt=0.4662 init_gold_top10=0.4953 init_gold_top100=0.5012
+step=172100 micro_steps=344200 elapsed=68.4s lr=3.000000e-04 loss_all=2.7749 acc_all=0.5959 loss_corrupt=2.7749 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7749 loss_recon=2.7749 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4999 init_acc_corrupt=0.4658 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=172200 micro_steps=344400 elapsed=66.6s lr=3.000000e-04 loss_all=2.7666 acc_all=0.5970 loss_corrupt=2.7666 acc_corrupt=0.5970 corrupt_frac=1.0000 loss=2.7666 loss_recon=2.7666 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4986 init_acc_corrupt=0.4674 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=172300 micro_steps=344600 elapsed=66.4s lr=3.000000e-04 loss_all=2.7847 acc_all=0.5948 loss_corrupt=2.7847 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7847 loss_recon=2.7847 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5001 init_acc_corrupt=0.4653 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=172400 micro_steps=344800 elapsed=66.5s lr=3.000000e-04 loss_all=2.8205 acc_all=0.5905 loss_corrupt=2.8205 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8205 loss_recon=2.8205 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5032 init_acc_corrupt=0.4617 init_gold_top10=0.4911 init_gold_top100=0.4969
+step=172500 micro_steps=345000 elapsed=87.2s lr=3.000000e-04 loss_all=2.7729 acc_all=0.5966 loss_corrupt=2.7729 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7729 loss_recon=2.7729 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4985 init_acc_corrupt=0.4677 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=172600 micro_steps=345200 elapsed=102.9s lr=3.000000e-04 loss_all=2.7738 acc_all=0.5959 loss_corrupt=2.7738 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7738 loss_recon=2.7738 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4983 init_acc_corrupt=0.4673 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=172700 micro_steps=345400 elapsed=90.0s lr=3.000000e-04 loss_all=2.7692 acc_all=0.5975 loss_corrupt=2.7692 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7692 loss_recon=2.7692 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4970 init_acc_corrupt=0.4683 init_gold_top10=0.4974 init_gold_top100=0.5032
+step=172800 micro_steps=345600 elapsed=100.8s lr=3.000000e-04 loss_all=2.7846 acc_all=0.5949 loss_corrupt=2.7846 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7846 loss_recon=2.7846 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5002 init_acc_corrupt=0.4647 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=172900 micro_steps=345800 elapsed=103.3s lr=3.000000e-04 loss_all=2.8066 acc_all=0.5918 loss_corrupt=2.8066 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8066 loss_recon=2.8066 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5033 init_acc_corrupt=0.4623 init_gold_top10=0.4912 init_gold_top100=0.4968
+step=173000 micro_steps=346000 elapsed=86.4s lr=3.000000e-04 loss_all=2.7729 acc_all=0.5957 loss_corrupt=2.7729 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7729 loss_recon=2.7729 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5006 init_acc_corrupt=0.4654 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=173100 micro_steps=346200 elapsed=73.6s lr=3.000000e-04 loss_all=2.7652 acc_all=0.5969 loss_corrupt=2.7652 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7652 loss_recon=2.7652 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4989 init_acc_corrupt=0.4669 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=173200 micro_steps=346400 elapsed=66.6s lr=3.000000e-04 loss_all=2.7924 acc_all=0.5932 loss_corrupt=2.7924 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7924 loss_recon=2.7924 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5015 init_acc_corrupt=0.4639 init_gold_top10=0.4929 init_gold_top100=0.4987
+step=173300 micro_steps=346600 elapsed=66.6s lr=3.000000e-04 loss_all=2.7931 acc_all=0.5943 loss_corrupt=2.7931 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7931 loss_recon=2.7931 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5005 init_acc_corrupt=0.4648 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=173400 micro_steps=346800 elapsed=66.5s lr=3.000000e-04 loss_all=2.7671 acc_all=0.5972 loss_corrupt=2.7671 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7671 loss_recon=2.7671 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4980 init_acc_corrupt=0.4680 init_gold_top10=0.4965 init_gold_top100=0.5021
+step=173500 micro_steps=347000 elapsed=66.0s lr=3.000000e-04 loss_all=2.7895 acc_all=0.5945 loss_corrupt=2.7895 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7895 loss_recon=2.7895 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4998 init_acc_corrupt=0.4654 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=173600 micro_steps=347200 elapsed=66.6s lr=3.000000e-04 loss_all=2.7648 acc_all=0.5974 loss_corrupt=2.7648 acc_corrupt=0.5974 corrupt_frac=1.0000 loss=2.7648 loss_recon=2.7648 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4984 init_acc_corrupt=0.4674 init_gold_top10=0.4961 init_gold_top100=0.5018
+step=173700 micro_steps=347400 elapsed=64.2s lr=3.000000e-04 loss_all=2.7697 acc_all=0.5969 loss_corrupt=2.7697 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7697 loss_recon=2.7697 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4976 init_acc_corrupt=0.4679 init_gold_top10=0.4970 init_gold_top100=0.5026
+step=173800 micro_steps=347600 elapsed=66.5s lr=3.000000e-04 loss_all=2.7864 acc_all=0.5948 loss_corrupt=2.7864 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7864 loss_recon=2.7864 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4991 init_acc_corrupt=0.4656 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=173900 micro_steps=347800 elapsed=62.1s lr=3.000000e-04 loss_all=2.7617 acc_all=0.5978 loss_corrupt=2.7617 acc_corrupt=0.5978 corrupt_frac=1.0000 loss=2.7617 loss_recon=2.7617 loss_meanflow=0.0000 mean_model_t=0.5039 mean_corrupt_t=0.5039 wrong_frac=0.4967 init_acc_corrupt=0.4695 init_gold_top10=0.4978 init_gold_top100=0.5035
+step=174000 micro_steps=348000 elapsed=62.1s lr=3.000000e-04 loss_all=2.7932 acc_all=0.5936 loss_corrupt=2.7932 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7932 loss_recon=2.7932 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5011 init_acc_corrupt=0.4640 init_gold_top10=0.4933 init_gold_top100=0.4991
+step=174100 micro_steps=348200 elapsed=64.0s lr=3.000000e-04 loss_all=2.7707 acc_all=0.5963 loss_corrupt=2.7707 acc_corrupt=0.5963 corrupt_frac=1.0000 loss=2.7707 loss_recon=2.7707 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4981 init_acc_corrupt=0.4669 init_gold_top10=0.4963 init_gold_top100=0.5021
+step=174200 micro_steps=348400 elapsed=59.8s lr=3.000000e-04 loss_all=2.7509 acc_all=0.5995 loss_corrupt=2.7509 acc_corrupt=0.5995 corrupt_frac=1.0000 loss=2.7509 loss_recon=2.7509 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.5041 wrong_frac=0.4954 init_acc_corrupt=0.4703 init_gold_top10=0.4990 init_gold_top100=0.5048
+step=174300 micro_steps=348600 elapsed=68.0s lr=3.000000e-04 loss_all=2.7662 acc_all=0.5973 loss_corrupt=2.7662 acc_corrupt=0.5973 corrupt_frac=1.0000 loss=2.7662 loss_recon=2.7662 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4973 init_acc_corrupt=0.4688 init_gold_top10=0.4971 init_gold_top100=0.5029
+step=174400 micro_steps=348800 elapsed=97.8s lr=3.000000e-04 loss_all=2.8008 acc_all=0.5930 loss_corrupt=2.8008 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.8008 loss_recon=2.8008 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5015 init_acc_corrupt=0.4637 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=174500 micro_steps=349000 elapsed=98.1s lr=3.000000e-04 loss_all=2.7656 acc_all=0.5974 loss_corrupt=2.7656 acc_corrupt=0.5974 corrupt_frac=1.0000 loss=2.7656 loss_recon=2.7656 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4973 init_acc_corrupt=0.4689 init_gold_top10=0.4972 init_gold_top100=0.5028
+step=174600 micro_steps=349200 elapsed=97.4s lr=3.000000e-04 loss_all=2.7874 acc_all=0.5948 loss_corrupt=2.7874 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7874 loss_recon=2.7874 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4986 init_acc_corrupt=0.4664 init_gold_top10=0.4957 init_gold_top100=0.5016
+step=174700 micro_steps=349400 elapsed=96.4s lr=3.000000e-04 loss_all=2.7769 acc_all=0.5957 loss_corrupt=2.7769 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7769 loss_recon=2.7769 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5000 init_acc_corrupt=0.4654 init_gold_top10=0.4943 init_gold_top100=0.5002
+step=174800 micro_steps=349600 elapsed=98.1s lr=3.000000e-04 loss_all=2.7677 acc_all=0.5976 loss_corrupt=2.7677 acc_corrupt=0.5976 corrupt_frac=1.0000 loss=2.7677 loss_recon=2.7677 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4969 init_acc_corrupt=0.4687 init_gold_top10=0.4975 init_gold_top100=0.5033
+step=174900 micro_steps=349800 elapsed=97.3s lr=3.000000e-04 loss_all=2.8064 acc_all=0.5917 loss_corrupt=2.8064 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8064 loss_recon=2.8064 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5023 init_acc_corrupt=0.4629 init_gold_top10=0.4919 init_gold_top100=0.4978
+step=175000 micro_steps=350000 elapsed=91.2s lr=3.000000e-04 loss_all=2.8045 acc_all=0.5922 loss_corrupt=2.8045 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8045 loss_recon=2.8045 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5022 init_acc_corrupt=0.4629 init_gold_top10=0.4922 init_gold_top100=0.4980
+step=175100 micro_steps=350200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7571 acc_all=0.5979 loss_corrupt=2.7571 acc_corrupt=0.5979 corrupt_frac=1.0000 loss=2.7571 loss_recon=2.7571 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4982 init_acc_corrupt=0.4677 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=175200 micro_steps=350400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7720 acc_all=0.5965 loss_corrupt=2.7720 acc_corrupt=0.5965 corrupt_frac=1.0000 loss=2.7720 loss_recon=2.7720 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4980 init_acc_corrupt=0.4678 init_gold_top10=0.4963 init_gold_top100=0.5021
+step=175300 micro_steps=350600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7883 acc_all=0.5946 loss_corrupt=2.7883 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7883 loss_recon=2.7883 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4995 init_acc_corrupt=0.4654 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=175400 micro_steps=350800 elapsed=66.6s lr=3.000000e-04 loss_all=2.7669 acc_all=0.5972 loss_corrupt=2.7669 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7669 loss_recon=2.7669 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4974 init_acc_corrupt=0.4691 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=175500 micro_steps=351000 elapsed=97.7s lr=3.000000e-04 loss_all=2.7505 acc_all=0.5998 loss_corrupt=2.7505 acc_corrupt=0.5998 corrupt_frac=1.0000 loss=2.7505 loss_recon=2.7505 loss_meanflow=0.0000 mean_model_t=0.5060 mean_corrupt_t=0.5060 wrong_frac=0.4949 init_acc_corrupt=0.4714 init_gold_top10=0.4997 init_gold_top100=0.5053
+step=175600 micro_steps=351200 elapsed=97.9s lr=3.000000e-04 loss_all=2.7940 acc_all=0.5941 loss_corrupt=2.7940 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7940 loss_recon=2.7940 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5008 init_acc_corrupt=0.4647 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=175700 micro_steps=351400 elapsed=97.2s lr=3.000000e-04 loss_all=2.7883 acc_all=0.5943 loss_corrupt=2.7883 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7883 loss_recon=2.7883 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5008 init_acc_corrupt=0.4649 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=175800 micro_steps=351600 elapsed=75.1s lr=3.000000e-04 loss_all=2.7609 acc_all=0.5974 loss_corrupt=2.7609 acc_corrupt=0.5974 corrupt_frac=1.0000 loss=2.7609 loss_recon=2.7609 loss_meanflow=0.0000 mean_model_t=0.5017 mean_corrupt_t=0.5017 wrong_frac=0.4974 init_acc_corrupt=0.4680 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=175900 micro_steps=351800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8298 acc_all=0.5890 loss_corrupt=2.8298 acc_corrupt=0.5890 corrupt_frac=1.0000 loss=2.8298 loss_recon=2.8298 loss_meanflow=0.0000 mean_model_t=0.4944 mean_corrupt_t=0.4944 wrong_frac=0.5057 init_acc_corrupt=0.4592 init_gold_top10=0.4885 init_gold_top100=0.4944
+step=176000 micro_steps=352000 elapsed=92.4s lr=3.000000e-04 loss_all=2.8128 acc_all=0.5913 loss_corrupt=2.8128 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8128 loss_recon=2.8128 loss_meanflow=0.0000 mean_model_t=0.4957 mean_corrupt_t=0.4957 wrong_frac=0.5038 init_acc_corrupt=0.4620 init_gold_top10=0.4904 init_gold_top100=0.4962
+step=176100 micro_steps=352200 elapsed=99.4s lr=3.000000e-04 loss_all=2.7486 acc_all=0.5995 loss_corrupt=2.7486 acc_corrupt=0.5995 corrupt_frac=1.0000 loss=2.7486 loss_recon=2.7486 loss_meanflow=0.0000 mean_model_t=0.5044 mean_corrupt_t=0.5044 wrong_frac=0.4959 init_acc_corrupt=0.4703 init_gold_top10=0.4986 init_gold_top100=0.5042
+step=176200 micro_steps=352400 elapsed=97.9s lr=3.000000e-04 loss_all=2.7441 acc_all=0.6004 loss_corrupt=2.7441 acc_corrupt=0.6004 corrupt_frac=1.0000 loss=2.7441 loss_recon=2.7441 loss_meanflow=0.0000 mean_model_t=0.5048 mean_corrupt_t=0.5048 wrong_frac=0.4949 init_acc_corrupt=0.4713 init_gold_top10=0.4996 init_gold_top100=0.5053
+step=176300 micro_steps=352600 elapsed=68.3s lr=3.000000e-04 loss_all=2.7764 acc_all=0.5962 loss_corrupt=2.7764 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7764 loss_recon=2.7764 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4987 init_acc_corrupt=0.4674 init_gold_top10=0.4958 init_gold_top100=0.5014
+step=176400 micro_steps=352800 elapsed=63.2s lr=3.000000e-04 loss_all=2.7646 acc_all=0.5976 loss_corrupt=2.7646 acc_corrupt=0.5976 corrupt_frac=1.0000 loss=2.7646 loss_recon=2.7646 loss_meanflow=0.0000 mean_model_t=0.5023 mean_corrupt_t=0.5023 wrong_frac=0.4978 init_acc_corrupt=0.4679 init_gold_top10=0.4968 init_gold_top100=0.5023
+step=176500 micro_steps=353000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8319 acc_all=0.5888 loss_corrupt=2.8319 acc_corrupt=0.5888 corrupt_frac=1.0000 loss=2.8319 loss_recon=2.8319 loss_meanflow=0.0000 mean_model_t=0.4949 mean_corrupt_t=0.4949 wrong_frac=0.5048 init_acc_corrupt=0.4602 init_gold_top10=0.4896 init_gold_top100=0.4953
+step=176600 micro_steps=353200 elapsed=96.0s lr=3.000000e-04 loss_all=2.7580 acc_all=0.5978 loss_corrupt=2.7580 acc_corrupt=0.5978 corrupt_frac=1.0000 loss=2.7580 loss_recon=2.7580 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4974 init_acc_corrupt=0.4688 init_gold_top10=0.4970 init_gold_top100=0.5028
+step=176700 micro_steps=353400 elapsed=95.8s lr=3.000000e-04 loss_all=2.7835 acc_all=0.5949 loss_corrupt=2.7835 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7835 loss_recon=2.7835 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4997 init_acc_corrupt=0.4652 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=176800 micro_steps=353600 elapsed=95.5s lr=3.000000e-04 loss_all=2.7634 acc_all=0.5979 loss_corrupt=2.7634 acc_corrupt=0.5979 corrupt_frac=1.0000 loss=2.7634 loss_recon=2.7634 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4973 init_acc_corrupt=0.4683 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=176900 micro_steps=353800 elapsed=96.2s lr=3.000000e-04 loss_all=2.8054 acc_all=0.5920 loss_corrupt=2.8054 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8054 loss_recon=2.8054 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5023 init_acc_corrupt=0.4629 init_gold_top10=0.4923 init_gold_top100=0.4978
+step=177000 micro_steps=354000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7884 acc_all=0.5939 loss_corrupt=2.7884 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7884 loss_recon=2.7884 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5011 init_acc_corrupt=0.4646 init_gold_top10=0.4933 init_gold_top100=0.4991
+step=177100 micro_steps=354200 elapsed=65.1s lr=3.000000e-04 loss_all=2.7458 acc_all=0.6001 loss_corrupt=2.7458 acc_corrupt=0.6001 corrupt_frac=1.0000 loss=2.7458 loss_recon=2.7458 loss_meanflow=0.0000 mean_model_t=0.5058 mean_corrupt_t=0.5058 wrong_frac=0.4946 init_acc_corrupt=0.4719 init_gold_top10=0.4999 init_gold_top100=0.5055
+step=177200 micro_steps=354400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7845 acc_all=0.5950 loss_corrupt=2.7845 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7845 loss_recon=2.7845 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5000 init_acc_corrupt=0.4657 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=177300 micro_steps=354600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7668 acc_all=0.5976 loss_corrupt=2.7668 acc_corrupt=0.5976 corrupt_frac=1.0000 loss=2.7668 loss_recon=2.7668 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4972 init_acc_corrupt=0.4683 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=177400 micro_steps=354800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7785 acc_all=0.5959 loss_corrupt=2.7785 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7785 loss_recon=2.7785 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4988 init_acc_corrupt=0.4667 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=177500 micro_steps=355000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7746 acc_all=0.5962 loss_corrupt=2.7746 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7746 loss_recon=2.7746 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4984 init_acc_corrupt=0.4672 init_gold_top10=0.4961 init_gold_top100=0.5017
+step=177600 micro_steps=355200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7735 acc_all=0.5962 loss_corrupt=2.7735 acc_corrupt=0.5962 corrupt_frac=1.0000 loss=2.7735 loss_recon=2.7735 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4987 init_acc_corrupt=0.4669 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=177700 micro_steps=355400 elapsed=63.3s lr=3.000000e-04 loss_all=2.8200 acc_all=0.5902 loss_corrupt=2.8200 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8200 loss_recon=2.8200 loss_meanflow=0.0000 mean_model_t=0.4951 mean_corrupt_t=0.4951 wrong_frac=0.5044 init_acc_corrupt=0.4608 init_gold_top10=0.4898 init_gold_top100=0.4957
+step=177800 micro_steps=355600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8172 acc_all=0.5905 loss_corrupt=2.8172 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8172 loss_recon=2.8172 loss_meanflow=0.0000 mean_model_t=0.4968 mean_corrupt_t=0.4968 wrong_frac=0.5038 init_acc_corrupt=0.4611 init_gold_top10=0.4905 init_gold_top100=0.4963
+step=177900 micro_steps=355800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7568 acc_all=0.5987 loss_corrupt=2.7568 acc_corrupt=0.5987 corrupt_frac=1.0000 loss=2.7568 loss_recon=2.7568 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4967 init_acc_corrupt=0.4697 init_gold_top10=0.4979 init_gold_top100=0.5035
+step=178000 micro_steps=356000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7795 acc_all=0.5959 loss_corrupt=2.7795 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7795 loss_recon=2.7795 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4991 init_acc_corrupt=0.4665 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=178100 micro_steps=356200 elapsed=65.1s lr=3.000000e-04 loss_all=2.8102 acc_all=0.5915 loss_corrupt=2.8102 acc_corrupt=0.5915 corrupt_frac=1.0000 loss=2.8102 loss_recon=2.8102 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5029 init_acc_corrupt=0.4622 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=178200 micro_steps=356400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7641 acc_all=0.5976 loss_corrupt=2.7641 acc_corrupt=0.5976 corrupt_frac=1.0000 loss=2.7641 loss_recon=2.7641 loss_meanflow=0.0000 mean_model_t=0.5028 mean_corrupt_t=0.5028 wrong_frac=0.4972 init_acc_corrupt=0.4689 init_gold_top10=0.4973 init_gold_top100=0.5030
+step=178300 micro_steps=356600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7798 acc_all=0.5956 loss_corrupt=2.7798 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7798 loss_recon=2.7798 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4987 init_acc_corrupt=0.4667 init_gold_top10=0.4958 init_gold_top100=0.5014
+step=178400 micro_steps=356800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7815 acc_all=0.5956 loss_corrupt=2.7815 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7815 loss_recon=2.7815 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4991 init_acc_corrupt=0.4663 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=178500 micro_steps=357000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8037 acc_all=0.5922 loss_corrupt=2.8037 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8037 loss_recon=2.8037 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5018 init_acc_corrupt=0.4629 init_gold_top10=0.4924 init_gold_top100=0.4983
+step=178600 micro_steps=357200 elapsed=63.2s lr=3.000000e-04 loss_all=2.7852 acc_all=0.5949 loss_corrupt=2.7852 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7852 loss_recon=2.7852 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4991 init_acc_corrupt=0.4661 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=178700 micro_steps=357400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7611 acc_all=0.5977 loss_corrupt=2.7611 acc_corrupt=0.5977 corrupt_frac=1.0000 loss=2.7611 loss_recon=2.7611 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4981 init_acc_corrupt=0.4684 init_gold_top10=0.4965 init_gold_top100=0.5021
+step=178800 micro_steps=357600 elapsed=63.6s lr=3.000000e-04 loss_all=2.7701 acc_all=0.5968 loss_corrupt=2.7701 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7701 loss_recon=2.7701 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4972 init_acc_corrupt=0.4689 init_gold_top10=0.4972 init_gold_top100=0.5030
+step=178900 micro_steps=357800 elapsed=86.4s lr=3.000000e-04 loss_all=2.7766 acc_all=0.5957 loss_corrupt=2.7766 acc_corrupt=0.5957 corrupt_frac=1.0000 loss=2.7766 loss_recon=2.7766 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4992 init_acc_corrupt=0.4664 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=179000 micro_steps=358000 elapsed=97.0s lr=3.000000e-04 loss_all=2.7826 acc_all=0.5954 loss_corrupt=2.7826 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7826 loss_recon=2.7826 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4997 init_acc_corrupt=0.4655 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=179100 micro_steps=358200 elapsed=98.6s lr=3.000000e-04 loss_all=2.8203 acc_all=0.5908 loss_corrupt=2.8203 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8203 loss_recon=2.8203 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5031 init_acc_corrupt=0.4617 init_gold_top10=0.4913 init_gold_top100=0.4971
+step=179200 micro_steps=358400 elapsed=96.6s lr=3.000000e-04 loss_all=2.7937 acc_all=0.5938 loss_corrupt=2.7937 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7937 loss_recon=2.7937 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.4998 init_acc_corrupt=0.4657 init_gold_top10=0.4945 init_gold_top100=0.5003
+step=179300 micro_steps=358600 elapsed=96.9s lr=3.000000e-04 loss_all=2.7747 acc_all=0.5961 loss_corrupt=2.7747 acc_corrupt=0.5961 corrupt_frac=1.0000 loss=2.7747 loss_recon=2.7747 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4985 init_acc_corrupt=0.4674 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=179400 micro_steps=358800 elapsed=96.8s lr=3.000000e-04 loss_all=2.7388 acc_all=0.6007 loss_corrupt=2.7388 acc_corrupt=0.6007 corrupt_frac=1.0000 loss=2.7388 loss_recon=2.7388 loss_meanflow=0.0000 mean_model_t=0.5058 mean_corrupt_t=0.5058 wrong_frac=0.4948 init_acc_corrupt=0.4717 init_gold_top10=0.4998 init_gold_top100=0.5054
+step=179500 micro_steps=359000 elapsed=97.2s lr=3.000000e-04 loss_all=2.7834 acc_all=0.5953 loss_corrupt=2.7834 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7834 loss_recon=2.7834 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4986 init_acc_corrupt=0.4665 init_gold_top10=0.4958 init_gold_top100=0.5016
+step=179600 micro_steps=359200 elapsed=96.9s lr=3.000000e-04 loss_all=2.7724 acc_all=0.5969 loss_corrupt=2.7724 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7724 loss_recon=2.7724 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4974 init_acc_corrupt=0.4682 init_gold_top10=0.4968 init_gold_top100=0.5027
+step=179700 micro_steps=359400 elapsed=96.8s lr=3.000000e-04 loss_all=2.7570 acc_all=0.5990 loss_corrupt=2.7570 acc_corrupt=0.5990 corrupt_frac=1.0000 loss=2.7570 loss_recon=2.7570 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4970 init_acc_corrupt=0.4693 init_gold_top10=0.4976 init_gold_top100=0.5032
+step=179800 micro_steps=359600 elapsed=82.1s lr=3.000000e-04 loss_all=2.8026 acc_all=0.5922 loss_corrupt=2.8026 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8026 loss_recon=2.8026 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5020 init_acc_corrupt=0.4630 init_gold_top10=0.4923 init_gold_top100=0.4982
+step=179900 micro_steps=359800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7969 acc_all=0.5930 loss_corrupt=2.7969 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.7969 loss_recon=2.7969 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5027 init_acc_corrupt=0.4633 init_gold_top10=0.4918 init_gold_top100=0.4974
+step=180000 micro_steps=360000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7212 acc_all=0.6034 loss_corrupt=2.7212 acc_corrupt=0.6034 corrupt_frac=1.0000 loss=2.7212 loss_recon=2.7212 loss_meanflow=0.0000 mean_model_t=0.5075 mean_corrupt_t=0.5075 wrong_frac=0.4927 init_acc_corrupt=0.4740 init_gold_top10=0.5019 init_gold_top100=0.5075
+step=180100 micro_steps=360200 elapsed=66.6s lr=3.000000e-04 loss_all=2.8091 acc_all=0.5918 loss_corrupt=2.8091 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8091 loss_recon=2.8091 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5029 init_acc_corrupt=0.4623 init_gold_top10=0.4916 init_gold_top100=0.4973
+step=180200 micro_steps=360400 elapsed=63.1s lr=3.000000e-04 loss_all=2.7795 acc_all=0.5959 loss_corrupt=2.7795 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7795 loss_recon=2.7795 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4986 init_acc_corrupt=0.4671 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=180300 micro_steps=360600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7832 acc_all=0.5950 loss_corrupt=2.7832 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7832 loss_recon=2.7832 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5006 init_acc_corrupt=0.4653 init_gold_top10=0.4937 init_gold_top100=0.4995
+step=180400 micro_steps=360800 elapsed=65.4s lr=3.000000e-04 loss_all=2.8178 acc_all=0.5905 loss_corrupt=2.8178 acc_corrupt=0.5905 corrupt_frac=1.0000 loss=2.8178 loss_recon=2.8178 loss_meanflow=0.0000 mean_model_t=0.4959 mean_corrupt_t=0.4959 wrong_frac=0.5041 init_acc_corrupt=0.4609 init_gold_top10=0.4903 init_gold_top100=0.4960
+step=180500 micro_steps=361000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7937 acc_all=0.5932 loss_corrupt=2.7937 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7937 loss_recon=2.7937 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5025 init_acc_corrupt=0.4626 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=180600 micro_steps=361200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7687 acc_all=0.5973 loss_corrupt=2.7687 acc_corrupt=0.5973 corrupt_frac=1.0000 loss=2.7687 loss_recon=2.7687 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4967 init_acc_corrupt=0.4686 init_gold_top10=0.4976 init_gold_top100=0.5034
+step=180700 micro_steps=361400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8145 acc_all=0.5906 loss_corrupt=2.8145 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8145 loss_recon=2.8145 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5042 init_acc_corrupt=0.4609 init_gold_top10=0.4902 init_gold_top100=0.4960
+step=180800 micro_steps=361600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7832 acc_all=0.5949 loss_corrupt=2.7832 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7832 loss_recon=2.7832 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5002 init_acc_corrupt=0.4649 init_gold_top10=0.4940 init_gold_top100=0.4998
+step=180900 micro_steps=361800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7874 acc_all=0.5941 loss_corrupt=2.7874 acc_corrupt=0.5941 corrupt_frac=1.0000 loss=2.7874 loss_recon=2.7874 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4997 init_acc_corrupt=0.4659 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=181000 micro_steps=362000 elapsed=63.7s lr=3.000000e-04 loss_all=2.7931 acc_all=0.5939 loss_corrupt=2.7931 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7931 loss_recon=2.7931 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5008 init_acc_corrupt=0.4644 init_gold_top10=0.4936 init_gold_top100=0.4993
+step=181100 micro_steps=362200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7820 acc_all=0.5952 loss_corrupt=2.7820 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7820 loss_recon=2.7820 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5000 init_acc_corrupt=0.4655 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=181200 micro_steps=362400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7779 acc_all=0.5954 loss_corrupt=2.7779 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7779 loss_recon=2.7779 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4989 init_acc_corrupt=0.4670 init_gold_top10=0.4956 init_gold_top100=0.5012
+step=181300 micro_steps=362600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7858 acc_all=0.5946 loss_corrupt=2.7858 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7858 loss_recon=2.7858 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5001 init_acc_corrupt=0.4656 init_gold_top10=0.4943 init_gold_top100=0.5001
+step=181400 micro_steps=362800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7404 acc_all=0.6006 loss_corrupt=2.7404 acc_corrupt=0.6006 corrupt_frac=1.0000 loss=2.7404 loss_recon=2.7404 loss_meanflow=0.0000 mean_model_t=0.5054 mean_corrupt_t=0.5054 wrong_frac=0.4947 init_acc_corrupt=0.4717 init_gold_top10=0.4998 init_gold_top100=0.5054
+step=181500 micro_steps=363000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7755 acc_all=0.5958 loss_corrupt=2.7755 acc_corrupt=0.5958 corrupt_frac=1.0000 loss=2.7755 loss_recon=2.7755 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4994 init_acc_corrupt=0.4667 init_gold_top10=0.4951 init_gold_top100=0.5007
+step=181600 micro_steps=363200 elapsed=63.1s lr=3.000000e-04 loss_all=2.8025 acc_all=0.5919 loss_corrupt=2.8025 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8025 loss_recon=2.8025 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5021 init_acc_corrupt=0.4629 init_gold_top10=0.4922 init_gold_top100=0.4981
+step=181700 micro_steps=363400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7910 acc_all=0.5943 loss_corrupt=2.7910 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7910 loss_recon=2.7910 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5005 init_acc_corrupt=0.4651 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=181800 micro_steps=363600 elapsed=63.3s lr=3.000000e-04 loss_all=2.8254 acc_all=0.5895 loss_corrupt=2.8254 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8254 loss_recon=2.8254 loss_meanflow=0.0000 mean_model_t=0.4944 mean_corrupt_t=0.4944 wrong_frac=0.5059 init_acc_corrupt=0.4589 init_gold_top10=0.4884 init_gold_top100=0.4943
+step=181900 micro_steps=363800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7635 acc_all=0.5979 loss_corrupt=2.7635 acc_corrupt=0.5979 corrupt_frac=1.0000 loss=2.7635 loss_recon=2.7635 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4970 init_acc_corrupt=0.4691 init_gold_top10=0.4974 init_gold_top100=0.5030
+step=182000 micro_steps=364000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7708 acc_all=0.5965 loss_corrupt=2.7708 acc_corrupt=0.5965 corrupt_frac=1.0000 loss=2.7708 loss_recon=2.7708 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4990 init_acc_corrupt=0.4671 init_gold_top10=0.4956 init_gold_top100=0.5012
+step=182100 micro_steps=364200 elapsed=64.9s lr=3.000000e-04 loss_all=2.7676 acc_all=0.5972 loss_corrupt=2.7676 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7676 loss_recon=2.7676 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4976 init_acc_corrupt=0.4683 init_gold_top10=0.4968 init_gold_top100=0.5025
+step=182200 micro_steps=364400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7916 acc_all=0.5935 loss_corrupt=2.7916 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7916 loss_recon=2.7916 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5013 init_acc_corrupt=0.4641 init_gold_top10=0.4931 init_gold_top100=0.4989
+step=182300 micro_steps=364600 elapsed=60.2s lr=3.000000e-04 loss_all=2.8200 acc_all=0.5901 loss_corrupt=2.8200 acc_corrupt=0.5901 corrupt_frac=1.0000 loss=2.8200 loss_recon=2.8200 loss_meanflow=0.0000 mean_model_t=0.4954 mean_corrupt_t=0.4954 wrong_frac=0.5046 init_acc_corrupt=0.4604 init_gold_top10=0.4897 init_gold_top100=0.4954
+step=182400 micro_steps=364800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7646 acc_all=0.5975 loss_corrupt=2.7646 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7646 loss_recon=2.7646 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4966 init_acc_corrupt=0.4690 init_gold_top10=0.4977 init_gold_top100=0.5035
+step=182500 micro_steps=365000 elapsed=63.1s lr=3.000000e-04 loss_all=2.7559 acc_all=0.5980 loss_corrupt=2.7559 acc_corrupt=0.5980 corrupt_frac=1.0000 loss=2.7559 loss_recon=2.7559 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4985 init_acc_corrupt=0.4674 init_gold_top10=0.4961 init_gold_top100=0.5016
+step=182600 micro_steps=365200 elapsed=63.4s lr=3.000000e-04 loss_all=2.8015 acc_all=0.5928 loss_corrupt=2.8015 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8015 loss_recon=2.8015 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5024 init_acc_corrupt=0.4625 init_gold_top10=0.4919 init_gold_top100=0.4977
+step=182700 micro_steps=365400 elapsed=63.5s lr=3.000000e-04 loss_all=2.8054 acc_all=0.5918 loss_corrupt=2.8054 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8054 loss_recon=2.8054 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5045 init_acc_corrupt=0.4613 init_gold_top10=0.4898 init_gold_top100=0.4956
+step=182800 micro_steps=365600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7675 acc_all=0.5972 loss_corrupt=2.7675 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7675 loss_recon=2.7675 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4985 init_acc_corrupt=0.4671 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=182900 micro_steps=365800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7724 acc_all=0.5964 loss_corrupt=2.7724 acc_corrupt=0.5964 corrupt_frac=1.0000 loss=2.7724 loss_recon=2.7724 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4995 init_acc_corrupt=0.4666 init_gold_top10=0.4950 init_gold_top100=0.5007
+step=183000 micro_steps=366000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7712 acc_all=0.5961 loss_corrupt=2.7712 acc_corrupt=0.5961 corrupt_frac=1.0000 loss=2.7712 loss_recon=2.7712 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4988 init_acc_corrupt=0.4664 init_gold_top10=0.4957 init_gold_top100=0.5014
+step=183100 micro_steps=366200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7688 acc_all=0.5966 loss_corrupt=2.7688 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7688 loss_recon=2.7688 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4983 init_acc_corrupt=0.4671 init_gold_top10=0.4961 init_gold_top100=0.5019
+step=183200 micro_steps=366400 elapsed=63.6s lr=3.000000e-04 loss_all=2.8335 acc_all=0.5882 loss_corrupt=2.8335 acc_corrupt=0.5882 corrupt_frac=1.0000 loss=2.8335 loss_recon=2.8335 loss_meanflow=0.0000 mean_model_t=0.4941 mean_corrupt_t=0.4941 wrong_frac=0.5062 init_acc_corrupt=0.4579 init_gold_top10=0.4880 init_gold_top100=0.4939
+step=183300 micro_steps=366600 elapsed=63.6s lr=3.000000e-04 loss_all=2.7735 acc_all=0.5963 loss_corrupt=2.7735 acc_corrupt=0.5963 corrupt_frac=1.0000 loss=2.7735 loss_recon=2.7735 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4996 init_acc_corrupt=0.4660 init_gold_top10=0.4948 init_gold_top100=0.5005
+step=183400 micro_steps=366800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8039 acc_all=0.5922 loss_corrupt=2.8039 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8039 loss_recon=2.8039 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5025 init_acc_corrupt=0.4624 init_gold_top10=0.4919 init_gold_top100=0.4976
+step=183500 micro_steps=367000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8061 acc_all=0.5923 loss_corrupt=2.8061 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8061 loss_recon=2.8061 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5029 init_acc_corrupt=0.4621 init_gold_top10=0.4914 init_gold_top100=0.4971
+step=183600 micro_steps=367200 elapsed=63.5s lr=3.000000e-04 loss_all=2.8240 acc_all=0.5899 loss_corrupt=2.8240 acc_corrupt=0.5899 corrupt_frac=1.0000 loss=2.8240 loss_recon=2.8240 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5035 init_acc_corrupt=0.4610 init_gold_top10=0.4907 init_gold_top100=0.4966
+step=183700 micro_steps=367400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7913 acc_all=0.5946 loss_corrupt=2.7913 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7913 loss_recon=2.7913 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5008 init_acc_corrupt=0.4648 init_gold_top10=0.4937 init_gold_top100=0.4994
+step=183800 micro_steps=367600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7868 acc_all=0.5942 loss_corrupt=2.7868 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7868 loss_recon=2.7868 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4980 wrong_frac=0.5015 init_acc_corrupt=0.4636 init_gold_top10=0.4929 init_gold_top100=0.4986
+step=183900 micro_steps=367800 elapsed=63.0s lr=3.000000e-04 loss_all=2.8093 acc_all=0.5917 loss_corrupt=2.8093 acc_corrupt=0.5917 corrupt_frac=1.0000 loss=2.8093 loss_recon=2.8093 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5022 init_acc_corrupt=0.4626 init_gold_top10=0.4921 init_gold_top100=0.4979
+step=184000 micro_steps=368000 elapsed=63.2s lr=3.000000e-04 loss_all=2.8015 acc_all=0.5927 loss_corrupt=2.8015 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.8015 loss_recon=2.8015 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5026 init_acc_corrupt=0.4623 init_gold_top10=0.4917 init_gold_top100=0.4974
+step=184100 micro_steps=368200 elapsed=64.7s lr=3.000000e-04 loss_all=2.7855 acc_all=0.5950 loss_corrupt=2.7855 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7855 loss_recon=2.7855 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4990 init_acc_corrupt=0.4661 init_gold_top10=0.4953 init_gold_top100=0.5012
+step=184200 micro_steps=368400 elapsed=66.5s lr=3.000000e-04 loss_all=2.7811 acc_all=0.5955 loss_corrupt=2.7811 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7811 loss_recon=2.7811 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4983 init_acc_corrupt=0.4669 init_gold_top10=0.4962 init_gold_top100=0.5019
+step=184300 micro_steps=368600 elapsed=96.9s lr=3.000000e-04 loss_all=2.7558 acc_all=0.5982 loss_corrupt=2.7558 acc_corrupt=0.5982 corrupt_frac=1.0000 loss=2.7558 loss_recon=2.7558 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4981 init_acc_corrupt=0.4685 init_gold_top10=0.4965 init_gold_top100=0.5022
+step=184400 micro_steps=368800 elapsed=96.6s lr=3.000000e-04 loss_all=2.7879 acc_all=0.5946 loss_corrupt=2.7879 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7879 loss_recon=2.7879 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5004 init_acc_corrupt=0.4647 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=184500 micro_steps=369000 elapsed=116.5s lr=3.000000e-04 loss_all=2.7868 acc_all=0.5945 loss_corrupt=2.7868 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7868 loss_recon=2.7868 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.5002 init_acc_corrupt=0.4654 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=184600 micro_steps=369200 elapsed=120.5s lr=3.000000e-04 loss_all=2.7538 acc_all=0.5992 loss_corrupt=2.7538 acc_corrupt=0.5992 corrupt_frac=1.0000 loss=2.7538 loss_recon=2.7538 loss_meanflow=0.0000 mean_model_t=0.5047 mean_corrupt_t=0.5047 wrong_frac=0.4955 init_acc_corrupt=0.4701 init_gold_top10=0.4991 init_gold_top100=0.5047
+step=184700 micro_steps=369400 elapsed=97.0s lr=3.000000e-04 loss_all=2.7806 acc_all=0.5946 loss_corrupt=2.7806 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7806 loss_recon=2.7806 loss_meanflow=0.0000 mean_model_t=0.4987 mean_corrupt_t=0.4987 wrong_frac=0.5012 init_acc_corrupt=0.4641 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=184800 micro_steps=369600 elapsed=97.2s lr=3.000000e-04 loss_all=2.7371 acc_all=0.6010 loss_corrupt=2.7371 acc_corrupt=0.6010 corrupt_frac=1.0000 loss=2.7371 loss_recon=2.7371 loss_meanflow=0.0000 mean_model_t=0.5053 mean_corrupt_t=0.5053 wrong_frac=0.4947 init_acc_corrupt=0.4714 init_gold_top10=0.4998 init_gold_top100=0.5054
+step=184900 micro_steps=369800 elapsed=122.3s lr=3.000000e-04 loss_all=2.8253 acc_all=0.5892 loss_corrupt=2.8253 acc_corrupt=0.5892 corrupt_frac=1.0000 loss=2.8253 loss_recon=2.8253 loss_meanflow=0.0000 mean_model_t=0.4926 mean_corrupt_t=0.4926 wrong_frac=0.5071 init_acc_corrupt=0.4574 init_gold_top10=0.4872 init_gold_top100=0.4931
+step=185000 micro_steps=370000 elapsed=133.3s lr=3.000000e-04 loss_all=2.7718 acc_all=0.5964 loss_corrupt=2.7718 acc_corrupt=0.5964 corrupt_frac=1.0000 loss=2.7718 loss_recon=2.7718 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4980 init_acc_corrupt=0.4677 init_gold_top10=0.4964 init_gold_top100=0.5021
+step=185100 micro_steps=370200 elapsed=132.5s lr=3.000000e-04 loss_all=2.8046 acc_all=0.5920 loss_corrupt=2.8046 acc_corrupt=0.5920 corrupt_frac=1.0000 loss=2.8046 loss_recon=2.8046 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5033 init_acc_corrupt=0.4619 init_gold_top10=0.4910 init_gold_top100=0.4968
+step=185200 micro_steps=370400 elapsed=131.8s lr=3.000000e-04 loss_all=2.8061 acc_all=0.5919 loss_corrupt=2.8061 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8061 loss_recon=2.8061 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5014 init_acc_corrupt=0.4633 init_gold_top10=0.4929 init_gold_top100=0.4988
+step=185300 micro_steps=370600 elapsed=131.2s lr=3.000000e-04 loss_all=2.7960 acc_all=0.5933 loss_corrupt=2.7960 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7960 loss_recon=2.7960 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5010 init_acc_corrupt=0.4646 init_gold_top10=0.4932 init_gold_top100=0.4992
+step=185400 micro_steps=370800 elapsed=132.9s lr=3.000000e-04 loss_all=2.8050 acc_all=0.5923 loss_corrupt=2.8050 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8050 loss_recon=2.8050 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5032 init_acc_corrupt=0.4623 init_gold_top10=0.4913 init_gold_top100=0.4970
+step=185500 micro_steps=371000 elapsed=132.2s lr=3.000000e-04 loss_all=2.7795 acc_all=0.5956 loss_corrupt=2.7795 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7795 loss_recon=2.7795 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4990 init_acc_corrupt=0.4670 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=185600 micro_steps=371200 elapsed=131.0s lr=3.000000e-04 loss_all=2.7450 acc_all=0.5996 loss_corrupt=2.7450 acc_corrupt=0.5996 corrupt_frac=1.0000 loss=2.7450 loss_recon=2.7450 loss_meanflow=0.0000 mean_model_t=0.5038 mean_corrupt_t=0.5038 wrong_frac=0.4962 init_acc_corrupt=0.4697 init_gold_top10=0.4985 init_gold_top100=0.5040
+step=185700 micro_steps=371400 elapsed=131.1s lr=3.000000e-04 loss_all=2.7989 acc_all=0.5935 loss_corrupt=2.7989 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7989 loss_recon=2.7989 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5005 init_acc_corrupt=0.4647 init_gold_top10=0.4938 init_gold_top100=0.4996
+step=185800 micro_steps=371600 elapsed=131.1s lr=3.000000e-04 loss_all=2.7528 acc_all=0.5990 loss_corrupt=2.7528 acc_corrupt=0.5990 corrupt_frac=1.0000 loss=2.7528 loss_recon=2.7528 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4970 init_acc_corrupt=0.4689 init_gold_top10=0.4975 init_gold_top100=0.5032
+step=185900 micro_steps=371800 elapsed=131.0s lr=3.000000e-04 loss_all=2.8063 acc_all=0.5918 loss_corrupt=2.8063 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8063 loss_recon=2.8063 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5029 init_acc_corrupt=0.4620 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=186000 micro_steps=372000 elapsed=121.7s lr=3.000000e-04 loss_all=2.7837 acc_all=0.5954 loss_corrupt=2.7837 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7837 loss_recon=2.7837 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5000 init_acc_corrupt=0.4657 init_gold_top10=0.4942 init_gold_top100=0.5001
+step=186100 micro_steps=372200 elapsed=64.9s lr=3.000000e-04 loss_all=2.7778 acc_all=0.5956 loss_corrupt=2.7778 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7778 loss_recon=2.7778 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4999 init_acc_corrupt=0.4656 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=186200 micro_steps=372400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7921 acc_all=0.5936 loss_corrupt=2.7921 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7921 loss_recon=2.7921 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5024 init_acc_corrupt=0.4628 init_gold_top10=0.4920 init_gold_top100=0.4977
+step=186300 micro_steps=372600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7692 acc_all=0.5973 loss_corrupt=2.7692 acc_corrupt=0.5973 corrupt_frac=1.0000 loss=2.7692 loss_recon=2.7692 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4982 init_acc_corrupt=0.4673 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=186400 micro_steps=372800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8056 acc_all=0.5926 loss_corrupt=2.8056 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8056 loss_recon=2.8056 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5020 init_acc_corrupt=0.4630 init_gold_top10=0.4925 init_gold_top100=0.4981
+step=186500 micro_steps=373000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7663 acc_all=0.5970 loss_corrupt=2.7663 acc_corrupt=0.5970 corrupt_frac=1.0000 loss=2.7663 loss_recon=2.7663 loss_meanflow=0.0000 mean_model_t=0.5020 mean_corrupt_t=0.5020 wrong_frac=0.4978 init_acc_corrupt=0.4683 init_gold_top10=0.4967 init_gold_top100=0.5024
+step=186600 micro_steps=373200 elapsed=63.3s lr=3.000000e-04 loss_all=2.7867 acc_all=0.5948 loss_corrupt=2.7867 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7867 loss_recon=2.7867 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5002 init_acc_corrupt=0.4656 init_gold_top10=0.4942 init_gold_top100=0.5000
+step=186700 micro_steps=373400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7463 acc_all=0.5997 loss_corrupt=2.7463 acc_corrupt=0.5997 corrupt_frac=1.0000 loss=2.7463 loss_recon=2.7463 loss_meanflow=0.0000 mean_model_t=0.5039 mean_corrupt_t=0.5039 wrong_frac=0.4962 init_acc_corrupt=0.4698 init_gold_top10=0.4983 init_gold_top100=0.5039
+step=186800 micro_steps=373600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7970 acc_all=0.5932 loss_corrupt=2.7970 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7970 loss_recon=2.7970 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5018 init_acc_corrupt=0.4632 init_gold_top10=0.4926 init_gold_top100=0.4983
+step=186900 micro_steps=373800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7998 acc_all=0.5925 loss_corrupt=2.7998 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.7998 loss_recon=2.7998 loss_meanflow=0.0000 mean_model_t=0.4972 mean_corrupt_t=0.4972 wrong_frac=0.5026 init_acc_corrupt=0.4622 init_gold_top10=0.4917 init_gold_top100=0.4975
+step=187000 micro_steps=374000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7831 acc_all=0.5949 loss_corrupt=2.7831 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7831 loss_recon=2.7831 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5004 init_acc_corrupt=0.4651 init_gold_top10=0.4940 init_gold_top100=0.4996
+step=187100 micro_steps=374200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7707 acc_all=0.5967 loss_corrupt=2.7707 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7707 loss_recon=2.7707 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4982 init_acc_corrupt=0.4672 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=187200 micro_steps=374400 elapsed=63.6s lr=3.000000e-04 loss_all=2.7820 acc_all=0.5956 loss_corrupt=2.7820 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7820 loss_recon=2.7820 loss_meanflow=0.0000 mean_model_t=0.5006 mean_corrupt_t=0.5006 wrong_frac=0.4992 init_acc_corrupt=0.4659 init_gold_top10=0.4951 init_gold_top100=0.5010
+step=187300 micro_steps=374600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7756 acc_all=0.5956 loss_corrupt=2.7756 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7756 loss_recon=2.7756 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4998 init_acc_corrupt=0.4653 init_gold_top10=0.4946 init_gold_top100=0.5004
+step=187400 micro_steps=374800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7897 acc_all=0.5942 loss_corrupt=2.7897 acc_corrupt=0.5942 corrupt_frac=1.0000 loss=2.7897 loss_recon=2.7897 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5010 init_acc_corrupt=0.4644 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=187500 micro_steps=375000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7621 acc_all=0.5984 loss_corrupt=2.7621 acc_corrupt=0.5984 corrupt_frac=1.0000 loss=2.7621 loss_recon=2.7621 loss_meanflow=0.0000 mean_model_t=0.5031 mean_corrupt_t=0.5031 wrong_frac=0.4966 init_acc_corrupt=0.4687 init_gold_top10=0.4979 init_gold_top100=0.5036
+step=187600 micro_steps=375200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7675 acc_all=0.5973 loss_corrupt=2.7675 acc_corrupt=0.5973 corrupt_frac=1.0000 loss=2.7675 loss_recon=2.7675 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4979 init_acc_corrupt=0.4672 init_gold_top10=0.4966 init_gold_top100=0.5022
+step=187700 micro_steps=375400 elapsed=63.7s lr=3.000000e-04 loss_all=2.7850 acc_all=0.5945 loss_corrupt=2.7850 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7850 loss_recon=2.7850 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5009 init_acc_corrupt=0.4644 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=187800 micro_steps=375600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8041 acc_all=0.5925 loss_corrupt=2.8041 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8041 loss_recon=2.8041 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5020 init_acc_corrupt=0.4630 init_gold_top10=0.4924 init_gold_top100=0.4982
+step=187900 micro_steps=375800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7791 acc_all=0.5954 loss_corrupt=2.7791 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7791 loss_recon=2.7791 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.4996 init_acc_corrupt=0.4658 init_gold_top10=0.4948 init_gold_top100=0.5006
+step=188000 micro_steps=376000 elapsed=59.9s lr=3.000000e-04 loss_all=2.7410 acc_all=0.6009 loss_corrupt=2.7410 acc_corrupt=0.6009 corrupt_frac=1.0000 loss=2.7410 loss_recon=2.7410 loss_meanflow=0.0000 mean_model_t=0.5058 mean_corrupt_t=0.5058 wrong_frac=0.4943 init_acc_corrupt=0.4713 init_gold_top10=0.5004 init_gold_top100=0.5060
+step=188100 micro_steps=376200 elapsed=65.1s lr=3.000000e-04 loss_all=2.7800 acc_all=0.5955 loss_corrupt=2.7800 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7800 loss_recon=2.7800 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4999 init_acc_corrupt=0.4657 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=188200 micro_steps=376400 elapsed=63.6s lr=3.000000e-04 loss_all=2.7996 acc_all=0.5930 loss_corrupt=2.7996 acc_corrupt=0.5930 corrupt_frac=1.0000 loss=2.7996 loss_recon=2.7996 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5005 init_acc_corrupt=0.4644 init_gold_top10=0.4937 init_gold_top100=0.4996
+step=188300 micro_steps=376600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7798 acc_all=0.5953 loss_corrupt=2.7798 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7798 loss_recon=2.7798 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4990 init_acc_corrupt=0.4660 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=188400 micro_steps=376800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8070 acc_all=0.5923 loss_corrupt=2.8070 acc_corrupt=0.5923 corrupt_frac=1.0000 loss=2.8070 loss_recon=2.8070 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5024 init_acc_corrupt=0.4626 init_gold_top10=0.4921 init_gold_top100=0.4978
+step=188500 micro_steps=377000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7914 acc_all=0.5935 loss_corrupt=2.7914 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7914 loss_recon=2.7914 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5014 init_acc_corrupt=0.4635 init_gold_top10=0.4930 init_gold_top100=0.4988
+step=188600 micro_steps=377200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7638 acc_all=0.5982 loss_corrupt=2.7638 acc_corrupt=0.5982 corrupt_frac=1.0000 loss=2.7638 loss_recon=2.7638 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4972 init_acc_corrupt=0.4687 init_gold_top10=0.4971 init_gold_top100=0.5028
+step=188700 micro_steps=377400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7864 acc_all=0.5947 loss_corrupt=2.7864 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7864 loss_recon=2.7864 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5008 init_acc_corrupt=0.4646 init_gold_top10=0.4935 init_gold_top100=0.4993
+step=188800 micro_steps=377600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7445 acc_all=0.5995 loss_corrupt=2.7445 acc_corrupt=0.5995 corrupt_frac=1.0000 loss=2.7445 loss_recon=2.7445 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4975 init_acc_corrupt=0.4684 init_gold_top10=0.4973 init_gold_top100=0.5027
+step=188900 micro_steps=377800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7750 acc_all=0.5960 loss_corrupt=2.7750 acc_corrupt=0.5960 corrupt_frac=1.0000 loss=2.7750 loss_recon=2.7750 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4991 init_acc_corrupt=0.4669 init_gold_top10=0.4952 init_gold_top100=0.5010
+step=189000 micro_steps=378000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8008 acc_all=0.5925 loss_corrupt=2.8008 acc_corrupt=0.5925 corrupt_frac=1.0000 loss=2.8008 loss_recon=2.8008 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5026 init_acc_corrupt=0.4628 init_gold_top10=0.4918 init_gold_top100=0.4975
+step=189100 micro_steps=378200 elapsed=65.1s lr=3.000000e-04 loss_all=2.7925 acc_all=0.5940 loss_corrupt=2.7925 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7925 loss_recon=2.7925 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5009 init_acc_corrupt=0.4640 init_gold_top10=0.4934 init_gold_top100=0.4993
+step=189200 micro_steps=378400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7901 acc_all=0.5938 loss_corrupt=2.7901 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7901 loss_recon=2.7901 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.4989 wrong_frac=0.5011 init_acc_corrupt=0.4644 init_gold_top10=0.4933 init_gold_top100=0.4991
+step=189300 micro_steps=378600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7674 acc_all=0.5971 loss_corrupt=2.7674 acc_corrupt=0.5971 corrupt_frac=1.0000 loss=2.7674 loss_recon=2.7674 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4985 init_acc_corrupt=0.4673 init_gold_top10=0.4959 init_gold_top100=0.5016
+step=189400 micro_steps=378800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7922 acc_all=0.5940 loss_corrupt=2.7922 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7922 loss_recon=2.7922 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5009 init_acc_corrupt=0.4642 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=189500 micro_steps=379000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7633 acc_all=0.5974 loss_corrupt=2.7633 acc_corrupt=0.5974 corrupt_frac=1.0000 loss=2.7633 loss_recon=2.7633 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4972 init_acc_corrupt=0.4684 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=189600 micro_steps=379200 elapsed=63.6s lr=3.000000e-04 loss_all=2.7577 acc_all=0.5980 loss_corrupt=2.7577 acc_corrupt=0.5980 corrupt_frac=1.0000 loss=2.7577 loss_recon=2.7577 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4985 init_acc_corrupt=0.4675 init_gold_top10=0.4961 init_gold_top100=0.5016
+step=189700 micro_steps=379400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7905 acc_all=0.5936 loss_corrupt=2.7905 acc_corrupt=0.5936 corrupt_frac=1.0000 loss=2.7905 loss_recon=2.7905 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4978 wrong_frac=0.5019 init_acc_corrupt=0.4633 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=189800 micro_steps=379600 elapsed=63.6s lr=3.000000e-04 loss_all=2.7901 acc_all=0.5946 loss_corrupt=2.7901 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7901 loss_recon=2.7901 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5005 init_acc_corrupt=0.4649 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=189900 micro_steps=379800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7527 acc_all=0.5986 loss_corrupt=2.7527 acc_corrupt=0.5986 corrupt_frac=1.0000 loss=2.7527 loss_recon=2.7527 loss_meanflow=0.0000 mean_model_t=0.5029 mean_corrupt_t=0.5029 wrong_frac=0.4970 init_acc_corrupt=0.4696 init_gold_top10=0.4975 init_gold_top100=0.5031
+step=190000 micro_steps=380000 elapsed=63.7s lr=3.000000e-04 loss_all=2.7945 acc_all=0.5935 loss_corrupt=2.7945 acc_corrupt=0.5935 corrupt_frac=1.0000 loss=2.7945 loss_recon=2.7945 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5017 init_acc_corrupt=0.4634 init_gold_top10=0.4926 init_gold_top100=0.4984
+step=190100 micro_steps=380200 elapsed=64.9s lr=3.000000e-04 loss_all=2.7801 acc_all=0.5953 loss_corrupt=2.7801 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7801 loss_recon=2.7801 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4999 init_acc_corrupt=0.4654 init_gold_top10=0.4945 init_gold_top100=0.5002
+step=190200 micro_steps=380400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7618 acc_all=0.5976 loss_corrupt=2.7618 acc_corrupt=0.5976 corrupt_frac=1.0000 loss=2.7618 loss_recon=2.7618 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4972 init_acc_corrupt=0.4684 init_gold_top10=0.4971 init_gold_top100=0.5030
+step=190300 micro_steps=380600 elapsed=63.5s lr=3.000000e-04 loss_all=2.8089 acc_all=0.5919 loss_corrupt=2.8089 acc_corrupt=0.5919 corrupt_frac=1.0000 loss=2.8089 loss_recon=2.8089 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5027 init_acc_corrupt=0.4627 init_gold_top10=0.4916 init_gold_top100=0.4974
+step=190400 micro_steps=380800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7790 acc_all=0.5960 loss_corrupt=2.7790 acc_corrupt=0.5960 corrupt_frac=1.0000 loss=2.7790 loss_recon=2.7790 loss_meanflow=0.0000 mean_model_t=0.5008 mean_corrupt_t=0.5008 wrong_frac=0.4993 init_acc_corrupt=0.4666 init_gold_top10=0.4950 init_gold_top100=0.5008
+step=190500 micro_steps=381000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8073 acc_all=0.5913 loss_corrupt=2.8073 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8073 loss_recon=2.8073 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5033 init_acc_corrupt=0.4615 init_gold_top10=0.4909 init_gold_top100=0.4968
+step=190600 micro_steps=381200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7653 acc_all=0.5975 loss_corrupt=2.7653 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7653 loss_recon=2.7653 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4979 init_acc_corrupt=0.4677 init_gold_top10=0.4966 init_gold_top100=0.5022
+step=190700 micro_steps=381400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7732 acc_all=0.5960 loss_corrupt=2.7732 acc_corrupt=0.5960 corrupt_frac=1.0000 loss=2.7732 loss_recon=2.7732 loss_meanflow=0.0000 mean_model_t=0.5001 mean_corrupt_t=0.5001 wrong_frac=0.5002 init_acc_corrupt=0.4655 init_gold_top10=0.4942 init_gold_top100=0.4999
+step=190800 micro_steps=381600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7640 acc_all=0.5972 loss_corrupt=2.7640 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7640 loss_recon=2.7640 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4984 init_acc_corrupt=0.4675 init_gold_top10=0.4960 init_gold_top100=0.5016
+step=190900 micro_steps=381800 elapsed=63.6s lr=3.000000e-04 loss_all=2.7760 acc_all=0.5961 loss_corrupt=2.7760 acc_corrupt=0.5961 corrupt_frac=1.0000 loss=2.7760 loss_recon=2.7760 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.5001 init_acc_corrupt=0.4656 init_gold_top10=0.4944 init_gold_top100=0.5001
+step=191000 micro_steps=382000 elapsed=61.0s lr=3.000000e-04 loss_all=2.7760 acc_all=0.5955 loss_corrupt=2.7760 acc_corrupt=0.5955 corrupt_frac=1.0000 loss=2.7760 loss_recon=2.7760 loss_meanflow=0.0000 mean_model_t=0.5003 mean_corrupt_t=0.5003 wrong_frac=0.4998 init_acc_corrupt=0.4660 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=191100 micro_steps=382200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7858 acc_all=0.5943 loss_corrupt=2.7858 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7858 loss_recon=2.7858 loss_meanflow=0.0000 mean_model_t=0.4988 mean_corrupt_t=0.4988 wrong_frac=0.5009 init_acc_corrupt=0.4638 init_gold_top10=0.4935 init_gold_top100=0.4992
+step=191200 micro_steps=382400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7911 acc_all=0.5943 loss_corrupt=2.7911 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7911 loss_recon=2.7911 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5010 init_acc_corrupt=0.4649 init_gold_top10=0.4934 init_gold_top100=0.4991
+step=191300 micro_steps=382600 elapsed=63.6s lr=3.000000e-04 loss_all=2.8003 acc_all=0.5928 loss_corrupt=2.8003 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8003 loss_recon=2.8003 loss_meanflow=0.0000 mean_model_t=0.4974 mean_corrupt_t=0.4974 wrong_frac=0.5029 init_acc_corrupt=0.4623 init_gold_top10=0.4915 init_gold_top100=0.4973
+step=191400 micro_steps=382800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7727 acc_all=0.5967 loss_corrupt=2.7727 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7727 loss_recon=2.7727 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4976 init_acc_corrupt=0.4682 init_gold_top10=0.4967 init_gold_top100=0.5025
+step=191500 micro_steps=383000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8191 acc_all=0.5903 loss_corrupt=2.8191 acc_corrupt=0.5903 corrupt_frac=1.0000 loss=2.8191 loss_recon=2.8191 loss_meanflow=0.0000 mean_model_t=0.4956 mean_corrupt_t=0.4956 wrong_frac=0.5039 init_acc_corrupt=0.4605 init_gold_top10=0.4904 init_gold_top100=0.4962
+step=191600 micro_steps=383200 elapsed=63.9s lr=3.000000e-04 loss_all=2.7902 acc_all=0.5940 loss_corrupt=2.7902 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7902 loss_recon=2.7902 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5012 init_acc_corrupt=0.4637 init_gold_top10=0.4933 init_gold_top100=0.4990
+step=191700 micro_steps=383400 elapsed=63.6s lr=3.000000e-04 loss_all=2.7861 acc_all=0.5943 loss_corrupt=2.7861 acc_corrupt=0.5943 corrupt_frac=1.0000 loss=2.7861 loss_recon=2.7861 loss_meanflow=0.0000 mean_model_t=0.4985 mean_corrupt_t=0.4985 wrong_frac=0.5012 init_acc_corrupt=0.4640 init_gold_top10=0.4933 init_gold_top100=0.4989
+step=191800 micro_steps=383600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7579 acc_all=0.5984 loss_corrupt=2.7579 acc_corrupt=0.5984 corrupt_frac=1.0000 loss=2.7579 loss_recon=2.7579 loss_meanflow=0.0000 mean_model_t=0.5030 mean_corrupt_t=0.5030 wrong_frac=0.4973 init_acc_corrupt=0.4686 init_gold_top10=0.4972 init_gold_top100=0.5028
+step=191900 micro_steps=383800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7731 acc_all=0.5966 loss_corrupt=2.7731 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7731 loss_recon=2.7731 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4986 init_acc_corrupt=0.4671 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=192000 micro_steps=384000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7986 acc_all=0.5929 loss_corrupt=2.7986 acc_corrupt=0.5929 corrupt_frac=1.0000 loss=2.7986 loss_recon=2.7986 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5041 init_acc_corrupt=0.4617 init_gold_top10=0.4904 init_gold_top100=0.4962
+step=192100 micro_steps=384200 elapsed=65.1s lr=3.000000e-04 loss_all=2.7657 acc_all=0.5975 loss_corrupt=2.7657 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7657 loss_recon=2.7657 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4970 init_acc_corrupt=0.4688 init_gold_top10=0.4975 init_gold_top100=0.5031
+step=192200 micro_steps=384400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7570 acc_all=0.5980 loss_corrupt=2.7570 acc_corrupt=0.5980 corrupt_frac=1.0000 loss=2.7570 loss_recon=2.7570 loss_meanflow=0.0000 mean_model_t=0.5021 mean_corrupt_t=0.5021 wrong_frac=0.4975 init_acc_corrupt=0.4679 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=192300 micro_steps=384600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8055 acc_all=0.5918 loss_corrupt=2.8055 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8055 loss_recon=2.8055 loss_meanflow=0.0000 mean_model_t=0.4955 mean_corrupt_t=0.4955 wrong_frac=0.5046 init_acc_corrupt=0.4603 init_gold_top10=0.4898 init_gold_top100=0.4956
+step=192400 micro_steps=384800 elapsed=63.5s lr=3.000000e-04 loss_all=2.8154 acc_all=0.5906 loss_corrupt=2.8154 acc_corrupt=0.5906 corrupt_frac=1.0000 loss=2.8154 loss_recon=2.8154 loss_meanflow=0.0000 mean_model_t=0.4950 mean_corrupt_t=0.4950 wrong_frac=0.5051 init_acc_corrupt=0.4599 init_gold_top10=0.4891 init_gold_top100=0.4950
+step=192500 micro_steps=385000 elapsed=63.6s lr=3.000000e-04 loss_all=2.7795 acc_all=0.5952 loss_corrupt=2.7795 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7795 loss_recon=2.7795 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5007 init_acc_corrupt=0.4648 init_gold_top10=0.4938 init_gold_top100=0.4994
+step=192600 micro_steps=385200 elapsed=63.4s lr=3.000000e-04 loss_all=2.8192 acc_all=0.5908 loss_corrupt=2.8192 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8192 loss_recon=2.8192 loss_meanflow=0.0000 mean_model_t=0.4965 mean_corrupt_t=0.4965 wrong_frac=0.5036 init_acc_corrupt=0.4619 init_gold_top10=0.4905 init_gold_top100=0.4964
+step=192700 micro_steps=385400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7723 acc_all=0.5964 loss_corrupt=2.7723 acc_corrupt=0.5964 corrupt_frac=1.0000 loss=2.7723 loss_recon=2.7723 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4990 init_acc_corrupt=0.4666 init_gold_top10=0.4953 init_gold_top100=0.5011
+step=192800 micro_steps=385600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7897 acc_all=0.5952 loss_corrupt=2.7897 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7897 loss_recon=2.7897 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4988 init_acc_corrupt=0.4662 init_gold_top10=0.4955 init_gold_top100=0.5013
+step=192900 micro_steps=385800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7944 acc_all=0.5940 loss_corrupt=2.7944 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7944 loss_recon=2.7944 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5012 init_acc_corrupt=0.4636 init_gold_top10=0.4931 init_gold_top100=0.4990
+step=193000 micro_steps=386000 elapsed=63.3s lr=3.000000e-04 loss_all=2.7970 acc_all=0.5933 loss_corrupt=2.7970 acc_corrupt=0.5933 corrupt_frac=1.0000 loss=2.7970 loss_recon=2.7970 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5010 init_acc_corrupt=0.4634 init_gold_top10=0.4933 init_gold_top100=0.4992
+step=193100 micro_steps=386200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7661 acc_all=0.5971 loss_corrupt=2.7661 acc_corrupt=0.5971 corrupt_frac=1.0000 loss=2.7661 loss_recon=2.7661 loss_meanflow=0.0000 mean_model_t=0.5016 mean_corrupt_t=0.5016 wrong_frac=0.4987 init_acc_corrupt=0.4675 init_gold_top10=0.4957 init_gold_top100=0.5015
+step=193200 micro_steps=386400 elapsed=63.4s lr=3.000000e-04 loss_all=2.8320 acc_all=0.5886 loss_corrupt=2.8320 acc_corrupt=0.5886 corrupt_frac=1.0000 loss=2.8320 loss_recon=2.8320 loss_meanflow=0.0000 mean_model_t=0.4941 mean_corrupt_t=0.4941 wrong_frac=0.5059 init_acc_corrupt=0.4586 init_gold_top10=0.4882 init_gold_top100=0.4942
+step=193300 micro_steps=386600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7865 acc_all=0.5946 loss_corrupt=2.7865 acc_corrupt=0.5946 corrupt_frac=1.0000 loss=2.7865 loss_recon=2.7865 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5011 init_acc_corrupt=0.4640 init_gold_top10=0.4933 init_gold_top100=0.4991
+step=193400 micro_steps=386800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7346 acc_all=0.6013 loss_corrupt=2.7346 acc_corrupt=0.6013 corrupt_frac=1.0000 loss=2.7346 loss_recon=2.7346 loss_meanflow=0.0000 mean_model_t=0.5054 mean_corrupt_t=0.5054 wrong_frac=0.4952 init_acc_corrupt=0.4716 init_gold_top10=0.4995 init_gold_top100=0.5050
+step=193500 micro_steps=387000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7925 acc_all=0.5938 loss_corrupt=2.7925 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7925 loss_recon=2.7925 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5011 init_acc_corrupt=0.4641 init_gold_top10=0.4932 init_gold_top100=0.4991
+step=193600 micro_steps=387200 elapsed=63.4s lr=3.000000e-04 loss_all=2.7650 acc_all=0.5971 loss_corrupt=2.7650 acc_corrupt=0.5971 corrupt_frac=1.0000 loss=2.7650 loss_recon=2.7650 loss_meanflow=0.0000 mean_model_t=0.5014 mean_corrupt_t=0.5014 wrong_frac=0.4982 init_acc_corrupt=0.4676 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=193700 micro_steps=387400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7666 acc_all=0.5967 loss_corrupt=2.7666 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7666 loss_recon=2.7666 loss_meanflow=0.0000 mean_model_t=0.5004 mean_corrupt_t=0.5004 wrong_frac=0.4999 init_acc_corrupt=0.4665 init_gold_top10=0.4946 init_gold_top100=0.5003
+step=193800 micro_steps=387600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8039 acc_all=0.5924 loss_corrupt=2.8039 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8039 loss_recon=2.8039 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5022 init_acc_corrupt=0.4625 init_gold_top10=0.4923 init_gold_top100=0.4980
+step=193900 micro_steps=387800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7919 acc_all=0.5938 loss_corrupt=2.7919 acc_corrupt=0.5938 corrupt_frac=1.0000 loss=2.7919 loss_recon=2.7919 loss_meanflow=0.0000 mean_model_t=0.4982 mean_corrupt_t=0.4982 wrong_frac=0.5014 init_acc_corrupt=0.4642 init_gold_top10=0.4930 init_gold_top100=0.4987
+step=194000 micro_steps=388000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7699 acc_all=0.5972 loss_corrupt=2.7699 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7699 loss_recon=2.7699 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4973 init_acc_corrupt=0.4678 init_gold_top10=0.4971 init_gold_top100=0.5029
+step=194100 micro_steps=388200 elapsed=61.4s lr=3.000000e-04 loss_all=2.7966 acc_all=0.5934 loss_corrupt=2.7966 acc_corrupt=0.5934 corrupt_frac=1.0000 loss=2.7966 loss_recon=2.7966 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5019 init_acc_corrupt=0.4634 init_gold_top10=0.4926 init_gold_top100=0.4983
+step=194200 micro_steps=388400 elapsed=64.0s lr=3.000000e-04 loss_all=2.7485 acc_all=0.5995 loss_corrupt=2.7485 acc_corrupt=0.5995 corrupt_frac=1.0000 loss=2.7485 loss_recon=2.7485 loss_meanflow=0.0000 mean_model_t=0.5042 mean_corrupt_t=0.5042 wrong_frac=0.4962 init_acc_corrupt=0.4698 init_gold_top10=0.4982 init_gold_top100=0.5039
+step=194300 micro_steps=388600 elapsed=64.6s lr=3.000000e-04 loss_all=2.7615 acc_all=0.5979 loss_corrupt=2.7615 acc_corrupt=0.5979 corrupt_frac=1.0000 loss=2.7615 loss_recon=2.7615 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4971 init_acc_corrupt=0.4683 init_gold_top10=0.4972 init_gold_top100=0.5029
+step=194400 micro_steps=388800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8085 acc_all=0.5913 loss_corrupt=2.8085 acc_corrupt=0.5913 corrupt_frac=1.0000 loss=2.8085 loss_recon=2.8085 loss_meanflow=0.0000 mean_model_t=0.4964 mean_corrupt_t=0.4964 wrong_frac=0.5038 init_acc_corrupt=0.4616 init_gold_top10=0.4904 init_gold_top100=0.4963
+step=194500 micro_steps=389000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7710 acc_all=0.5968 loss_corrupt=2.7710 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7710 loss_recon=2.7710 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4989 init_acc_corrupt=0.4666 init_gold_top10=0.4957 init_gold_top100=0.5012
+step=194600 micro_steps=389200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7883 acc_all=0.5940 loss_corrupt=2.7883 acc_corrupt=0.5940 corrupt_frac=1.0000 loss=2.7883 loss_recon=2.7883 loss_meanflow=0.0000 mean_model_t=0.4993 mean_corrupt_t=0.4993 wrong_frac=0.5006 init_acc_corrupt=0.4645 init_gold_top10=0.4938 init_gold_top100=0.4996
+step=194700 micro_steps=389400 elapsed=63.3s lr=3.000000e-04 loss_all=2.7887 acc_all=0.5947 loss_corrupt=2.7887 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7887 loss_recon=2.7887 loss_meanflow=0.0000 mean_model_t=0.4998 mean_corrupt_t=0.4998 wrong_frac=0.5004 init_acc_corrupt=0.4654 init_gold_top10=0.4940 init_gold_top100=0.4997
+step=194800 micro_steps=389600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8105 acc_all=0.5911 loss_corrupt=2.8105 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8105 loss_recon=2.8105 loss_meanflow=0.0000 mean_model_t=0.4969 mean_corrupt_t=0.4969 wrong_frac=0.5037 init_acc_corrupt=0.4609 init_gold_top10=0.4906 init_gold_top100=0.4965
+step=194900 micro_steps=389800 elapsed=63.1s lr=3.000000e-04 loss_all=2.7268 acc_all=0.6026 loss_corrupt=2.7268 acc_corrupt=0.6026 corrupt_frac=1.0000 loss=2.7268 loss_recon=2.7268 loss_meanflow=0.0000 mean_model_t=0.5069 mean_corrupt_t=0.5069 wrong_frac=0.4931 init_acc_corrupt=0.4727 init_gold_top10=0.5014 init_gold_top100=0.5070
+step=195000 micro_steps=390000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7716 acc_all=0.5968 loss_corrupt=2.7716 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7716 loss_recon=2.7716 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4990 init_acc_corrupt=0.4669 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=195100 micro_steps=390200 elapsed=64.9s lr=3.000000e-04 loss_all=2.7863 acc_all=0.5944 loss_corrupt=2.7863 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7863 loss_recon=2.7863 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5003 init_acc_corrupt=0.4649 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=195200 micro_steps=390400 elapsed=63.1s lr=3.000000e-04 loss_all=2.7877 acc_all=0.5944 loss_corrupt=2.7877 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7877 loss_recon=2.7877 loss_meanflow=0.0000 mean_model_t=0.4995 mean_corrupt_t=0.4995 wrong_frac=0.5003 init_acc_corrupt=0.4649 init_gold_top10=0.4941 init_gold_top100=0.4998
+step=195300 micro_steps=390600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7848 acc_all=0.5949 loss_corrupt=2.7848 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7848 loss_recon=2.7848 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4999 wrong_frac=0.4999 init_acc_corrupt=0.4658 init_gold_top10=0.4943 init_gold_top100=0.5002
+step=195400 micro_steps=390800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8062 acc_all=0.5914 loss_corrupt=2.8062 acc_corrupt=0.5914 corrupt_frac=1.0000 loss=2.8062 loss_recon=2.8062 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5037 init_acc_corrupt=0.4617 init_gold_top10=0.4907 init_gold_top100=0.4964
+step=195500 micro_steps=391000 elapsed=63.4s lr=3.000000e-04 loss_all=2.7874 acc_all=0.5944 loss_corrupt=2.7874 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7874 loss_recon=2.7874 loss_meanflow=0.0000 mean_model_t=0.4986 mean_corrupt_t=0.4986 wrong_frac=0.5015 init_acc_corrupt=0.4642 init_gold_top10=0.4930 init_gold_top100=0.4987
+step=195600 micro_steps=391200 elapsed=63.5s lr=3.000000e-04 loss_all=2.8033 acc_all=0.5922 loss_corrupt=2.8033 acc_corrupt=0.5922 corrupt_frac=1.0000 loss=2.8033 loss_recon=2.8033 loss_meanflow=0.0000 mean_model_t=0.4967 mean_corrupt_t=0.4967 wrong_frac=0.5031 init_acc_corrupt=0.4614 init_gold_top10=0.4911 init_gold_top100=0.4971
+step=195700 micro_steps=391400 elapsed=63.6s lr=3.000000e-04 loss_all=2.8048 acc_all=0.5926 loss_corrupt=2.8048 acc_corrupt=0.5926 corrupt_frac=1.0000 loss=2.8048 loss_recon=2.8048 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5020 init_acc_corrupt=0.4633 init_gold_top10=0.4923 init_gold_top100=0.4981
+step=195800 micro_steps=391600 elapsed=63.5s lr=3.000000e-04 loss_all=2.7765 acc_all=0.5958 loss_corrupt=2.7765 acc_corrupt=0.5958 corrupt_frac=1.0000 loss=2.7765 loss_recon=2.7765 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4992 init_acc_corrupt=0.4659 init_gold_top10=0.4952 init_gold_top100=0.5009
+step=195900 micro_steps=391800 elapsed=63.4s lr=3.000000e-04 loss_all=2.8116 acc_all=0.5916 loss_corrupt=2.8116 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8116 loss_recon=2.8116 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5024 init_acc_corrupt=0.4624 init_gold_top10=0.4919 init_gold_top100=0.4977
+step=196000 micro_steps=392000 elapsed=63.2s lr=3.000000e-04 loss_all=2.7599 acc_all=0.5978 loss_corrupt=2.7599 acc_corrupt=0.5978 corrupt_frac=1.0000 loss=2.7599 loss_recon=2.7599 loss_meanflow=0.0000 mean_model_t=0.5024 mean_corrupt_t=0.5024 wrong_frac=0.4975 init_acc_corrupt=0.4683 init_gold_top10=0.4971 init_gold_top100=0.5027
+step=196100 micro_steps=392200 elapsed=71.5s lr=3.000000e-04 loss_all=2.7761 acc_all=0.5959 loss_corrupt=2.7761 acc_corrupt=0.5959 corrupt_frac=1.0000 loss=2.7761 loss_recon=2.7761 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4987 init_acc_corrupt=0.4667 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=196200 micro_steps=392400 elapsed=98.7s lr=3.000000e-04 loss_all=2.7881 acc_all=0.5939 loss_corrupt=2.7881 acc_corrupt=0.5939 corrupt_frac=1.0000 loss=2.7881 loss_recon=2.7881 loss_meanflow=0.0000 mean_model_t=0.4984 mean_corrupt_t=0.4984 wrong_frac=0.5016 init_acc_corrupt=0.4639 init_gold_top10=0.4927 init_gold_top100=0.4985
+step=196300 micro_steps=392600 elapsed=99.0s lr=3.000000e-04 loss_all=2.7581 acc_all=0.5981 loss_corrupt=2.7581 acc_corrupt=0.5981 corrupt_frac=1.0000 loss=2.7581 loss_recon=2.7581 loss_meanflow=0.0000 mean_model_t=0.5025 mean_corrupt_t=0.5025 wrong_frac=0.4977 init_acc_corrupt=0.4682 init_gold_top10=0.4969 init_gold_top100=0.5025
+step=196400 micro_steps=392800 elapsed=78.3s lr=3.000000e-04 loss_all=2.7703 acc_all=0.5967 loss_corrupt=2.7703 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7703 loss_recon=2.7703 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4983 init_acc_corrupt=0.4670 init_gold_top10=0.4963 init_gold_top100=0.5019
+step=196500 micro_steps=393000 elapsed=62.7s lr=3.000000e-04 loss_all=2.7924 acc_all=0.5932 loss_corrupt=2.7924 acc_corrupt=0.5932 corrupt_frac=1.0000 loss=2.7924 loss_recon=2.7924 loss_meanflow=0.0000 mean_model_t=0.4977 mean_corrupt_t=0.4977 wrong_frac=0.5028 init_acc_corrupt=0.4626 init_gold_top10=0.4916 init_gold_top100=0.4974
+step=196600 micro_steps=393200 elapsed=63.3s lr=3.000000e-04 loss_all=2.7682 acc_all=0.5966 loss_corrupt=2.7682 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7682 loss_recon=2.7682 loss_meanflow=0.0000 mean_model_t=0.5005 mean_corrupt_t=0.5005 wrong_frac=0.4998 init_acc_corrupt=0.4659 init_gold_top10=0.4947 init_gold_top100=0.5004
+step=196700 micro_steps=393400 elapsed=63.4s lr=3.000000e-04 loss_all=2.7635 acc_all=0.5976 loss_corrupt=2.7635 acc_corrupt=0.5976 corrupt_frac=1.0000 loss=2.7635 loss_recon=2.7635 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4985 init_acc_corrupt=0.4674 init_gold_top10=0.4962 init_gold_top100=0.5017
+step=196800 micro_steps=393600 elapsed=63.3s lr=3.000000e-04 loss_all=2.7460 acc_all=0.5997 loss_corrupt=2.7460 acc_corrupt=0.5997 corrupt_frac=1.0000 loss=2.7460 loss_recon=2.7460 loss_meanflow=0.0000 mean_model_t=0.5037 mean_corrupt_t=0.5037 wrong_frac=0.4968 init_acc_corrupt=0.4696 init_gold_top10=0.4977 init_gold_top100=0.5034
+step=196900 micro_steps=393800 elapsed=63.3s lr=3.000000e-04 loss_all=2.8010 acc_all=0.5924 loss_corrupt=2.8010 acc_corrupt=0.5924 corrupt_frac=1.0000 loss=2.8010 loss_recon=2.8010 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5033 init_acc_corrupt=0.4624 init_gold_top10=0.4912 init_gold_top100=0.4969
+step=197000 micro_steps=394000 elapsed=60.0s lr=3.000000e-04 loss_all=2.7838 acc_all=0.5953 loss_corrupt=2.7838 acc_corrupt=0.5953 corrupt_frac=1.0000 loss=2.7838 loss_recon=2.7838 loss_meanflow=0.0000 mean_model_t=0.5002 mean_corrupt_t=0.5002 wrong_frac=0.4992 init_acc_corrupt=0.4658 init_gold_top10=0.4950 init_gold_top100=0.5009
+step=197100 micro_steps=394200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7804 acc_all=0.5944 loss_corrupt=2.7804 acc_corrupt=0.5944 corrupt_frac=1.0000 loss=2.7804 loss_recon=2.7804 loss_meanflow=0.0000 mean_model_t=0.4983 mean_corrupt_t=0.4983 wrong_frac=0.5020 init_acc_corrupt=0.4638 init_gold_top10=0.4925 init_gold_top100=0.4982
+step=197200 micro_steps=394400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7864 acc_all=0.5945 loss_corrupt=2.7864 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7864 loss_recon=2.7864 loss_meanflow=0.0000 mean_model_t=0.4991 mean_corrupt_t=0.4991 wrong_frac=0.5007 init_acc_corrupt=0.4647 init_gold_top10=0.4936 init_gold_top100=0.4995
+step=197300 micro_steps=394600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7847 acc_all=0.5950 loss_corrupt=2.7847 acc_corrupt=0.5950 corrupt_frac=1.0000 loss=2.7847 loss_recon=2.7847 loss_meanflow=0.0000 mean_model_t=0.4997 mean_corrupt_t=0.4997 wrong_frac=0.5005 init_acc_corrupt=0.4649 init_gold_top10=0.4938 init_gold_top100=0.4996
+step=197400 micro_steps=394800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7324 acc_all=0.6014 loss_corrupt=2.7324 acc_corrupt=0.6014 corrupt_frac=1.0000 loss=2.7324 loss_recon=2.7324 loss_meanflow=0.0000 mean_model_t=0.5053 mean_corrupt_t=0.5053 wrong_frac=0.4948 init_acc_corrupt=0.4715 init_gold_top10=0.4999 init_gold_top100=0.5054
+step=197500 micro_steps=395000 elapsed=61.7s lr=3.000000e-04 loss_all=2.7815 acc_all=0.5954 loss_corrupt=2.7815 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7815 loss_recon=2.7815 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5002 init_acc_corrupt=0.4664 init_gold_top10=0.4941 init_gold_top100=0.4999
+step=197600 micro_steps=395200 elapsed=63.4s lr=3.000000e-04 loss_all=2.7717 acc_all=0.5966 loss_corrupt=2.7717 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7717 loss_recon=2.7717 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5011 wrong_frac=0.4992 init_acc_corrupt=0.4667 init_gold_top10=0.4953 init_gold_top100=0.5010
+step=197700 micro_steps=395400 elapsed=63.3s lr=3.000000e-04 loss_all=2.7826 acc_all=0.5954 loss_corrupt=2.7826 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7826 loss_recon=2.7826 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4994 init_acc_corrupt=0.4663 init_gold_top10=0.4949 init_gold_top100=0.5007
+step=197800 micro_steps=395600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7697 acc_all=0.5971 loss_corrupt=2.7697 acc_corrupt=0.5971 corrupt_frac=1.0000 loss=2.7697 loss_recon=2.7697 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4988 init_acc_corrupt=0.4666 init_gold_top10=0.4957 init_gold_top100=0.5013
+step=197900 micro_steps=395800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7519 acc_all=0.5997 loss_corrupt=2.7519 acc_corrupt=0.5997 corrupt_frac=1.0000 loss=2.7519 loss_recon=2.7519 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4971 init_acc_corrupt=0.4691 init_gold_top10=0.4974 init_gold_top100=0.5030
+step=198000 micro_steps=396000 elapsed=63.4s lr=3.000000e-04 loss_all=2.8110 acc_all=0.5911 loss_corrupt=2.8110 acc_corrupt=0.5911 corrupt_frac=1.0000 loss=2.8110 loss_recon=2.8110 loss_meanflow=0.0000 mean_model_t=0.4962 mean_corrupt_t=0.4962 wrong_frac=0.5041 init_acc_corrupt=0.4609 init_gold_top10=0.4904 init_gold_top100=0.4961
+step=198100 micro_steps=396200 elapsed=65.0s lr=3.000000e-04 loss_all=2.7511 acc_all=0.5988 loss_corrupt=2.7511 acc_corrupt=0.5988 corrupt_frac=1.0000 loss=2.7511 loss_recon=2.7511 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4976 init_acc_corrupt=0.4688 init_gold_top10=0.4969 init_gold_top100=0.5026
+step=198200 micro_steps=396400 elapsed=63.6s lr=3.000000e-04 loss_all=2.8151 acc_all=0.5909 loss_corrupt=2.8151 acc_corrupt=0.5909 corrupt_frac=1.0000 loss=2.8151 loss_recon=2.8151 loss_meanflow=0.0000 mean_model_t=0.4952 mean_corrupt_t=0.4952 wrong_frac=0.5047 init_acc_corrupt=0.4602 init_gold_top10=0.4895 init_gold_top100=0.4954
+step=198300 micro_steps=396600 elapsed=63.4s lr=3.000000e-04 loss_all=2.8525 acc_all=0.5860 loss_corrupt=2.8525 acc_corrupt=0.5860 corrupt_frac=1.0000 loss=2.8525 loss_recon=2.8525 loss_meanflow=0.0000 mean_model_t=0.4931 mean_corrupt_t=0.4931 wrong_frac=0.5073 init_acc_corrupt=0.4570 init_gold_top10=0.4868 init_gold_top100=0.4928
+step=198400 micro_steps=396800 elapsed=63.4s lr=3.000000e-04 loss_all=2.7654 acc_all=0.5973 loss_corrupt=2.7654 acc_corrupt=0.5973 corrupt_frac=1.0000 loss=2.7654 loss_recon=2.7654 loss_meanflow=0.0000 mean_model_t=0.5022 mean_corrupt_t=0.5022 wrong_frac=0.4982 init_acc_corrupt=0.4675 init_gold_top10=0.4964 init_gold_top100=0.5019
+step=198500 micro_steps=397000 elapsed=63.5s lr=3.000000e-04 loss_all=2.7851 acc_all=0.5948 loss_corrupt=2.7851 acc_corrupt=0.5948 corrupt_frac=1.0000 loss=2.7851 loss_recon=2.7851 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5005 init_acc_corrupt=0.4648 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=198600 micro_steps=397200 elapsed=63.5s lr=3.000000e-04 loss_all=2.7285 acc_all=0.6021 loss_corrupt=2.7285 acc_corrupt=0.6021 corrupt_frac=1.0000 loss=2.7285 loss_recon=2.7285 loss_meanflow=0.0000 mean_model_t=0.5061 mean_corrupt_t=0.5061 wrong_frac=0.4938 init_acc_corrupt=0.4724 init_gold_top10=0.5008 init_gold_top100=0.5063
+step=198700 micro_steps=397400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7625 acc_all=0.5976 loss_corrupt=2.7625 acc_corrupt=0.5976 corrupt_frac=1.0000 loss=2.7625 loss_recon=2.7625 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4981 init_acc_corrupt=0.4677 init_gold_top10=0.4963 init_gold_top100=0.5019
+step=198800 micro_steps=397600 elapsed=63.4s lr=3.000000e-04 loss_all=2.7892 acc_all=0.5945 loss_corrupt=2.7892 acc_corrupt=0.5945 corrupt_frac=1.0000 loss=2.7892 loss_recon=2.7892 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5000 wrong_frac=0.4998 init_acc_corrupt=0.4654 init_gold_top10=0.4945 init_gold_top100=0.5003
+step=198900 micro_steps=397800 elapsed=80.2s lr=3.000000e-04 loss_all=2.7883 acc_all=0.5937 loss_corrupt=2.7883 acc_corrupt=0.5937 corrupt_frac=1.0000 loss=2.7883 loss_recon=2.7883 loss_meanflow=0.0000 mean_model_t=0.4979 mean_corrupt_t=0.4979 wrong_frac=0.5018 init_acc_corrupt=0.4639 init_gold_top10=0.4926 init_gold_top100=0.4983
+step=199000 micro_steps=398000 elapsed=99.8s lr=3.000000e-04 loss_all=2.7858 acc_all=0.5947 loss_corrupt=2.7858 acc_corrupt=0.5947 corrupt_frac=1.0000 loss=2.7858 loss_recon=2.7858 loss_meanflow=0.0000 mean_model_t=0.4992 mean_corrupt_t=0.4992 wrong_frac=0.5010 init_acc_corrupt=0.4645 init_gold_top10=0.4934 init_gold_top100=0.4992
+step=199100 micro_steps=398200 elapsed=92.1s lr=3.000000e-04 loss_all=2.7676 acc_all=0.5975 loss_corrupt=2.7676 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7676 loss_recon=2.7676 loss_meanflow=0.0000 mean_model_t=0.5015 mean_corrupt_t=0.5015 wrong_frac=0.4982 init_acc_corrupt=0.4679 init_gold_top10=0.4962 init_gold_top100=0.5020
+step=199200 micro_steps=398400 elapsed=100.1s lr=3.000000e-04 loss_all=2.7863 acc_all=0.5949 loss_corrupt=2.7863 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7863 loss_recon=2.7863 loss_meanflow=0.0000 mean_model_t=0.5007 mean_corrupt_t=0.5007 wrong_frac=0.4997 init_acc_corrupt=0.4659 init_gold_top10=0.4946 init_gold_top100=0.5005
+step=199300 micro_steps=398600 elapsed=100.1s lr=3.000000e-04 loss_all=2.7413 acc_all=0.6008 loss_corrupt=2.7413 acc_corrupt=0.6008 corrupt_frac=1.0000 loss=2.7413 loss_recon=2.7413 loss_meanflow=0.0000 mean_model_t=0.5056 mean_corrupt_t=0.5056 wrong_frac=0.4943 init_acc_corrupt=0.4716 init_gold_top10=0.5003 init_gold_top100=0.5059
+step=199400 micro_steps=398800 elapsed=100.2s lr=3.000000e-04 loss_all=2.8263 acc_all=0.5895 loss_corrupt=2.8263 acc_corrupt=0.5895 corrupt_frac=1.0000 loss=2.8263 loss_recon=2.8263 loss_meanflow=0.0000 mean_model_t=0.4948 mean_corrupt_t=0.4948 wrong_frac=0.5053 init_acc_corrupt=0.4603 init_gold_top10=0.4891 init_gold_top100=0.4948
+step=199500 micro_steps=399000 elapsed=100.0s lr=3.000000e-04 loss_all=2.7479 acc_all=0.5995 loss_corrupt=2.7479 acc_corrupt=0.5995 corrupt_frac=1.0000 loss=2.7479 loss_recon=2.7479 loss_meanflow=0.0000 mean_model_t=0.5044 mean_corrupt_t=0.5044 wrong_frac=0.4961 init_acc_corrupt=0.4707 init_gold_top10=0.4984 init_gold_top100=0.5040
+step=199600 micro_steps=399200 elapsed=100.1s lr=3.000000e-04 loss_all=2.8073 acc_all=0.5928 loss_corrupt=2.8073 acc_corrupt=0.5928 corrupt_frac=1.0000 loss=2.8073 loss_recon=2.8073 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5024 init_acc_corrupt=0.4624 init_gold_top10=0.4918 init_gold_top100=0.4977
+step=199700 micro_steps=399400 elapsed=100.2s lr=3.000000e-04 loss_all=2.7800 acc_all=0.5954 loss_corrupt=2.7800 acc_corrupt=0.5954 corrupt_frac=1.0000 loss=2.7800 loss_recon=2.7800 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4990 wrong_frac=0.5005 init_acc_corrupt=0.4647 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=199800 micro_steps=399600 elapsed=100.2s lr=3.000000e-04 loss_all=2.7423 acc_all=0.5997 loss_corrupt=2.7423 acc_corrupt=0.5997 corrupt_frac=1.0000 loss=2.7423 loss_recon=2.7423 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4973 init_acc_corrupt=0.4687 init_gold_top10=0.4973 init_gold_top100=0.5029
+step=199900 micro_steps=399800 elapsed=100.1s lr=3.000000e-04 loss_all=2.7656 acc_all=0.5969 loss_corrupt=2.7656 acc_corrupt=0.5969 corrupt_frac=1.0000 loss=2.7656 loss_recon=2.7656 loss_meanflow=0.0000 mean_model_t=0.5018 mean_corrupt_t=0.5018 wrong_frac=0.4982 init_acc_corrupt=0.4675 init_gold_top10=0.4963 init_gold_top100=0.5020
+step=200000 micro_steps=400000 elapsed=100.1s lr=3.000000e-04 loss_all=2.8068 acc_all=0.5921 loss_corrupt=2.8068 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8068 loss_recon=2.8068 loss_meanflow=0.0000 mean_model_t=0.4970 mean_corrupt_t=0.4970 wrong_frac=0.5030 init_acc_corrupt=0.4619 init_gold_top10=0.4915 init_gold_top100=0.4972
+step=200100 micro_steps=400200 elapsed=103.2s lr=3.000000e-04 loss_all=2.7824 acc_all=0.5952 loss_corrupt=2.7824 acc_corrupt=0.5952 corrupt_frac=1.0000 loss=2.7824 loss_recon=2.7824 loss_meanflow=0.0000 mean_model_t=0.4996 mean_corrupt_t=0.4996 wrong_frac=0.5004 init_acc_corrupt=0.4647 init_gold_top10=0.4939 init_gold_top100=0.4997
+step=200200 micro_steps=400400 elapsed=100.1s lr=3.000000e-04 loss_all=2.8435 acc_all=0.5870 loss_corrupt=2.8435 acc_corrupt=0.5870 corrupt_frac=1.0000 loss=2.8435 loss_recon=2.8435 loss_meanflow=0.0000 mean_model_t=0.4923 mean_corrupt_t=0.4923 wrong_frac=0.5075 init_acc_corrupt=0.4573 init_gold_top10=0.4867 init_gold_top100=0.4926
+step=200300 micro_steps=400600 elapsed=72.7s lr=3.000000e-04 loss_all=2.7697 acc_all=0.5967 loss_corrupt=2.7697 acc_corrupt=0.5967 corrupt_frac=1.0000 loss=2.7697 loss_recon=2.7697 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4988 init_acc_corrupt=0.4666 init_gold_top10=0.4956 init_gold_top100=0.5013
+step=200400 micro_steps=400800 elapsed=101.9s lr=3.000000e-04 loss_all=2.8058 acc_all=0.5918 loss_corrupt=2.8058 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8058 loss_recon=2.8058 loss_meanflow=0.0000 mean_model_t=0.4973 mean_corrupt_t=0.4973 wrong_frac=0.5029 init_acc_corrupt=0.4622 init_gold_top10=0.4913 init_gold_top100=0.4972
+step=200500 micro_steps=401000 elapsed=100.1s lr=3.000000e-04 loss_all=2.7693 acc_all=0.5972 loss_corrupt=2.7693 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7693 loss_recon=2.7693 loss_meanflow=0.0000 mean_model_t=0.5019 mean_corrupt_t=0.5019 wrong_frac=0.4985 init_acc_corrupt=0.4671 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=200600 micro_steps=401200 elapsed=100.3s lr=3.000000e-04 loss_all=2.7633 acc_all=0.5978 loss_corrupt=2.7633 acc_corrupt=0.5978 corrupt_frac=1.0000 loss=2.7633 loss_recon=2.7633 loss_meanflow=0.0000 mean_model_t=0.5027 mean_corrupt_t=0.5027 wrong_frac=0.4975 init_acc_corrupt=0.4683 init_gold_top10=0.4970 init_gold_top100=0.5027
+step=200700 micro_steps=401400 elapsed=100.1s lr=3.000000e-04 loss_all=2.7634 acc_all=0.5972 loss_corrupt=2.7634 acc_corrupt=0.5972 corrupt_frac=1.0000 loss=2.7634 loss_recon=2.7634 loss_meanflow=0.0000 mean_model_t=0.5009 mean_corrupt_t=0.5009 wrong_frac=0.4989 init_acc_corrupt=0.4665 init_gold_top10=0.4954 init_gold_top100=0.5012
+step=200800 micro_steps=401600 elapsed=100.2s lr=3.000000e-04 loss_all=2.7778 acc_all=0.5956 loss_corrupt=2.7778 acc_corrupt=0.5956 corrupt_frac=1.0000 loss=2.7778 loss_recon=2.7778 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5005 init_acc_corrupt=0.4652 init_gold_top10=0.4939 init_gold_top100=0.4996
+step=200900 micro_steps=401800 elapsed=100.2s lr=3.000000e-04 loss_all=2.7633 acc_all=0.5977 loss_corrupt=2.7633 acc_corrupt=0.5977 corrupt_frac=1.0000 loss=2.7633 loss_recon=2.7633 loss_meanflow=0.0000 mean_model_t=0.5013 mean_corrupt_t=0.5013 wrong_frac=0.4991 init_acc_corrupt=0.4667 init_gold_top10=0.4954 init_gold_top100=0.5011
+step=201000 micro_steps=402000 elapsed=100.2s lr=3.000000e-04 loss_all=2.7685 acc_all=0.5966 loss_corrupt=2.7685 acc_corrupt=0.5966 corrupt_frac=1.0000 loss=2.7685 loss_recon=2.7685 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4985 init_acc_corrupt=0.4669 init_gold_top10=0.4960 init_gold_top100=0.5017
+step=201100 micro_steps=402200 elapsed=101.8s lr=3.000000e-04 loss_all=2.7695 acc_all=0.5965 loss_corrupt=2.7695 acc_corrupt=0.5965 corrupt_frac=1.0000 loss=2.7695 loss_recon=2.7695 loss_meanflow=0.0000 mean_model_t=0.5010 mean_corrupt_t=0.5010 wrong_frac=0.4994 init_acc_corrupt=0.4664 init_gold_top10=0.4952 init_gold_top100=0.5007
+step=201200 micro_steps=402400 elapsed=100.2s lr=3.000000e-04 loss_all=2.7696 acc_all=0.5971 loss_corrupt=2.7696 acc_corrupt=0.5971 corrupt_frac=1.0000 loss=2.7696 loss_recon=2.7696 loss_meanflow=0.0000 mean_model_t=0.5026 mean_corrupt_t=0.5026 wrong_frac=0.4977 init_acc_corrupt=0.4680 init_gold_top10=0.4966 init_gold_top100=0.5024
+step=201300 micro_steps=402600 elapsed=100.0s lr=3.000000e-04 loss_all=2.7551 acc_all=0.5987 loss_corrupt=2.7551 acc_corrupt=0.5987 corrupt_frac=1.0000 loss=2.7551 loss_recon=2.7551 loss_meanflow=0.0000 mean_model_t=0.5032 mean_corrupt_t=0.5032 wrong_frac=0.4973 init_acc_corrupt=0.4689 init_gold_top10=0.4973 init_gold_top100=0.5028
+step=201400 micro_steps=402800 elapsed=100.3s lr=3.000000e-04 loss_all=2.7975 acc_all=0.5927 loss_corrupt=2.7975 acc_corrupt=0.5927 corrupt_frac=1.0000 loss=2.7975 loss_recon=2.7975 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.4975 wrong_frac=0.5028 init_acc_corrupt=0.4624 init_gold_top10=0.4917 init_gold_top100=0.4974
+step=201500 micro_steps=403000 elapsed=100.3s lr=3.000000e-04 loss_all=2.8054 acc_all=0.5921 loss_corrupt=2.8054 acc_corrupt=0.5921 corrupt_frac=1.0000 loss=2.8054 loss_recon=2.8054 loss_meanflow=0.0000 mean_model_t=0.4976 mean_corrupt_t=0.4976 wrong_frac=0.5027 init_acc_corrupt=0.4624 init_gold_top10=0.4916 init_gold_top100=0.4974
+step=201600 micro_steps=403200 elapsed=100.2s lr=3.000000e-04 loss_all=2.8177 acc_all=0.5902 loss_corrupt=2.8177 acc_corrupt=0.5902 corrupt_frac=1.0000 loss=2.8177 loss_recon=2.8177 loss_meanflow=0.0000 mean_model_t=0.4953 mean_corrupt_t=0.4953 wrong_frac=0.5048 init_acc_corrupt=0.4602 init_gold_top10=0.4895 init_gold_top100=0.4954
+step=201700 micro_steps=403400 elapsed=100.3s lr=3.000000e-04 loss_all=2.7630 acc_all=0.5975 loss_corrupt=2.7630 acc_corrupt=0.5975 corrupt_frac=1.0000 loss=2.7630 loss_recon=2.7630 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4987 init_acc_corrupt=0.4672 init_gold_top10=0.4956 init_gold_top100=0.5014
+step=201800 micro_steps=403600 elapsed=100.3s lr=3.000000e-04 loss_all=2.7461 acc_all=0.5995 loss_corrupt=2.7461 acc_corrupt=0.5995 corrupt_frac=1.0000 loss=2.7461 loss_recon=2.7461 loss_meanflow=0.0000 mean_model_t=0.5034 mean_corrupt_t=0.5034 wrong_frac=0.4971 init_acc_corrupt=0.4689 init_gold_top10=0.4975 init_gold_top100=0.5030
+step=201900 micro_steps=403800 elapsed=81.6s lr=3.000000e-04 loss_all=2.7725 acc_all=0.5968 loss_corrupt=2.7725 acc_corrupt=0.5968 corrupt_frac=1.0000 loss=2.7725 loss_recon=2.7725 loss_meanflow=0.0000 mean_model_t=0.5012 mean_corrupt_t=0.5012 wrong_frac=0.4985 init_acc_corrupt=0.4668 init_gold_top10=0.4958 init_gold_top100=0.5015
+step=202000 micro_steps=404000 elapsed=61.9s lr=3.000000e-04 loss_all=2.8271 acc_all=0.5894 loss_corrupt=2.8271 acc_corrupt=0.5894 corrupt_frac=1.0000 loss=2.8271 loss_recon=2.8271 loss_meanflow=0.0000 mean_model_t=0.4948 mean_corrupt_t=0.4948 wrong_frac=0.5053 init_acc_corrupt=0.4593 init_gold_top10=0.4889 init_gold_top100=0.4948
+step=202100 micro_steps=404200 elapsed=65.0s lr=3.000000e-04 loss_all=2.8055 acc_all=0.5916 loss_corrupt=2.8055 acc_corrupt=0.5916 corrupt_frac=1.0000 loss=2.8055 loss_recon=2.8055 loss_meanflow=0.0000 mean_model_t=0.4961 mean_corrupt_t=0.4961 wrong_frac=0.5045 init_acc_corrupt=0.4610 init_gold_top10=0.4899 init_gold_top100=0.4956
+step=202200 micro_steps=404400 elapsed=63.5s lr=3.000000e-04 loss_all=2.7799 acc_all=0.5949 loss_corrupt=2.7799 acc_corrupt=0.5949 corrupt_frac=1.0000 loss=2.7799 loss_recon=2.7799 loss_meanflow=0.0000 mean_model_t=0.4981 mean_corrupt_t=0.4981 wrong_frac=0.5016 init_acc_corrupt=0.4634 init_gold_top10=0.4928 init_gold_top100=0.4985
+step=202300 micro_steps=404600 elapsed=63.3s lr=3.000000e-04 loss_all=2.8103 acc_all=0.5908 loss_corrupt=2.8103 acc_corrupt=0.5908 corrupt_frac=1.0000 loss=2.8103 loss_recon=2.8103 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4963 wrong_frac=0.5033 init_acc_corrupt=0.4613 init_gold_top10=0.4911 init_gold_top100=0.4968
+step=202400 micro_steps=404800 elapsed=63.5s lr=3.000000e-04 loss_all=2.7750 acc_all=0.5958 loss_corrupt=2.7750 acc_corrupt=0.5958 corrupt_frac=1.0000 loss=2.7750 loss_recon=2.7750 loss_meanflow=0.0000 mean_model_t=0.4994 mean_corrupt_t=0.4994 wrong_frac=0.5006 init_acc_corrupt=0.4652 init_gold_top10=0.4938 init_gold_top100=0.4995
+step=202500 micro_steps=405000 elapsed=63.5s lr=3.000000e-04 loss_all=2.8066 acc_all=0.5918 loss_corrupt=2.8066 acc_corrupt=0.5918 corrupt_frac=1.0000 loss=2.8066 loss_recon=2.8066 loss_meanflow=0.0000 mean_model_t=0.4954 mean_corrupt_t=0.4954 wrong_frac=0.5043 init_acc_corrupt=0.4609 init_gold_top10=0.4901 init_gold_top100=0.4958
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/ada.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/ada.py
new file mode 100644
index 0000000000000000000000000000000000000000..7f6fec7492861ec47272a8540227c957b97c0b58
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/ada.py
@@ -0,0 +1,144 @@
+"""
+    pygments.lexers.ada
+    ~~~~~~~~~~~~~~~~~~~
+
+    Lexers for Ada family languages.
+
+    :copyright: Copyright 2006-present by the Pygments team, see AUTHORS.
+    :license: BSD, see LICENSE for details.
+"""
+
+import re
+
+from pygments.lexer import RegexLexer, include, bygroups, words, using, this, \
+    default
+from pygments.token import Text, Comment, Operator, Keyword, Name, String, \
+    Number, Punctuation
+from pygments.lexers._ada_builtins import KEYWORD_LIST, BUILTIN_LIST
+
+__all__ = ['AdaLexer']
+
+
+class AdaLexer(RegexLexer):
+    """
+    For Ada source code.
+    """
+
+    name = 'Ada'
+    aliases = ['ada', 'ada95', 'ada2005']
+    filenames = ['*.adb', '*.ads', '*.ada']
+    mimetypes = ['text/x-ada']
+    url = 'https://www.adaic.org'
+    version_added = '1.3'
+
+    flags = re.MULTILINE | re.IGNORECASE
+
+    tokens = {
+        'root': [
+            (r'[^\S\n]+', Text),
+            (r'--.*?\n', Comment.Single),
+            (r'[^\S\n]+', Text),
+            (r'function|procedure|entry', Keyword.Declaration, 'subprogram'),
+            (r'(subtype|type)(\s+)(\w+)',
+             bygroups(Keyword.Declaration, Text, Keyword.Type), 'type_def'),
+            (r'task|protected', Keyword.Declaration),
+            (r'(subtype)(\s+)', bygroups(Keyword.Declaration, Text)),
+            (r'(end)(\s+)', bygroups(Keyword.Reserved, Text), 'end'),
+            (r'(pragma)(\s+)(\w+)', bygroups(Keyword.Reserved, Text,
+                                             Comment.Preproc)),
+            (r'(true|false|null)\b', Keyword.Constant),
+            # builtin types
+            (words(BUILTIN_LIST, suffix=r'\b'), Keyword.Type),
+            (r'(and(\s+then)?|in|mod|not|or(\s+else)|rem)\b', Operator.Word),
+            (r'generic|private', Keyword.Declaration),
+            (r'package', Keyword.Declaration, 'package'),
+            (r'array\b', Keyword.Reserved, 'array_def'),
+            (r'(with|use)(\s+)', bygroups(Keyword.Namespace, Text), 'import'),
+            (r'(\w+)(\s*)(:)(\s*)(constant)',
+             bygroups(Name.Constant, Text, Punctuation, Text,
+                      Keyword.Reserved)),
+            (r'<<\w+>>', Name.Label),
+            (r'(\w+)(\s*)(:)(\s*)(declare|begin|loop|for|while)',
+             bygroups(Name.Label, Text, Punctuation, Text, Keyword.Reserved)),
+            # keywords
+            (words(KEYWORD_LIST, prefix=r'\b', suffix=r'\b'),
+             Keyword.Reserved),
+            (r'"[^"]*"', String),
+            include('attribute'),
+            include('numbers'),
+            (r"'[^']'", String.Character),
+            (r'(\w+)(\s*|[(,])', bygroups(Name, using(this))),
+            (r"(<>|=>|:=|@|[\[\]]|[()|:;,.'])", Punctuation),
+            (r'[*<>+=/&-]', Operator),
+            (r'\n+', Text),
+        ],
+        'numbers': [
+            (r'[0-9_]+#[0-9a-f_\.]+#', Number.Hex),
+            (r'[0-9_]+\.[0-9_]*', Number.Float),
+            (r'[0-9_]+', Number.Integer),
+        ],
+        'attribute': [
+            (r"(')(\w+)", bygroups(Punctuation, Name.Attribute)),
+        ],
+        'subprogram': [
+            (r'\(', Punctuation, ('#pop', 'formal_part')),
+            (r';', Punctuation, '#pop'),
+            (r'is\b', Keyword.Reserved, '#pop'),
+            (r'"[^"]+"|\w+', Name.Function),
+            include('root'),
+        ],
+        'end': [
+            ('(if|case|record|loop|select)', Keyword.Reserved),
+            (r'"[^"]+"|[\w.]+', Name.Function),
+            (r'\s+', Text),
+            (';', Punctuation, '#pop'),
+        ],
+        'type_def': [
+            (r';', Punctuation, '#pop'),
+            (r'\(', Punctuation, 'formal_part'),
+            (r'\[', Punctuation, 'formal_part'),
+            (r'with|and|use', Keyword.Reserved),
+            (r'array\b', Keyword.Reserved, ('#pop', 'array_def')),
+            (r'record\b', Keyword.Reserved, ('record_def')),
+            (r'(null record)(;)', bygroups(Keyword.Reserved, Punctuation), '#pop'),
+            include('root'),
+        ],
+        'array_def': [
+            (r';', Punctuation, '#pop'),
+            (r'(\w+)(\s+)(range)', bygroups(Keyword.Type, Text, Keyword.Reserved)),
+            include('root'),
+        ],
+        'record_def': [
+            (r'end record', Keyword.Reserved, '#pop'),
+            include('root'),
+        ],
+        'import': [
+            # TODO: use Name.Namespace if appropriate.  This needs
+            # work to disinguish imports from aspects.
+            (r'[\w.]+', Name, '#pop'),
+            default('#pop'),
+        ],
+        'formal_part': [
+            (r'\)', Punctuation, '#pop'),
+            (r'\]', Punctuation, '#pop'),
+            (r'\w+', Name.Variable),
+            (r',|:[^=]', Punctuation),
+            (r'(in|not|null|out|access)\b', Keyword.Reserved),
+            include('root'),
+        ],
+        'package': [
+            ('body', Keyword.Declaration),
+            (r'is\s+new|renames', Keyword.Reserved),
+            ('is', Keyword.Reserved, '#pop'),
+            (';', Punctuation, '#pop'),
+            (r'\(', Punctuation, 'package_instantiation'),
+            (r'([\w.]+)', Name.Class),
+            include('root'),
+        ],
+        'package_instantiation': [
+            (r'("[^"]+"|\w+)(\s+)(=>)', bygroups(Name.Variable, Text, Punctuation)),
+            (r'[\w.\'"]', Text),
+            (r'\)', Punctuation, '#pop'),
+            include('root'),
+        ],
+    }
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/elm.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/elm.py
new file mode 100644
index 0000000000000000000000000000000000000000..201ffc340e2b56d393c1888024666bb8f824e020
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/elm.py
@@ -0,0 +1,123 @@
+"""
+    pygments.lexers.elm
+    ~~~~~~~~~~~~~~~~~~~
+
+    Lexer for the Elm programming language.
+
+    :copyright: Copyright 2006-present by the Pygments team, see AUTHORS.
+    :license: BSD, see LICENSE for details.
+"""
+
+from pygments.lexer import RegexLexer, words, include, bygroups
+from pygments.token import Comment, Keyword, Name, Number, Punctuation, \
+    String, Whitespace
+
+__all__ = ['ElmLexer']
+
+
+class ElmLexer(RegexLexer):
+    """
+    For Elm source code.
+    """
+
+    name = 'Elm'
+    url = 'https://elm-lang.org/'
+    aliases = ['elm']
+    filenames = ['*.elm']
+    mimetypes = ['text/x-elm']
+    version_added = '2.1'
+
+    validName = r'[a-z_][a-zA-Z0-9_\']*'
+
+    specialName = r'^main '
+
+    builtinOps = (
+        '~', '||', '|>', '|', '`', '^', '\\', '\'', '>>', '>=', '>', '==',
+        '=', '<~', '<|', '<=', '<<', '<-', '<', '::', ':', '/=', '//', '/',
+        '..', '.', '->', '-', '++', '+', '*', '&&', '%',
+    )
+
+    reservedWords = words((
+        'alias', 'as', 'case', 'else', 'if', 'import', 'in',
+        'let', 'module', 'of', 'port', 'then', 'type', 'where',
+    ), suffix=r'\b')
+
+    tokens = {
+        'root': [
+
+            # Comments
+            (r'\{-', Comment.Multiline, 'comment'),
+            (r'--.*', Comment.Single),
+
+            # Whitespace
+            (r'\s+', Whitespace),
+
+            # Strings
+            (r'"', String, 'doublequote'),
+
+            # Modules
+            (r'^(\s*)(module)(\s*)', bygroups(Whitespace, Keyword.Namespace,
+                Whitespace), 'imports'),
+
+            # Imports
+            (r'^(\s*)(import)(\s*)', bygroups(Whitespace, Keyword.Namespace,
+                Whitespace), 'imports'),
+
+            # Shaders
+            (r'\[glsl\|.*', Name.Entity, 'shader'),
+
+            # Keywords
+            (reservedWords, Keyword.Reserved),
+
+            # Types
+            (r'[A-Z][a-zA-Z0-9_]*', Keyword.Type),
+
+            # Main
+            (specialName, Keyword.Reserved),
+
+            # Prefix Operators
+            (words((builtinOps), prefix=r'\(', suffix=r'\)'), Name.Function),
+
+            # Infix Operators
+            (words(builtinOps), Name.Function),
+
+            # Numbers
+            include('numbers'),
+
+            # Variable Names
+            (validName, Name.Variable),
+
+            # Parens
+            (r'[,()\[\]{}]', Punctuation),
+
+        ],
+
+        'comment': [
+            (r'-(?!\})', Comment.Multiline),
+            (r'\{-', Comment.Multiline, 'comment'),
+            (r'[^-}]', Comment.Multiline),
+            (r'-\}', Comment.Multiline, '#pop'),
+        ],
+
+        'doublequote': [
+            (r'\\u[0-9a-fA-F]{4}', String.Escape),
+            (r'\\[nrfvb\\"]', String.Escape),
+            (r'[^"]', String),
+            (r'"', String, '#pop'),
+        ],
+
+        'imports': [
+            (r'\w+(\.\w+)*', Name.Class, '#pop'),
+        ],
+
+        'numbers': [
+            (r'_?\d+\.(?=\d+)', Number.Float),
+            (r'_?\d+', Number.Integer),
+        ],
+
+        'shader': [
+            (r'\|(?!\])', Name.Entity),
+            (r'\|\]', Name.Entity, '#pop'),
+            (r'(.*)(\n)', bygroups(Name.Entity, Whitespace)),
+        ],
+    }
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/modula2.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/modula2.py
new file mode 100644
index 0000000000000000000000000000000000000000..c0ebd5c14c33767f8374de499baa938b2c943425
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/modula2.py
@@ -0,0 +1,1579 @@
+"""
+    pygments.lexers.modula2
+    ~~~~~~~~~~~~~~~~~~~~~~~
+
+    Multi-Dialect Lexer for Modula-2.
+
+    :copyright: Copyright 2006-present by the Pygments team, see AUTHORS.
+    :license: BSD, see LICENSE for details.
+"""
+
+import re
+
+from pygments.lexer import RegexLexer, include
+from pygments.util import get_bool_opt, get_list_opt
+from pygments.token import Text, Comment, Operator, Keyword, Name, \
+    String, Number, Punctuation, Error
+
+__all__ = ['Modula2Lexer']
+
+
+# Multi-Dialect Modula-2 Lexer
+class Modula2Lexer(RegexLexer):
+    """
+    For Modula-2 source code.
+
+    The Modula-2 lexer supports several dialects.  By default, it operates in
+    fallback mode, recognising the *combined* literals, punctuation symbols
+    and operators of all supported dialects, and the *combined* reserved words
+    and builtins of PIM Modula-2, ISO Modula-2 and Modula-2 R10, while not
+    differentiating between library defined identifiers.
+
+    To select a specific dialect, a dialect option may be passed
+    or a dialect tag may be embedded into a source file.
+
+    Dialect Options:
+
+    `m2pim`
+        Select PIM Modula-2 dialect.
+    `m2iso`
+        Select ISO Modula-2 dialect.
+    `m2r10`
+        Select Modula-2 R10 dialect.
+    `objm2`
+        Select Objective Modula-2 dialect.
+
+    The PIM and ISO dialect options may be qualified with a language extension.
+
+    Language Extensions:
+
+    `+aglet`
+        Select Aglet Modula-2 extensions, available with m2iso.
+    `+gm2`
+        Select GNU Modula-2 extensions, available with m2pim.
+    `+p1`
+        Select p1 Modula-2 extensions, available with m2iso.
+    `+xds`
+        Select XDS Modula-2 extensions, available with m2iso.
+
+
+    Passing a Dialect Option via Unix Commandline Interface
+
+    Dialect options may be passed to the lexer using the `dialect` key.
+    Only one such option should be passed. If multiple dialect options are
+    passed, the first valid option is used, any subsequent options are ignored.
+
+    Examples:
+
+    `$ pygmentize -O full,dialect=m2iso -f html -o /path/to/output /path/to/input`
+        Use ISO dialect to render input to HTML output
+    `$ pygmentize -O full,dialect=m2iso+p1 -f rtf -o /path/to/output /path/to/input`
+        Use ISO dialect with p1 extensions to render input to RTF output
+
+
+    Embedding a Dialect Option within a source file
+
+    A dialect option may be embedded in a source file in form of a dialect
+    tag, a specially formatted comment that specifies a dialect option.
+
+    Dialect Tag EBNF::
+
+       dialectTag :
+           OpeningCommentDelim Prefix dialectOption ClosingCommentDelim ;
+
+       dialectOption :
+           'm2pim' | 'm2iso' | 'm2r10' | 'objm2' |
+           'm2iso+aglet' | 'm2pim+gm2' | 'm2iso+p1' | 'm2iso+xds' ;
+
+       Prefix : '!' ;
+
+       OpeningCommentDelim : '(*' ;
+
+       ClosingCommentDelim : '*)' ;
+
+    No whitespace is permitted between the tokens of a dialect tag.
+
+    In the event that a source file contains multiple dialect tags, the first
+    tag that contains a valid dialect option will be used and any subsequent
+    dialect tags will be ignored.  Ideally, a dialect tag should be placed
+    at the beginning of a source file.
+
+    An embedded dialect tag overrides a dialect option set via command line.
+
+    Examples:
+
+    ``(*!m2r10*) DEFINITION MODULE Foobar; ...``
+        Use Modula2 R10 dialect to render this source file.
+    ``(*!m2pim+gm2*) DEFINITION MODULE Bazbam; ...``
+        Use PIM dialect with GNU extensions to render this source file.
+
+
+    Algol Publication Mode:
+
+    In Algol publication mode, source text is rendered for publication of
+    algorithms in scientific papers and academic texts, following the format
+    of the Revised Algol-60 Language Report.  It is activated by passing
+    one of two corresponding styles as an option:
+
+    `algol`
+        render reserved words lowercase underline boldface
+        and builtins lowercase boldface italic
+    `algol_nu`
+        render reserved words lowercase boldface (no underlining)
+        and builtins lowercase boldface italic
+
+    The lexer automatically performs the required lowercase conversion when
+    this mode is activated.
+
+    Example:
+
+    ``$ pygmentize -O full,style=algol -f latex -o /path/to/output /path/to/input``
+        Render input file in Algol publication mode to LaTeX output.
+
+
+    Rendering Mode of First Class ADT Identifiers:
+
+    The rendering of standard library first class ADT identifiers is controlled
+    by option flag "treat_stdlib_adts_as_builtins".
+
+    When this option is turned on, standard library ADT identifiers are rendered
+    as builtins.  When it is turned off, they are rendered as ordinary library
+    identifiers.
+
+    `treat_stdlib_adts_as_builtins` (default: On)
+
+    The option is useful for dialects that support ADTs as first class objects
+    and provide ADTs in the standard library that would otherwise be built-in.
+
+    At present, only Modula-2 R10 supports library ADTs as first class objects
+    and therefore, no ADT identifiers are defined for any other dialects.
+
+    Example:
+
+    ``$ pygmentize -O full,dialect=m2r10,treat_stdlib_adts_as_builtins=Off ...``
+        Render standard library ADTs as ordinary library types.
+
+    .. versionchanged:: 2.1
+       Added multi-dialect support.
+    """
+    name = 'Modula-2'
+    url = 'http://www.modula2.org/'
+    aliases = ['modula2', 'm2']
+    filenames = ['*.def', '*.mod']
+    mimetypes = ['text/x-modula2']
+    version_added = '1.3'
+
+    flags = re.MULTILINE | re.DOTALL
+
+    tokens = {
+        'whitespace': [
+            (r'\n+', Text),  # blank lines
+            (r'\s+', Text),  # whitespace
+        ],
+        'dialecttags': [
+            # PIM Dialect Tag
+            (r'\(\*!m2pim\*\)', Comment.Special),
+            # ISO Dialect Tag
+            (r'\(\*!m2iso\*\)', Comment.Special),
+            # M2R10 Dialect Tag
+            (r'\(\*!m2r10\*\)', Comment.Special),
+            # ObjM2 Dialect Tag
+            (r'\(\*!objm2\*\)', Comment.Special),
+            # Aglet Extensions Dialect Tag
+            (r'\(\*!m2iso\+aglet\*\)', Comment.Special),
+            # GNU Extensions Dialect Tag
+            (r'\(\*!m2pim\+gm2\*\)', Comment.Special),
+            # p1 Extensions Dialect Tag
+            (r'\(\*!m2iso\+p1\*\)', Comment.Special),
+            # XDS Extensions Dialect Tag
+            (r'\(\*!m2iso\+xds\*\)', Comment.Special),
+        ],
+        'identifiers': [
+            (r'([a-zA-Z_$][\w$]*)', Name),
+        ],
+        'prefixed_number_literals': [
+            #
+            # Base-2, whole number
+            (r'0b[01]+(\'[01]+)*', Number.Bin),
+            #
+            # Base-16, whole number
+            (r'0[ux][0-9A-F]+(\'[0-9A-F]+)*', Number.Hex),
+        ],
+        'plain_number_literals': [
+            #
+            # Base-10, real number with exponent
+            (r'[0-9]+(\'[0-9]+)*'  # integral part
+             r'\.[0-9]+(\'[0-9]+)*'  # fractional part
+             r'[eE][+-]?[0-9]+(\'[0-9]+)*',  # exponent
+             Number.Float),
+            #
+            # Base-10, real number without exponent
+            (r'[0-9]+(\'[0-9]+)*'  # integral part
+             r'\.[0-9]+(\'[0-9]+)*',  # fractional part
+             Number.Float),
+            #
+            # Base-10, whole number
+            (r'[0-9]+(\'[0-9]+)*', Number.Integer),
+        ],
+        'suffixed_number_literals': [
+            #
+            # Base-8, whole number
+            (r'[0-7]+B', Number.Oct),
+            #
+            # Base-8, character code
+            (r'[0-7]+C', Number.Oct),
+            #
+            # Base-16, number
+            (r'[0-9A-F]+H', Number.Hex),
+        ],
+        'string_literals': [
+            (r'"(\\\\|\\[^\\]|[^"\\])*"', String.Double),
+            (r"'(\\\\|\\[^\\]|[^'\\])*'", String.Single),
+        ],
+        'digraph_operators': [
+            # Dot Product Operator
+            (r'\*\.', Operator),
+            # Array Concatenation Operator
+            (r'\+>', Operator),  # M2R10 + ObjM2
+            # Inequality Operator
+            (r'<>', Operator),  # ISO + PIM
+            # Less-Or-Equal, Subset
+            (r'<=', Operator),
+            # Greater-Or-Equal, Superset
+            (r'>=', Operator),
+            # Identity Operator
+            (r'==', Operator),  # M2R10 + ObjM2
+            # Type Conversion Operator
+            (r'::', Operator),  # M2R10 + ObjM2
+            # Assignment Symbol
+            (r':=', Operator),
+            # Postfix Increment Mutator
+            (r'\+\+', Operator),  # M2R10 + ObjM2
+            # Postfix Decrement Mutator
+            (r'--', Operator),  # M2R10 + ObjM2
+        ],
+        'unigraph_operators': [
+            # Arithmetic Operators
+            (r'[+-]', Operator),
+            (r'[*/]', Operator),
+            # ISO 80000-2 compliant Set Difference Operator
+            (r'\\', Operator),  # M2R10 + ObjM2
+            # Relational Operators
+            (r'[=#<>]', Operator),
+            # Dereferencing Operator
+            (r'\^', Operator),
+            # Dereferencing Operator Synonym
+            (r'@', Operator),  # ISO
+            # Logical AND Operator Synonym
+            (r'&', Operator),  # PIM + ISO
+            # Logical NOT Operator Synonym
+            (r'~', Operator),  # PIM + ISO
+            # Smalltalk Message Prefix
+            (r'`', Operator),  # ObjM2
+        ],
+        'digraph_punctuation': [
+            # Range Constructor
+            (r'\.\.', Punctuation),
+            # Opening Chevron Bracket
+            (r'<<', Punctuation),  # M2R10 + ISO
+            # Closing Chevron Bracket
+            (r'>>', Punctuation),  # M2R10 + ISO
+            # Blueprint Punctuation
+            (r'->', Punctuation),  # M2R10 + ISO
+            # Distinguish |# and # in M2 R10
+            (r'\|#', Punctuation),
+            # Distinguish ## and # in M2 R10
+            (r'##', Punctuation),
+            # Distinguish |* and * in M2 R10
+            (r'\|\*', Punctuation),
+        ],
+        'unigraph_punctuation': [
+            # Common Punctuation
+            (r'[()\[\]{},.:;|]', Punctuation),
+            # Case Label Separator Synonym
+            (r'!', Punctuation),  # ISO
+            # Blueprint Punctuation
+            (r'\?', Punctuation),  # M2R10 + ObjM2
+        ],
+        'comments': [
+            # Single Line Comment
+            (r'^//.*?\n', Comment.Single),  # M2R10 + ObjM2
+            # Block Comment
+            (r'\(\*([^$].*?)\*\)', Comment.Multiline),
+            # Template Block Comment
+            (r'/\*(.*?)\*/', Comment.Multiline),  # M2R10 + ObjM2
+        ],
+        'pragmas': [
+            # ISO Style Pragmas
+            (r'<\*.*?\*>', Comment.Preproc),  # ISO, M2R10 + ObjM2
+            # Pascal Style Pragmas
+            (r'\(\*\$.*?\*\)', Comment.Preproc),  # PIM
+        ],
+        'root': [
+            include('whitespace'),
+            include('dialecttags'),
+            include('pragmas'),
+            include('comments'),
+            include('identifiers'),
+            include('suffixed_number_literals'),  # PIM + ISO
+            include('prefixed_number_literals'),  # M2R10 + ObjM2
+            include('plain_number_literals'),
+            include('string_literals'),
+            include('digraph_punctuation'),
+            include('digraph_operators'),
+            include('unigraph_punctuation'),
+            include('unigraph_operators'),
+        ]
+    }
+
+#  C o m m o n   D a t a s e t s
+
+    # Common Reserved Words Dataset
+    common_reserved_words = (
+        # 37 common reserved words
+        'AND', 'ARRAY', 'BEGIN', 'BY', 'CASE', 'CONST', 'DEFINITION', 'DIV',
+        'DO', 'ELSE', 'ELSIF', 'END', 'EXIT', 'FOR', 'FROM', 'IF',
+        'IMPLEMENTATION', 'IMPORT', 'IN', 'LOOP', 'MOD', 'MODULE', 'NOT',
+        'OF', 'OR', 'POINTER', 'PROCEDURE', 'RECORD', 'REPEAT', 'RETURN',
+        'SET', 'THEN', 'TO', 'TYPE', 'UNTIL', 'VAR', 'WHILE',
+    )
+
+    # Common Builtins Dataset
+    common_builtins = (
+        # 16 common builtins
+        'ABS', 'BOOLEAN', 'CARDINAL', 'CHAR', 'CHR', 'FALSE', 'INTEGER',
+        'LONGINT', 'LONGREAL', 'MAX', 'MIN', 'NIL', 'ODD', 'ORD', 'REAL',
+        'TRUE',
+    )
+
+    # Common Pseudo-Module Builtins Dataset
+    common_pseudo_builtins = (
+        # 4 common pseudo builtins
+        'ADDRESS', 'BYTE', 'WORD', 'ADR'
+    )
+
+#  P I M   M o d u l a - 2   D a t a s e t s
+
+    # Lexemes to Mark as Error Tokens for PIM Modula-2
+    pim_lexemes_to_reject = (
+        '!', '`', '@', '$', '%', '?', '\\', '==', '++', '--', '::', '*.',
+        '+>', '->', '<<', '>>', '|#', '##',
+    )
+
+    # PIM Modula-2 Additional Reserved Words Dataset
+    pim_additional_reserved_words = (
+        # 3 additional reserved words
+        'EXPORT', 'QUALIFIED', 'WITH',
+    )
+
+    # PIM Modula-2 Additional Builtins Dataset
+    pim_additional_builtins = (
+        # 16 additional builtins
+        'BITSET', 'CAP', 'DEC', 'DISPOSE', 'EXCL', 'FLOAT', 'HALT', 'HIGH',
+        'INC', 'INCL', 'NEW', 'NIL', 'PROC', 'SIZE', 'TRUNC', 'VAL',
+    )
+
+    # PIM Modula-2 Additional Pseudo-Module Builtins Dataset
+    pim_additional_pseudo_builtins = (
+        # 5 additional pseudo builtins
+        'SYSTEM', 'PROCESS', 'TSIZE', 'NEWPROCESS', 'TRANSFER',
+    )
+
+#  I S O   M o d u l a - 2   D a t a s e t s
+
+    # Lexemes to Mark as Error Tokens for ISO Modula-2
+    iso_lexemes_to_reject = (
+        '`', '$', '%', '?', '\\', '==', '++', '--', '::', '*.', '+>', '->',
+        '<<', '>>', '|#', '##',
+    )
+
+    # ISO Modula-2 Additional Reserved Words Dataset
+    iso_additional_reserved_words = (
+        # 9 additional reserved words (ISO 10514-1)
+        'EXCEPT', 'EXPORT', 'FINALLY', 'FORWARD', 'PACKEDSET', 'QUALIFIED',
+        'REM', 'RETRY', 'WITH',
+        # 10 additional reserved words (ISO 10514-2 & ISO 10514-3)
+        'ABSTRACT', 'AS', 'CLASS', 'GUARD', 'INHERIT', 'OVERRIDE', 'READONLY',
+        'REVEAL', 'TRACED', 'UNSAFEGUARDED',
+    )
+
+    # ISO Modula-2 Additional Builtins Dataset
+    iso_additional_builtins = (
+        # 26 additional builtins (ISO 10514-1)
+        'BITSET', 'CAP', 'CMPLX', 'COMPLEX', 'DEC', 'DISPOSE', 'EXCL', 'FLOAT',
+        'HALT', 'HIGH', 'IM', 'INC', 'INCL', 'INT', 'INTERRUPTIBLE',  'LENGTH',
+        'LFLOAT', 'LONGCOMPLEX', 'NEW', 'PROC', 'PROTECTION', 'RE', 'SIZE',
+        'TRUNC', 'UNINTERRUBTIBLE', 'VAL',
+        # 5 additional builtins (ISO 10514-2 & ISO 10514-3)
+        'CREATE', 'DESTROY', 'EMPTY', 'ISMEMBER', 'SELF',
+    )
+
+    # ISO Modula-2 Additional Pseudo-Module Builtins Dataset
+    iso_additional_pseudo_builtins = (
+        # 14 additional builtins (SYSTEM)
+        'SYSTEM', 'BITSPERLOC', 'LOCSPERBYTE', 'LOCSPERWORD', 'LOC',
+        'ADDADR', 'SUBADR', 'DIFADR', 'MAKEADR', 'ADR',
+        'ROTATE', 'SHIFT', 'CAST', 'TSIZE',
+        # 13 additional builtins (COROUTINES)
+        'COROUTINES', 'ATTACH', 'COROUTINE', 'CURRENT', 'DETACH', 'HANDLER',
+        'INTERRUPTSOURCE', 'IOTRANSFER', 'IsATTACHED', 'LISTEN',
+        'NEWCOROUTINE', 'PROT', 'TRANSFER',
+        # 9 additional builtins (EXCEPTIONS)
+        'EXCEPTIONS', 'AllocateSource', 'CurrentNumber', 'ExceptionNumber',
+        'ExceptionSource', 'GetMessage', 'IsCurrentSource',
+        'IsExceptionalExecution', 'RAISE',
+        # 3 additional builtins (TERMINATION)
+        'TERMINATION', 'IsTerminating', 'HasHalted',
+        # 4 additional builtins (M2EXCEPTION)
+        'M2EXCEPTION', 'M2Exceptions', 'M2Exception', 'IsM2Exception',
+        'indexException', 'rangeException', 'caseSelectException',
+        'invalidLocation', 'functionException', 'wholeValueException',
+        'wholeDivException', 'realValueException', 'realDivException',
+        'complexValueException', 'complexDivException', 'protException',
+        'sysException', 'coException', 'exException',
+    )
+
+#  M o d u l a - 2   R 1 0   D a t a s e t s
+
+    # Lexemes to Mark as Error Tokens for Modula-2 R10
+    m2r10_lexemes_to_reject = (
+        '!', '`', '@', '$', '%', '&', '<>',
+    )
+
+    # Modula-2 R10 reserved words in addition to the common set
+    m2r10_additional_reserved_words = (
+        # 12 additional reserved words
+        'ALIAS', 'ARGLIST', 'BLUEPRINT', 'COPY', 'GENLIB', 'INDETERMINATE',
+        'NEW', 'NONE', 'OPAQUE', 'REFERENTIAL', 'RELEASE', 'RETAIN',
+        # 2 additional reserved words with symbolic assembly option
+        'ASM', 'REG',
+    )
+
+    # Modula-2 R10 builtins in addition to the common set
+    m2r10_additional_builtins = (
+        # 26 additional builtins
+        'CARDINAL', 'COUNT', 'EMPTY', 'EXISTS', 'INSERT', 'LENGTH', 'LONGCARD',
+        'OCTET', 'PTR', 'PRED', 'READ', 'READNEW', 'REMOVE', 'RETRIEVE', 'SORT',
+        'STORE', 'SUBSET', 'SUCC', 'TLIMIT', 'TMAX', 'TMIN', 'TRUE', 'TSIZE',
+        'UNICHAR', 'WRITE', 'WRITEF',
+    )
+
+    # Modula-2 R10 Additional Pseudo-Module Builtins Dataset
+    m2r10_additional_pseudo_builtins = (
+        # 13 additional builtins (TPROPERTIES)
+        'TPROPERTIES', 'PROPERTY', 'LITERAL', 'TPROPERTY', 'TLITERAL',
+        'TBUILTIN', 'TDYN', 'TREFC', 'TNIL', 'TBASE', 'TPRECISION',
+        'TMAXEXP', 'TMINEXP',
+        # 4 additional builtins (CONVERSION)
+        'CONVERSION', 'TSXFSIZE', 'SXF', 'VAL',
+        # 35 additional builtins (UNSAFE)
+        'UNSAFE', 'CAST', 'INTRINSIC', 'AVAIL', 'ADD', 'SUB', 'ADDC', 'SUBC',
+        'FETCHADD', 'FETCHSUB', 'SHL', 'SHR', 'ASHR', 'ROTL', 'ROTR', 'ROTLC',
+        'ROTRC', 'BWNOT', 'BWAND', 'BWOR', 'BWXOR', 'BWNAND', 'BWNOR',
+        'SETBIT', 'TESTBIT', 'LSBIT', 'MSBIT', 'CSBITS', 'BAIL', 'HALT',
+        'TODO', 'FFI', 'ADDR', 'VARGLIST', 'VARGC',
+        # 11 additional builtins (ATOMIC)
+        'ATOMIC', 'INTRINSIC', 'AVAIL', 'SWAP', 'CAS', 'INC', 'DEC', 'BWAND',
+        'BWNAND', 'BWOR', 'BWXOR',
+        # 7 additional builtins (COMPILER)
+        'COMPILER', 'DEBUG', 'MODNAME', 'PROCNAME', 'LINENUM', 'DEFAULT',
+        'HASH',
+        # 5 additional builtins (ASSEMBLER)
+        'ASSEMBLER', 'REGISTER', 'SETREG', 'GETREG', 'CODE',
+    )
+
+#  O b j e c t i v e   M o d u l a - 2   D a t a s e t s
+
+    # Lexemes to Mark as Error Tokens for Objective Modula-2
+    objm2_lexemes_to_reject = (
+        '!', '$', '%', '&', '<>',
+    )
+
+    # Objective Modula-2 Extensions
+    # reserved words in addition to Modula-2 R10
+    objm2_additional_reserved_words = (
+        # 16 additional reserved words
+        'BYCOPY', 'BYREF', 'CLASS', 'CONTINUE', 'CRITICAL', 'INOUT', 'METHOD',
+        'ON', 'OPTIONAL', 'OUT', 'PRIVATE', 'PROTECTED', 'PROTOCOL', 'PUBLIC',
+        'SUPER', 'TRY',
+    )
+
+    # Objective Modula-2 Extensions
+    # builtins in addition to Modula-2 R10
+    objm2_additional_builtins = (
+        # 3 additional builtins
+        'OBJECT', 'NO', 'YES',
+    )
+
+    # Objective Modula-2 Extensions
+    # pseudo-module builtins in addition to Modula-2 R10
+    objm2_additional_pseudo_builtins = (
+        # None
+    )
+
+#  A g l e t   M o d u l a - 2   D a t a s e t s
+
+    # Aglet Extensions
+    # reserved words in addition to ISO Modula-2
+    aglet_additional_reserved_words = (
+        # None
+    )
+
+    # Aglet Extensions
+    # builtins in addition to ISO Modula-2
+    aglet_additional_builtins = (
+        # 9 additional builtins
+        'BITSET8', 'BITSET16', 'BITSET32', 'CARDINAL8', 'CARDINAL16',
+        'CARDINAL32', 'INTEGER8', 'INTEGER16', 'INTEGER32',
+    )
+
+    # Aglet Modula-2 Extensions
+    # pseudo-module builtins in addition to ISO Modula-2
+    aglet_additional_pseudo_builtins = (
+        # None
+    )
+
+#  G N U   M o d u l a - 2   D a t a s e t s
+
+    # GNU Extensions
+    # reserved words in addition to PIM Modula-2
+    gm2_additional_reserved_words = (
+        # 10 additional reserved words
+        'ASM', '__ATTRIBUTE__', '__BUILTIN__', '__COLUMN__', '__DATE__',
+        '__FILE__', '__FUNCTION__', '__LINE__', '__MODULE__', 'VOLATILE',
+    )
+
+    # GNU Extensions
+    # builtins in addition to PIM Modula-2
+    gm2_additional_builtins = (
+        # 21 additional builtins
+        'BITSET8', 'BITSET16', 'BITSET32', 'CARDINAL8', 'CARDINAL16',
+        'CARDINAL32', 'CARDINAL64', 'COMPLEX32', 'COMPLEX64', 'COMPLEX96',
+        'COMPLEX128', 'INTEGER8', 'INTEGER16', 'INTEGER32', 'INTEGER64',
+        'REAL8', 'REAL16', 'REAL32', 'REAL96', 'REAL128', 'THROW',
+    )
+
+    # GNU Extensions
+    # pseudo-module builtins in addition to PIM Modula-2
+    gm2_additional_pseudo_builtins = (
+        # None
+    )
+
+#  p 1   M o d u l a - 2   D a t a s e t s
+
+    # p1 Extensions
+    # reserved words in addition to ISO Modula-2
+    p1_additional_reserved_words = (
+        # None
+    )
+
+    # p1 Extensions
+    # builtins in addition to ISO Modula-2
+    p1_additional_builtins = (
+        # None
+    )
+
+    # p1 Modula-2 Extensions
+    # pseudo-module builtins in addition to ISO Modula-2
+    p1_additional_pseudo_builtins = (
+        # 1 additional builtin
+        'BCD',
+    )
+
+#  X D S   M o d u l a - 2   D a t a s e t s
+
+    # XDS Extensions
+    # reserved words in addition to ISO Modula-2
+    xds_additional_reserved_words = (
+        # 1 additional reserved word
+        'SEQ',
+    )
+
+    # XDS Extensions
+    # builtins in addition to ISO Modula-2
+    xds_additional_builtins = (
+        # 9 additional builtins
+        'ASH', 'ASSERT', 'DIFFADR_TYPE', 'ENTIER', 'INDEX', 'LEN',
+        'LONGCARD', 'SHORTCARD', 'SHORTINT',
+    )
+
+    # XDS Modula-2 Extensions
+    # pseudo-module builtins in addition to ISO Modula-2
+    xds_additional_pseudo_builtins = (
+        # 22 additional builtins (SYSTEM)
+        'PROCESS', 'NEWPROCESS', 'BOOL8', 'BOOL16', 'BOOL32', 'CARD8',
+        'CARD16', 'CARD32', 'INT8', 'INT16', 'INT32', 'REF', 'MOVE',
+        'FILL', 'GET', 'PUT', 'CC', 'int', 'unsigned', 'size_t', 'void'
+        # 3 additional builtins (COMPILER)
+        'COMPILER', 'OPTION', 'EQUATION'
+    )
+
+#  P I M   S t a n d a r d   L i b r a r y   D a t a s e t s
+
+    # PIM Modula-2 Standard Library Modules Dataset
+    pim_stdlib_module_identifiers = (
+        'Terminal', 'FileSystem', 'InOut', 'RealInOut', 'MathLib0', 'Storage',
+    )
+
+    # PIM Modula-2 Standard Library Types Dataset
+    pim_stdlib_type_identifiers = (
+        'Flag', 'FlagSet', 'Response', 'Command', 'Lock', 'Permission',
+        'MediumType', 'File', 'FileProc', 'DirectoryProc', 'FileCommand',
+        'DirectoryCommand',
+    )
+
+    # PIM Modula-2 Standard Library Procedures Dataset
+    pim_stdlib_proc_identifiers = (
+        'Read', 'BusyRead', 'ReadAgain', 'Write', 'WriteString', 'WriteLn',
+        'Create', 'Lookup', 'Close', 'Delete', 'Rename', 'SetRead', 'SetWrite',
+        'SetModify', 'SetOpen', 'Doio', 'SetPos', 'GetPos', 'Length', 'Reset',
+        'Again', 'ReadWord', 'WriteWord', 'ReadChar', 'WriteChar',
+        'CreateMedium', 'DeleteMedium', 'AssignName', 'DeassignName',
+        'ReadMedium', 'LookupMedium', 'OpenInput', 'OpenOutput', 'CloseInput',
+        'CloseOutput', 'ReadString', 'ReadInt', 'ReadCard', 'ReadWrd',
+        'WriteInt', 'WriteCard', 'WriteOct', 'WriteHex', 'WriteWrd',
+        'ReadReal', 'WriteReal', 'WriteFixPt', 'WriteRealOct', 'sqrt', 'exp',
+        'ln', 'sin', 'cos', 'arctan', 'entier', 'ALLOCATE', 'DEALLOCATE',
+    )
+
+    # PIM Modula-2 Standard Library Variables Dataset
+    pim_stdlib_var_identifiers = (
+        'Done', 'termCH', 'in', 'out'
+    )
+
+    # PIM Modula-2 Standard Library Constants Dataset
+    pim_stdlib_const_identifiers = (
+        'EOL',
+    )
+
+#  I S O   S t a n d a r d   L i b r a r y   D a t a s e t s
+
+    # ISO Modula-2 Standard Library Modules Dataset
+    iso_stdlib_module_identifiers = (
+        # TO DO
+    )
+
+    # ISO Modula-2 Standard Library Types Dataset
+    iso_stdlib_type_identifiers = (
+        # TO DO
+    )
+
+    # ISO Modula-2 Standard Library Procedures Dataset
+    iso_stdlib_proc_identifiers = (
+        # TO DO
+    )
+
+    # ISO Modula-2 Standard Library Variables Dataset
+    iso_stdlib_var_identifiers = (
+        # TO DO
+    )
+
+    # ISO Modula-2 Standard Library Constants Dataset
+    iso_stdlib_const_identifiers = (
+        # TO DO
+    )
+
+#  M 2   R 1 0   S t a n d a r d   L i b r a r y   D a t a s e t s
+
+    # Modula-2 R10 Standard Library ADTs Dataset
+    m2r10_stdlib_adt_identifiers = (
+        'BCD', 'LONGBCD', 'BITSET', 'SHORTBITSET', 'LONGBITSET',
+        'LONGLONGBITSET', 'COMPLEX', 'LONGCOMPLEX', 'SHORTCARD', 'LONGLONGCARD',
+        'SHORTINT', 'LONGLONGINT', 'POSINT', 'SHORTPOSINT', 'LONGPOSINT',
+        'LONGLONGPOSINT', 'BITSET8', 'BITSET16', 'BITSET32', 'BITSET64',
+        'BITSET128', 'BS8', 'BS16', 'BS32', 'BS64', 'BS128', 'CARDINAL8',
+        'CARDINAL16', 'CARDINAL32', 'CARDINAL64', 'CARDINAL128', 'CARD8',
+        'CARD16', 'CARD32', 'CARD64', 'CARD128', 'INTEGER8', 'INTEGER16',
+        'INTEGER32', 'INTEGER64', 'INTEGER128', 'INT8', 'INT16', 'INT32',
+        'INT64', 'INT128', 'STRING', 'UNISTRING',
+    )
+
+    # Modula-2 R10 Standard Library Blueprints Dataset
+    m2r10_stdlib_blueprint_identifiers = (
+        'ProtoRoot', 'ProtoComputational', 'ProtoNumeric', 'ProtoScalar',
+        'ProtoNonScalar', 'ProtoCardinal', 'ProtoInteger', 'ProtoReal',
+        'ProtoComplex', 'ProtoVector', 'ProtoTuple', 'ProtoCompArray',
+        'ProtoCollection', 'ProtoStaticArray', 'ProtoStaticSet',
+        'ProtoStaticString', 'ProtoArray', 'ProtoString', 'ProtoSet',
+        'ProtoMultiSet', 'ProtoDictionary', 'ProtoMultiDict', 'ProtoExtension',
+        'ProtoIO', 'ProtoCardMath', 'ProtoIntMath', 'ProtoRealMath',
+    )
+
+    # Modula-2 R10 Standard Library Modules Dataset
+    m2r10_stdlib_module_identifiers = (
+        'ASCII', 'BooleanIO', 'CharIO', 'UnicharIO', 'OctetIO',
+        'CardinalIO', 'LongCardIO', 'IntegerIO', 'LongIntIO', 'RealIO',
+        'LongRealIO', 'BCDIO', 'LongBCDIO', 'CardMath', 'LongCardMath',
+        'IntMath', 'LongIntMath', 'RealMath', 'LongRealMath', 'BCDMath',
+        'LongBCDMath', 'FileIO', 'FileSystem', 'Storage', 'IOSupport',
+    )
+
+    # Modula-2 R10 Standard Library Types Dataset
+    m2r10_stdlib_type_identifiers = (
+        'File', 'Status',
+        # TO BE COMPLETED
+    )
+
+    # Modula-2 R10 Standard Library Procedures Dataset
+    m2r10_stdlib_proc_identifiers = (
+        'ALLOCATE', 'DEALLOCATE', 'SIZE',
+        # TO BE COMPLETED
+    )
+
+    # Modula-2 R10 Standard Library Variables Dataset
+    m2r10_stdlib_var_identifiers = (
+        'stdIn', 'stdOut', 'stdErr',
+    )
+
+    # Modula-2 R10 Standard Library Constants Dataset
+    m2r10_stdlib_const_identifiers = (
+        'pi', 'tau',
+    )
+
+#  D i a l e c t s
+
+    # Dialect modes
+    dialects = (
+        'unknown',
+        'm2pim', 'm2iso', 'm2r10', 'objm2',
+        'm2iso+aglet', 'm2pim+gm2', 'm2iso+p1', 'm2iso+xds',
+    )
+
+#   D a t a b a s e s
+
+    # Lexemes to Mark as Errors Database
+    lexemes_to_reject_db = {
+        # Lexemes to reject for unknown dialect
+        'unknown': (
+            # LEAVE THIS EMPTY
+        ),
+        # Lexemes to reject for PIM Modula-2
+        'm2pim': (
+            pim_lexemes_to_reject,
+        ),
+        # Lexemes to reject for ISO Modula-2
+        'm2iso': (
+            iso_lexemes_to_reject,
+        ),
+        # Lexemes to reject for Modula-2 R10
+        'm2r10': (
+            m2r10_lexemes_to_reject,
+        ),
+        # Lexemes to reject for Objective Modula-2
+        'objm2': (
+            objm2_lexemes_to_reject,
+        ),
+        # Lexemes to reject for Aglet Modula-2
+        'm2iso+aglet': (
+            iso_lexemes_to_reject,
+        ),
+        # Lexemes to reject for GNU Modula-2
+        'm2pim+gm2': (
+            pim_lexemes_to_reject,
+        ),
+        # Lexemes to reject for p1 Modula-2
+        'm2iso+p1': (
+            iso_lexemes_to_reject,
+        ),
+        # Lexemes to reject for XDS Modula-2
+        'm2iso+xds': (
+            iso_lexemes_to_reject,
+        ),
+    }
+
+    # Reserved Words Database
+    reserved_words_db = {
+        # Reserved words for unknown dialect
+        'unknown': (
+            common_reserved_words,
+            pim_additional_reserved_words,
+            iso_additional_reserved_words,
+            m2r10_additional_reserved_words,
+        ),
+
+        # Reserved words for PIM Modula-2
+        'm2pim': (
+            common_reserved_words,
+            pim_additional_reserved_words,
+        ),
+
+        # Reserved words for Modula-2 R10
+        'm2iso': (
+            common_reserved_words,
+            iso_additional_reserved_words,
+        ),
+
+        # Reserved words for ISO Modula-2
+        'm2r10': (
+            common_reserved_words,
+            m2r10_additional_reserved_words,
+        ),
+
+        # Reserved words for Objective Modula-2
+        'objm2': (
+            common_reserved_words,
+            m2r10_additional_reserved_words,
+            objm2_additional_reserved_words,
+        ),
+
+        # Reserved words for Aglet Modula-2 Extensions
+        'm2iso+aglet': (
+            common_reserved_words,
+            iso_additional_reserved_words,
+            aglet_additional_reserved_words,
+        ),
+
+        # Reserved words for GNU Modula-2 Extensions
+        'm2pim+gm2': (
+            common_reserved_words,
+            pim_additional_reserved_words,
+            gm2_additional_reserved_words,
+        ),
+
+        # Reserved words for p1 Modula-2 Extensions
+        'm2iso+p1': (
+            common_reserved_words,
+            iso_additional_reserved_words,
+            p1_additional_reserved_words,
+        ),
+
+        # Reserved words for XDS Modula-2 Extensions
+        'm2iso+xds': (
+            common_reserved_words,
+            iso_additional_reserved_words,
+            xds_additional_reserved_words,
+        ),
+    }
+
+    # Builtins Database
+    builtins_db = {
+        # Builtins for unknown dialect
+        'unknown': (
+            common_builtins,
+            pim_additional_builtins,
+            iso_additional_builtins,
+            m2r10_additional_builtins,
+        ),
+
+        # Builtins for PIM Modula-2
+        'm2pim': (
+            common_builtins,
+            pim_additional_builtins,
+        ),
+
+        # Builtins for ISO Modula-2
+        'm2iso': (
+            common_builtins,
+            iso_additional_builtins,
+        ),
+
+        # Builtins for ISO Modula-2
+        'm2r10': (
+            common_builtins,
+            m2r10_additional_builtins,
+        ),
+
+        # Builtins for Objective Modula-2
+        'objm2': (
+            common_builtins,
+            m2r10_additional_builtins,
+            objm2_additional_builtins,
+        ),
+
+        # Builtins for Aglet Modula-2 Extensions
+        'm2iso+aglet': (
+            common_builtins,
+            iso_additional_builtins,
+            aglet_additional_builtins,
+        ),
+
+        # Builtins for GNU Modula-2 Extensions
+        'm2pim+gm2': (
+            common_builtins,
+            pim_additional_builtins,
+            gm2_additional_builtins,
+        ),
+
+        # Builtins for p1 Modula-2 Extensions
+        'm2iso+p1': (
+            common_builtins,
+            iso_additional_builtins,
+            p1_additional_builtins,
+        ),
+
+        # Builtins for XDS Modula-2 Extensions
+        'm2iso+xds': (
+            common_builtins,
+            iso_additional_builtins,
+            xds_additional_builtins,
+        ),
+    }
+
+    # Pseudo-Module Builtins Database
+    pseudo_builtins_db = {
+        # Builtins for unknown dialect
+        'unknown': (
+            common_pseudo_builtins,
+            pim_additional_pseudo_builtins,
+            iso_additional_pseudo_builtins,
+            m2r10_additional_pseudo_builtins,
+        ),
+
+        # Builtins for PIM Modula-2
+        'm2pim': (
+            common_pseudo_builtins,
+            pim_additional_pseudo_builtins,
+        ),
+
+        # Builtins for ISO Modula-2
+        'm2iso': (
+            common_pseudo_builtins,
+            iso_additional_pseudo_builtins,
+        ),
+
+        # Builtins for ISO Modula-2
+        'm2r10': (
+            common_pseudo_builtins,
+            m2r10_additional_pseudo_builtins,
+        ),
+
+        # Builtins for Objective Modula-2
+        'objm2': (
+            common_pseudo_builtins,
+            m2r10_additional_pseudo_builtins,
+            objm2_additional_pseudo_builtins,
+        ),
+
+        # Builtins for Aglet Modula-2 Extensions
+        'm2iso+aglet': (
+            common_pseudo_builtins,
+            iso_additional_pseudo_builtins,
+            aglet_additional_pseudo_builtins,
+        ),
+
+        # Builtins for GNU Modula-2 Extensions
+        'm2pim+gm2': (
+            common_pseudo_builtins,
+            pim_additional_pseudo_builtins,
+            gm2_additional_pseudo_builtins,
+        ),
+
+        # Builtins for p1 Modula-2 Extensions
+        'm2iso+p1': (
+            common_pseudo_builtins,
+            iso_additional_pseudo_builtins,
+            p1_additional_pseudo_builtins,
+        ),
+
+        # Builtins for XDS Modula-2 Extensions
+        'm2iso+xds': (
+            common_pseudo_builtins,
+            iso_additional_pseudo_builtins,
+            xds_additional_pseudo_builtins,
+        ),
+    }
+
+    # Standard Library ADTs Database
+    stdlib_adts_db = {
+        # Empty entry for unknown dialect
+        'unknown': (
+            # LEAVE THIS EMPTY
+        ),
+        # Standard Library ADTs for PIM Modula-2
+        'm2pim': (
+            # No first class library types
+        ),
+
+        # Standard Library ADTs for ISO Modula-2
+        'm2iso': (
+            # No first class library types
+        ),
+
+        # Standard Library ADTs for Modula-2 R10
+        'm2r10': (
+            m2r10_stdlib_adt_identifiers,
+        ),
+
+        # Standard Library ADTs for Objective Modula-2
+        'objm2': (
+            m2r10_stdlib_adt_identifiers,
+        ),
+
+        # Standard Library ADTs for Aglet Modula-2
+        'm2iso+aglet': (
+            # No first class library types
+        ),
+
+        # Standard Library ADTs for GNU Modula-2
+        'm2pim+gm2': (
+            # No first class library types
+        ),
+
+        # Standard Library ADTs for p1 Modula-2
+        'm2iso+p1': (
+            # No first class library types
+        ),
+
+        # Standard Library ADTs for XDS Modula-2
+        'm2iso+xds': (
+            # No first class library types
+        ),
+    }
+
+    # Standard Library Modules Database
+    stdlib_modules_db = {
+        # Empty entry for unknown dialect
+        'unknown': (
+            # LEAVE THIS EMPTY
+        ),
+        # Standard Library Modules for PIM Modula-2
+        'm2pim': (
+            pim_stdlib_module_identifiers,
+        ),
+
+        # Standard Library Modules for ISO Modula-2
+        'm2iso': (
+            iso_stdlib_module_identifiers,
+        ),
+
+        # Standard Library Modules for Modula-2 R10
+        'm2r10': (
+            m2r10_stdlib_blueprint_identifiers,
+            m2r10_stdlib_module_identifiers,
+            m2r10_stdlib_adt_identifiers,
+        ),
+
+        # Standard Library Modules for Objective Modula-2
+        'objm2': (
+            m2r10_stdlib_blueprint_identifiers,
+            m2r10_stdlib_module_identifiers,
+        ),
+
+        # Standard Library Modules for Aglet Modula-2
+        'm2iso+aglet': (
+            iso_stdlib_module_identifiers,
+        ),
+
+        # Standard Library Modules for GNU Modula-2
+        'm2pim+gm2': (
+            pim_stdlib_module_identifiers,
+        ),
+
+        # Standard Library Modules for p1 Modula-2
+        'm2iso+p1': (
+            iso_stdlib_module_identifiers,
+        ),
+
+        # Standard Library Modules for XDS Modula-2
+        'm2iso+xds': (
+            iso_stdlib_module_identifiers,
+        ),
+    }
+
+    # Standard Library Types Database
+    stdlib_types_db = {
+        # Empty entry for unknown dialect
+        'unknown': (
+            # LEAVE THIS EMPTY
+        ),
+        # Standard Library Types for PIM Modula-2
+        'm2pim': (
+            pim_stdlib_type_identifiers,
+        ),
+
+        # Standard Library Types for ISO Modula-2
+        'm2iso': (
+            iso_stdlib_type_identifiers,
+        ),
+
+        # Standard Library Types for Modula-2 R10
+        'm2r10': (
+            m2r10_stdlib_type_identifiers,
+        ),
+
+        # Standard Library Types for Objective Modula-2
+        'objm2': (
+            m2r10_stdlib_type_identifiers,
+        ),
+
+        # Standard Library Types for Aglet Modula-2
+        'm2iso+aglet': (
+            iso_stdlib_type_identifiers,
+        ),
+
+        # Standard Library Types for GNU Modula-2
+        'm2pim+gm2': (
+            pim_stdlib_type_identifiers,
+        ),
+
+        # Standard Library Types for p1 Modula-2
+        'm2iso+p1': (
+            iso_stdlib_type_identifiers,
+        ),
+
+        # Standard Library Types for XDS Modula-2
+        'm2iso+xds': (
+            iso_stdlib_type_identifiers,
+        ),
+    }
+
+    # Standard Library Procedures Database
+    stdlib_procedures_db = {
+        # Empty entry for unknown dialect
+        'unknown': (
+            # LEAVE THIS EMPTY
+        ),
+        # Standard Library Procedures for PIM Modula-2
+        'm2pim': (
+            pim_stdlib_proc_identifiers,
+        ),
+
+        # Standard Library Procedures for ISO Modula-2
+        'm2iso': (
+            iso_stdlib_proc_identifiers,
+        ),
+
+        # Standard Library Procedures for Modula-2 R10
+        'm2r10': (
+            m2r10_stdlib_proc_identifiers,
+        ),
+
+        # Standard Library Procedures for Objective Modula-2
+        'objm2': (
+            m2r10_stdlib_proc_identifiers,
+        ),
+
+        # Standard Library Procedures for Aglet Modula-2
+        'm2iso+aglet': (
+            iso_stdlib_proc_identifiers,
+        ),
+
+        # Standard Library Procedures for GNU Modula-2
+        'm2pim+gm2': (
+            pim_stdlib_proc_identifiers,
+        ),
+
+        # Standard Library Procedures for p1 Modula-2
+        'm2iso+p1': (
+            iso_stdlib_proc_identifiers,
+        ),
+
+        # Standard Library Procedures for XDS Modula-2
+        'm2iso+xds': (
+            iso_stdlib_proc_identifiers,
+        ),
+    }
+
+    # Standard Library Variables Database
+    stdlib_variables_db = {
+        # Empty entry for unknown dialect
+        'unknown': (
+            # LEAVE THIS EMPTY
+        ),
+        # Standard Library Variables for PIM Modula-2
+        'm2pim': (
+            pim_stdlib_var_identifiers,
+        ),
+
+        # Standard Library Variables for ISO Modula-2
+        'm2iso': (
+            iso_stdlib_var_identifiers,
+        ),
+
+        # Standard Library Variables for Modula-2 R10
+        'm2r10': (
+            m2r10_stdlib_var_identifiers,
+        ),
+
+        # Standard Library Variables for Objective Modula-2
+        'objm2': (
+            m2r10_stdlib_var_identifiers,
+        ),
+
+        # Standard Library Variables for Aglet Modula-2
+        'm2iso+aglet': (
+            iso_stdlib_var_identifiers,
+        ),
+
+        # Standard Library Variables for GNU Modula-2
+        'm2pim+gm2': (
+            pim_stdlib_var_identifiers,
+        ),
+
+        # Standard Library Variables for p1 Modula-2
+        'm2iso+p1': (
+            iso_stdlib_var_identifiers,
+        ),
+
+        # Standard Library Variables for XDS Modula-2
+        'm2iso+xds': (
+            iso_stdlib_var_identifiers,
+        ),
+    }
+
+    # Standard Library Constants Database
+    stdlib_constants_db = {
+        # Empty entry for unknown dialect
+        'unknown': (
+            # LEAVE THIS EMPTY
+        ),
+        # Standard Library Constants for PIM Modula-2
+        'm2pim': (
+            pim_stdlib_const_identifiers,
+        ),
+
+        # Standard Library Constants for ISO Modula-2
+        'm2iso': (
+            iso_stdlib_const_identifiers,
+        ),
+
+        # Standard Library Constants for Modula-2 R10
+        'm2r10': (
+            m2r10_stdlib_const_identifiers,
+        ),
+
+        # Standard Library Constants for Objective Modula-2
+        'objm2': (
+            m2r10_stdlib_const_identifiers,
+        ),
+
+        # Standard Library Constants for Aglet Modula-2
+        'm2iso+aglet': (
+            iso_stdlib_const_identifiers,
+        ),
+
+        # Standard Library Constants for GNU Modula-2
+        'm2pim+gm2': (
+            pim_stdlib_const_identifiers,
+        ),
+
+        # Standard Library Constants for p1 Modula-2
+        'm2iso+p1': (
+            iso_stdlib_const_identifiers,
+        ),
+
+        # Standard Library Constants for XDS Modula-2
+        'm2iso+xds': (
+            iso_stdlib_const_identifiers,
+        ),
+    }
+
+#   M e t h o d s
+
+    # initialise a lexer instance
+    def __init__(self, **options):
+        #
+        # check dialect options
+        #
+        dialects = get_list_opt(options, 'dialect', [])
+        #
+        for dialect_option in dialects:
+            if dialect_option in self.dialects[1:-1]:
+                # valid dialect option found
+                self.set_dialect(dialect_option)
+                break
+        #
+        # Fallback Mode (DEFAULT)
+        else:
+            # no valid dialect option
+            self.set_dialect('unknown')
+        #
+        self.dialect_set_by_tag = False
+        #
+        # check style options
+        #
+        styles = get_list_opt(options, 'style', [])
+        #
+        # use lowercase mode for Algol style
+        if 'algol' in styles or 'algol_nu' in styles:
+            self.algol_publication_mode = True
+        else:
+            self.algol_publication_mode = False
+        #
+        # Check option flags
+        #
+        self.treat_stdlib_adts_as_builtins = get_bool_opt(
+            options, 'treat_stdlib_adts_as_builtins', True)
+        #
+        # call superclass initialiser
+        RegexLexer.__init__(self, **options)
+
+    # Set lexer to a specified dialect
+    def set_dialect(self, dialect_id):
+        #
+        # if __debug__:
+        #    print 'entered set_dialect with arg: ', dialect_id
+        #
+        # check dialect name against known dialects
+        if dialect_id not in self.dialects:
+            dialect = 'unknown'  # default
+        else:
+            dialect = dialect_id
+        #
+        # compose lexemes to reject set
+        lexemes_to_reject_set = set()
+        # add each list of reject lexemes for this dialect
+        for list in self.lexemes_to_reject_db[dialect]:
+            lexemes_to_reject_set.update(set(list))
+        #
+        # compose reserved words set
+        reswords_set = set()
+        # add each list of reserved words for this dialect
+        for list in self.reserved_words_db[dialect]:
+            reswords_set.update(set(list))
+        #
+        # compose builtins set
+        builtins_set = set()
+        # add each list of builtins for this dialect excluding reserved words
+        for list in self.builtins_db[dialect]:
+            builtins_set.update(set(list).difference(reswords_set))
+        #
+        # compose pseudo-builtins set
+        pseudo_builtins_set = set()
+        # add each list of builtins for this dialect excluding reserved words
+        for list in self.pseudo_builtins_db[dialect]:
+            pseudo_builtins_set.update(set(list).difference(reswords_set))
+        #
+        # compose ADTs set
+        adts_set = set()
+        # add each list of ADTs for this dialect excluding reserved words
+        for list in self.stdlib_adts_db[dialect]:
+            adts_set.update(set(list).difference(reswords_set))
+        #
+        # compose modules set
+        modules_set = set()
+        # add each list of builtins for this dialect excluding builtins
+        for list in self.stdlib_modules_db[dialect]:
+            modules_set.update(set(list).difference(builtins_set))
+        #
+        # compose types set
+        types_set = set()
+        # add each list of types for this dialect excluding builtins
+        for list in self.stdlib_types_db[dialect]:
+            types_set.update(set(list).difference(builtins_set))
+        #
+        # compose procedures set
+        procedures_set = set()
+        # add each list of procedures for this dialect excluding builtins
+        for list in self.stdlib_procedures_db[dialect]:
+            procedures_set.update(set(list).difference(builtins_set))
+        #
+        # compose variables set
+        variables_set = set()
+        # add each list of variables for this dialect excluding builtins
+        for list in self.stdlib_variables_db[dialect]:
+            variables_set.update(set(list).difference(builtins_set))
+        #
+        # compose constants set
+        constants_set = set()
+        # add each list of constants for this dialect excluding builtins
+        for list in self.stdlib_constants_db[dialect]:
+            constants_set.update(set(list).difference(builtins_set))
+        #
+        # update lexer state
+        self.dialect = dialect
+        self.lexemes_to_reject = lexemes_to_reject_set
+        self.reserved_words = reswords_set
+        self.builtins = builtins_set
+        self.pseudo_builtins = pseudo_builtins_set
+        self.adts = adts_set
+        self.modules = modules_set
+        self.types = types_set
+        self.procedures = procedures_set
+        self.variables = variables_set
+        self.constants = constants_set
+        #
+        # if __debug__:
+        #    print 'exiting set_dialect'
+        #    print ' self.dialect: ', self.dialect
+        #    print ' self.lexemes_to_reject: ', self.lexemes_to_reject
+        #    print ' self.reserved_words: ', self.reserved_words
+        #    print ' self.builtins: ', self.builtins
+        #    print ' self.pseudo_builtins: ', self.pseudo_builtins
+        #    print ' self.adts: ', self.adts
+        #    print ' self.modules: ', self.modules
+        #    print ' self.types: ', self.types
+        #    print ' self.procedures: ', self.procedures
+        #    print ' self.variables: ', self.variables
+        #    print ' self.types: ', self.types
+        #    print ' self.constants: ', self.constants
+
+    # Extracts a dialect name from a dialect tag comment string  and checks
+    # the extracted name against known dialects.  If a match is found,  the
+    # matching name is returned, otherwise dialect id 'unknown' is returned
+    def get_dialect_from_dialect_tag(self, dialect_tag):
+        #
+        # if __debug__:
+        #    print 'entered get_dialect_from_dialect_tag with arg: ', dialect_tag
+        #
+        # constants
+        left_tag_delim = '(*!'
+        right_tag_delim = '*)'
+        left_tag_delim_len = len(left_tag_delim)
+        right_tag_delim_len = len(right_tag_delim)
+        indicator_start = left_tag_delim_len
+        indicator_end = -(right_tag_delim_len)
+        #
+        # check comment string for dialect indicator
+        if len(dialect_tag) > (left_tag_delim_len + right_tag_delim_len) \
+           and dialect_tag.startswith(left_tag_delim) \
+           and dialect_tag.endswith(right_tag_delim):
+            #
+            # if __debug__:
+            #    print 'dialect tag found'
+            #
+            # extract dialect indicator
+            indicator = dialect_tag[indicator_start:indicator_end]
+            #
+            # if __debug__:
+            #    print 'extracted: ', indicator
+            #
+            # check against known dialects
+            for index in range(1, len(self.dialects)):
+                #
+                # if __debug__:
+                #    print 'dialects[', index, ']: ', self.dialects[index]
+                #
+                if indicator == self.dialects[index]:
+                    #
+                    # if __debug__:
+                    #    print 'matching dialect found'
+                    #
+                    # indicator matches known dialect
+                    return indicator
+            else:
+                # indicator does not match any dialect
+                return 'unknown'  # default
+        else:
+            # invalid indicator string
+            return 'unknown'  # default
+
+    # intercept the token stream, modify token attributes and return them
+    def get_tokens_unprocessed(self, text):
+        for index, token, value in RegexLexer.get_tokens_unprocessed(self, text):
+            #
+            # check for dialect tag if dialect has not been set by tag
+            if not self.dialect_set_by_tag and token == Comment.Special:
+                indicated_dialect = self.get_dialect_from_dialect_tag(value)
+                if indicated_dialect != 'unknown':
+                    # token is a dialect indicator
+                    # reset reserved words and builtins
+                    self.set_dialect(indicated_dialect)
+                    self.dialect_set_by_tag = True
+            #
+            # check for reserved words, predefined and stdlib identifiers
+            if token is Name:
+                if value in self.reserved_words:
+                    token = Keyword.Reserved
+                    if self.algol_publication_mode:
+                        value = value.lower()
+                #
+                elif value in self.builtins:
+                    token = Name.Builtin
+                    if self.algol_publication_mode:
+                        value = value.lower()
+                #
+                elif value in self.pseudo_builtins:
+                    token = Name.Builtin.Pseudo
+                    if self.algol_publication_mode:
+                        value = value.lower()
+                #
+                elif value in self.adts:
+                    if not self.treat_stdlib_adts_as_builtins:
+                        token = Name.Namespace
+                    else:
+                        token = Name.Builtin.Pseudo
+                        if self.algol_publication_mode:
+                            value = value.lower()
+                #
+                elif value in self.modules:
+                    token = Name.Namespace
+                #
+                elif value in self.types:
+                    token = Name.Class
+                #
+                elif value in self.procedures:
+                    token = Name.Function
+                #
+                elif value in self.variables:
+                    token = Name.Variable
+                #
+                elif value in self.constants:
+                    token = Name.Constant
+            #
+            elif token in Number:
+                #
+                # mark prefix number literals as error for PIM and ISO dialects
+                if self.dialect not in ('unknown', 'm2r10', 'objm2'):
+                    if "'" in value or value[0:2] in ('0b', '0x', '0u'):
+                        token = Error
+                #
+                elif self.dialect in ('m2r10', 'objm2'):
+                    # mark base-8 number literals as errors for M2 R10 and ObjM2
+                    if token is Number.Oct:
+                        token = Error
+                    # mark suffix base-16 literals as errors for M2 R10 and ObjM2
+                    elif token is Number.Hex and 'H' in value:
+                        token = Error
+                    # mark real numbers with E as errors for M2 R10 and ObjM2
+                    elif token is Number.Float and 'E' in value:
+                        token = Error
+            #
+            elif token in Comment:
+                #
+                # mark single line comment as error for PIM and ISO dialects
+                if token is Comment.Single:
+                    if self.dialect not in ('unknown', 'm2r10', 'objm2'):
+                        token = Error
+                #
+                if token is Comment.Preproc:
+                    # mark ISO pragma as error for PIM dialects
+                    if value.startswith('<*') and \
+                       self.dialect.startswith('m2pim'):
+                        token = Error
+                    # mark PIM pragma as comment for other dialects
+                    elif value.startswith('(*$') and \
+                            self.dialect != 'unknown' and \
+                            not self.dialect.startswith('m2pim'):
+                        token = Comment.Multiline
+            #
+            else:  # token is neither Name nor Comment
+                #
+                # mark lexemes matching the dialect's error token set as errors
+                if value in self.lexemes_to_reject:
+                    token = Error
+                #
+                # substitute lexemes when in Algol mode
+                if self.algol_publication_mode:
+                    if value == '#':
+                        value = '≠'
+                    elif value == '<=':
+                        value = '≤'
+                    elif value == '>=':
+                        value = '≥'
+                    elif value == '==':
+                        value = '≡'
+                    elif value == '*.':
+                        value = '•'
+
+            # return result
+            yield index, token, value
+
+    def analyse_text(text):
+        """It's Pascal-like, but does not use FUNCTION -- uses PROCEDURE
+        instead."""
+
+        # Check if this looks like Pascal, if not, bail out early
+        if not ('(*' in text and '*)' in text and ':=' in text):
+            return
+
+        result = 0
+        # Procedure is in Modula2
+        if re.search(r'\bPROCEDURE\b', text):
+            result += 0.6
+
+        # FUNCTION is only valid in Pascal, but not in Modula2
+        if re.search(r'\bFUNCTION\b', text):
+            result = 0.0
+
+        return result
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/smithy.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/smithy.py
new file mode 100644
index 0000000000000000000000000000000000000000..2642bb879901595f88e605364703db92de9707bd
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/smithy.py
@@ -0,0 +1,77 @@
+"""
+    pygments.lexers.smithy
+    ~~~~~~~~~~~~~~~~~~~~~~
+
+    Lexers for the Smithy IDL.
+
+    :copyright: Copyright 2006-present by the Pygments team, see AUTHORS.
+    :license: BSD, see LICENSE for details.
+"""
+
+from pygments.lexer import RegexLexer, bygroups, words
+from pygments.token import Text, Comment, Keyword, Name, String, \
+    Number, Whitespace, Punctuation
+
+__all__ = ['SmithyLexer']
+
+
+class SmithyLexer(RegexLexer):
+    """
+    For Smithy IDL
+    """
+    name = 'Smithy'
+    url = 'https://awslabs.github.io/smithy/'
+    filenames = ['*.smithy']
+    aliases = ['smithy']
+    version_added = '2.10'
+
+    unquoted = r'[A-Za-z0-9_\.#$-]+'
+    identifier = r"[A-Za-z0-9_\.#$-]+"
+
+    simple_shapes = (
+        'use', 'byte', 'short', 'integer', 'long', 'float', 'document',
+        'double', 'bigInteger', 'bigDecimal', 'boolean', 'blob', 'string',
+        'timestamp',
+    )
+
+    aggregate_shapes = (
+       'apply', 'list', 'map', 'set', 'structure', 'union', 'resource',
+       'operation', 'service', 'trait'
+    )
+
+    tokens = {
+        'root': [
+            (r'///.*$', Comment.Multiline),
+            (r'//.*$', Comment),
+            (r'@[0-9a-zA-Z\.#-]*', Name.Decorator),
+            (r'(=)', Name.Decorator),
+            (r'^(\$version)(:)(.+)',
+                bygroups(Keyword.Declaration, Name.Decorator, Name.Class)),
+            (r'^(namespace)(\s+' + identifier + r')\b',
+                bygroups(Keyword.Declaration, Name.Class)),
+            (words(simple_shapes,
+                   prefix=r'^', suffix=r'(\s+' + identifier + r')\b'),
+                bygroups(Keyword.Declaration, Name.Class)),
+            (words(aggregate_shapes,
+                   prefix=r'^', suffix=r'(\s+' + identifier + r')'),
+                bygroups(Keyword.Declaration, Name.Class)),
+            (r'^(metadata)(\s+)((?:\S+)|(?:\"[^"]+\"))(\s*)(=)',
+                bygroups(Keyword.Declaration, Whitespace, Name.Class,
+                         Whitespace, Name.Decorator)),
+            (r"(true|false|null)", Keyword.Constant),
+            (r"(-?(?:0|[1-9]\d*)(?:\.\d+)?(?:[eE][+-]?\d+)?)", Number),
+            (identifier + ":", Name.Label),
+            (identifier, Name.Variable.Class),
+            (r'\[', Text, "#push"),
+            (r'\]', Text, "#pop"),
+            (r'\(', Text, "#push"),
+            (r'\)', Text, "#pop"),
+            (r'\{', Text, "#push"),
+            (r'\}', Text, "#pop"),
+            (r'"{3}(\\\\|\n|\\")*"{3}', String.Doc),
+            (r'"(\\\\|\n|\\"|[^"])*"', String.Double),
+            (r"'(\\\\|\n|\\'|[^'])*'", String.Single),
+            (r'[:,]+', Punctuation),
+            (r'\s+', Whitespace),
+        ]
+    }
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/textedit.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/textedit.py
new file mode 100644
index 0000000000000000000000000000000000000000..e8e94ed5b0ecbe521d2aa3e753eb8ffa94f27224
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/pygments/lexers/textedit.py
@@ -0,0 +1,205 @@
+"""
+    pygments.lexers.textedit
+    ~~~~~~~~~~~~~~~~~~~~~~~~
+
+    Lexers for languages related to text processing.
+
+    :copyright: Copyright 2006-present by the Pygments team, see AUTHORS.
+    :license: BSD, see LICENSE for details.
+"""
+
+import re
+from bisect import bisect
+
+from pygments.lexer import RegexLexer, bygroups, default, include, this, using
+from pygments.lexers.python import PythonLexer
+from pygments.token import Comment, Keyword, Name, Number, Operator, \
+    Punctuation, String, Text, Whitespace
+
+__all__ = ['AwkLexer', 'SedLexer', 'VimLexer']
+
+
+class AwkLexer(RegexLexer):
+    """
+    For Awk scripts.
+    """
+
+    name = 'Awk'
+    aliases = ['awk', 'gawk', 'mawk', 'nawk']
+    filenames = ['*.awk']
+    mimetypes = ['application/x-awk']
+    url = 'https://en.wikipedia.org/wiki/AWK'
+    version_added = '1.5'
+
+    tokens = {
+        'commentsandwhitespace': [
+            (r'\s+', Text),
+            (r'#.*$', Comment.Single)
+        ],
+        'slashstartsregex': [
+            include('commentsandwhitespace'),
+            (r'/(\\.|[^[/\\\n]|\[(\\.|[^\]\\\n])*])+/'
+             r'\B', String.Regex, '#pop'),
+            (r'(?=/)', Text, ('#pop', 'badregex')),
+            default('#pop')
+        ],
+        'badregex': [
+            (r'\n', Text, '#pop')
+        ],
+        'root': [
+            (r'^(?=\s|/)', Text, 'slashstartsregex'),
+            include('commentsandwhitespace'),
+            (r'\+\+|--|\|\||&&|in\b|\$|!?~|\?|:|'
+             r'(\*\*|[-<>+*%\^/!=|])=?', Operator, 'slashstartsregex'),
+            (r'[{(\[;,]', Punctuation, 'slashstartsregex'),
+            (r'[})\].]', Punctuation),
+            (r'(break|continue|do|while|exit|for|if|else|'
+             r'return)\b', Keyword, 'slashstartsregex'),
+            (r'function\b', Keyword.Declaration, 'slashstartsregex'),
+            (r'(atan2|cos|exp|int|log|rand|sin|sqrt|srand|gensub|gsub|index|'
+             r'length|match|split|sprintf|sub|substr|tolower|toupper|close|'
+             r'fflush|getline|next|nextfile|print|printf|strftime|systime|'
+             r'delete|system)\b', Keyword.Reserved),
+            (r'(ARGC|ARGIND|ARGV|BEGIN|CONVFMT|ENVIRON|END|ERRNO|FIELDWIDTHS|'
+             r'FILENAME|FNR|FS|IGNORECASE|NF|NR|OFMT|OFS|ORFS|RLENGTH|RS|'
+             r'RSTART|RT|SUBSEP)\b', Name.Builtin),
+            (r'[$a-zA-Z_]\w*', Name.Other),
+            (r'[0-9][0-9]*\.[0-9]+([eE][0-9]+)?[fd]?', Number.Float),
+            (r'0x[0-9a-fA-F]+', Number.Hex),
+            (r'[0-9]+', Number.Integer),
+            (r'"(\\\\|\\[^\\]|[^"\\])*"', String.Double),
+            (r"'(\\\\|\\[^\\]|[^'\\])*'", String.Single),
+        ]
+    }
+
+
+class SedLexer(RegexLexer):
+    """
+    Lexer for Sed script files.
+    """
+    name = 'Sed'
+    aliases = ['sed', 'gsed', 'ssed']
+    filenames = ['*.sed', '*.[gs]sed']
+    mimetypes = ['text/x-sed']
+    url = 'https://en.wikipedia.org/wiki/Sed'
+    version_added = ''
+    flags = re.MULTILINE
+
+    # Match the contents within delimiters such as /<contents>/
+    _inside_delims = r'((?:(?:\\[^\n]|[^\\])*?\\\n)*?(?:\\.|[^\\])*?)'
+
+    tokens = {
+        'root': [
+            (r'\s+', Whitespace),
+            (r'#.*$', Comment.Single),
+            (r'[0-9]+', Number.Integer),
+            (r'\$', Operator),
+            (r'[{};,!]', Punctuation),
+            (r'[dDFgGhHlnNpPqQxz=]', Keyword),
+            (r'([berRtTvwW:])([^;\n]*)', bygroups(Keyword, String.Single)),
+            (r'([aci])((?:.*?\\\n)*(?:.*?[^\\]$))', bygroups(Keyword, String.Double)),
+            (r'([qQ])([0-9]*)', bygroups(Keyword, Number.Integer)),
+            (r'(/)' + _inside_delims + r'(/)', bygroups(Punctuation, String.Regex, Punctuation)),
+            (r'(\\(.))' + _inside_delims + r'(\2)',
+             bygroups(Punctuation, None, String.Regex, Punctuation)),
+            (r'(y)(.)' + _inside_delims + r'(\2)' + _inside_delims + r'(\2)',
+             bygroups(Keyword, Punctuation, String.Single, Punctuation, String.Single, Punctuation)),
+            (r'(s)(.)' + _inside_delims + r'(\2)' + _inside_delims + r'(\2)((?:[gpeIiMm]|[0-9])*)',
+             bygroups(Keyword, Punctuation, String.Regex, Punctuation, String.Single, Punctuation,
+                      Keyword))
+        ]
+    }
+
+class VimLexer(RegexLexer):
+    """
+    Lexer for VimL script files.
+    """
+    name = 'VimL'
+    aliases = ['vim']
+    filenames = ['*.vim', '.vimrc', '.exrc', '.gvimrc',
+                 '_vimrc', '_exrc', '_gvimrc', 'vimrc', 'gvimrc']
+    mimetypes = ['text/x-vim']
+    url = 'https://www.vim.org'
+    version_added = '0.8'
+
+    flags = re.MULTILINE
+
+    _python = r'py(?:t(?:h(?:o(?:n)?)?)?)?'
+
+    tokens = {
+        'root': [
+            (r'^([ \t:]*)(' + _python + r')([ \t]*)(<<)([ \t]*)(.*)((?:\n|.)*)(\6)',
+             bygroups(using(this), Keyword, Text, Operator, Text, Text,
+                      using(PythonLexer), Text)),
+            (r'^([ \t:]*)(' + _python + r')([ \t])(.*)',
+             bygroups(using(this), Keyword, Text, using(PythonLexer))),
+
+            (r'^\s*".*', Comment),
+
+            (r'[ \t]+', Text),
+            # TODO: regexes can have other delims
+            (r'/[^/\\\n]*(?:\\[\s\S][^/\\\n]*)*/', String.Regex),
+            (r'"[^"\\\n]*(?:\\[\s\S][^"\\\n]*)*"', String.Double),
+            (r"'[^\n']*(?:''[^\n']*)*'", String.Single),
+
+            # Who decided that doublequote was a good comment character??
+            (r'(?<=\s)"[^\-:.%#=*].*', Comment),
+            (r'-?\d+', Number),
+            (r'#[0-9a-f]{6}', Number.Hex),
+            (r'^:', Punctuation),
+            (r'[()<>+=!|,~-]', Punctuation),  # Inexact list.  Looks decent.
+            (r'\b(let|if|else|endif|elseif|fun|function|endfunction)\b',
+             Keyword),
+            (r'\b(NONE|bold|italic|underline|dark|light)\b', Name.Builtin),
+            (r'\b\w+\b', Name.Other),  # These are postprocessed below
+            (r'.', Text),
+        ],
+    }
+
+    def __init__(self, **options):
+        from pygments.lexers._vim_builtins import auto, command, option
+        self._cmd = command
+        self._opt = option
+        self._aut = auto
+
+        RegexLexer.__init__(self, **options)
+
+    def is_in(self, w, mapping):
+        r"""
+        It's kind of difficult to decide if something might be a keyword
+        in VimL because it allows you to abbreviate them.  In fact,
+        'ab[breviate]' is a good example.  :ab, :abbre, or :abbreviate are
+        valid ways to call it so rather than making really awful regexps
+        like::
+
+            \bab(?:b(?:r(?:e(?:v(?:i(?:a(?:t(?:e)?)?)?)?)?)?)?)?\b
+
+        we match `\b\w+\b` and then call is_in() on those tokens.  See
+        `scripts/get_vimkw.py` for how the lists are extracted.
+        """
+        p = bisect(mapping, (w,))
+        if p > 0:
+            if mapping[p-1][0] == w[:len(mapping[p-1][0])] and \
+               mapping[p-1][1][:len(w)] == w:
+                return True
+        if p < len(mapping):
+            return mapping[p][0] == w[:len(mapping[p][0])] and \
+                mapping[p][1][:len(w)] == w
+        return False
+
+    def get_tokens_unprocessed(self, text):
+        # TODO: builtins are only subsequent tokens on lines
+        #       and 'keywords' only happen at the beginning except
+        #       for :au ones
+        for index, token, value in \
+                RegexLexer.get_tokens_unprocessed(self, text):
+            if token is Name.Other:
+                if self.is_in(value, self._cmd):
+                    yield index, Keyword, value
+                elif self.is_in(value, self._opt) or \
+                        self.is_in(value, self._aut):
+                    yield index, Name.Builtin, value
+                else:
+                    yield index, Text, value
+            else:
+                yield index, token, value
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/__init__.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..298f4c968375e6d90728a3d4e78c5046e4591c01
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/__init__.py
@@ -0,0 +1,27 @@
+# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import TYPE_CHECKING
+
+from ...utils import _LazyModule
+from ...utils.import_utils import define_import_structure
+
+
+if TYPE_CHECKING:
+    from .configuration_deepseek_v3 import *
+    from .modeling_deepseek_v3 import *
+else:
+    import sys
+
+    _file = globals()["__file__"]
+    sys.modules[__name__] = _LazyModule(__name__, _file, define_import_structure(_file), module_spec=__spec__)
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/configuration_deepseek_v3.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/configuration_deepseek_v3.py
new file mode 100644
index 0000000000000000000000000000000000000000..4178547a5ff230c0694e07f2068c544c8c62665c
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/configuration_deepseek_v3.py
@@ -0,0 +1,115 @@
+# Copyright 2025 bzantium and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on the DeepSeekV3 implementations from the DeepSeek AI team. (https://huggingface.co/deepseek-ai/DeepSeek-V3)
+
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""DeepSeekV3 model configuration"""
+
+from huggingface_hub.dataclasses import strict
+
+from ...configuration_utils import PreTrainedConfig
+from ...modeling_rope_utils import RopeParameters
+from ...utils import auto_docstring
+
+
+@auto_docstring(checkpoint="bzantium/tiny-deepseek-v3")
+@strict
+class DeepseekV3Config(PreTrainedConfig):
+    r"""
+    n_group (`int`, *optional*, defaults to 8):
+        Number of groups for routed experts.
+    first_k_dense_replace (`int`, *optional*, defaults to 3):
+        Number of dense layers in shallow layers(embed->dense->dense->...->dense->moe->moe...->lm_head).
+                                                        \--k dense layers--/
+    rope_interleave (`bool`, *optional*, defaults to `True`):
+        Whether to interleave the rotary position embeddings.
+
+    Example:
+
+    ```python
+    >>> from transformers import DeepseekV3Model, DeepseekV3Config
+
+    >>> # Initializing a Deepseek-V3 style configuration
+    >>> configuration = DeepseekV3Config()
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "deepseek_v3"
+    keys_to_ignore_at_inference = ["past_key_values"]
+    base_model_tp_plan = {
+        "layers.*.mlp.experts.gate_up_proj": "packed_colwise",
+        "layers.*.mlp.experts.down_proj": "rowwise",
+        "layers.*.mlp.experts": "moe_tp_experts",
+        "layers.*.mlp.shared_experts.gate_proj": "colwise",
+        "layers.*.mlp.shared_experts.up_proj": "colwise",
+        "layers.*.mlp.shared_experts.down_proj": "rowwise",
+        "layers.*.mlp.gate_proj": "colwise",
+        "layers.*.mlp.up_proj": "colwise",
+        "layers.*.mlp.down_proj": "rowwise",
+    }
+    base_model_pp_plan = {
+        "embed_tokens": (["input_ids"], ["inputs_embeds"]),
+        "layers": (["hidden_states", "attention_mask"], ["hidden_states"]),
+        "norm": (["hidden_states"], ["hidden_states"]),
+    }
+    attribute_map = {
+        "num_local_experts": "n_routed_experts",
+    }
+
+    vocab_size: int = 129280
+    hidden_size: int = 7168
+    intermediate_size: int = 18432
+    moe_intermediate_size: int = 2048
+    num_hidden_layers: int = 61
+    num_attention_heads: int = 128
+    num_key_value_heads: int | None = 128
+    n_shared_experts: int = 1
+    n_routed_experts: int = 256
+    routed_scaling_factor: float = 2.5
+    kv_lora_rank: int = 512
+    q_lora_rank: int | None = 1536
+    qk_rope_head_dim: int = 64
+    v_head_dim: int | None = 128
+    qk_nope_head_dim: int = 128
+    n_group: int | None = 8
+    topk_group: int | None = 4
+    num_experts_per_tok: int | None = 8
+    first_k_dense_replace: int | None = 3
+    norm_topk_prob: bool | None = True
+    hidden_act: str = "silu"
+    max_position_embeddings: int = 4096
+    initializer_range: float = 0.02
+    rms_norm_eps: float = 1e-6
+    use_cache: bool = True
+    pad_token_id: int | None = None
+    bos_token_id: int | None = 0
+    eos_token_id: int | list[int] | None = 1
+    pretraining_tp: int | None = 1
+    tie_word_embeddings: bool = False
+    rope_parameters: RopeParameters | dict | None = None
+    rope_interleave: bool | None = True
+    attention_bias: bool = False
+    attention_dropout: float | int | None = 0.0
+
+    def __post_init__(self, **kwargs):
+        if self.num_key_value_heads is None:
+            self.num_key_value_heads = self.num_attention_heads
+
+        self.qk_head_dim = self.qk_nope_head_dim + self.qk_rope_head_dim
+        self.head_dim = self.qk_rope_head_dim
+        super().__post_init__(**kwargs)
+
+
+__all__ = ["DeepseekV3Config"]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/modeling_deepseek_v3.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/modeling_deepseek_v3.py
new file mode 100644
index 0000000000000000000000000000000000000000..fe3acd9aeddd7c050a72dfcc02760e5eea23f632
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/modeling_deepseek_v3.py
@@ -0,0 +1,722 @@
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+#           This file was automatically generated from src/transformers/models/deepseek_v3/modular_deepseek_v3.py.
+#               Do NOT edit this file manually as any edits will be overwritten by the generation of
+#             the file from the modular. If any change should be done, please apply the change to the
+#                          modular_deepseek_v3.py file directly. One of our CI enforces this.
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+import math
+from collections.abc import Callable
+from typing import Optional
+
+import torch
+import torch.nn.functional as F
+from torch import nn
+
+from ... import initialization as init
+from ...activations import ACT2FN
+from ...cache_utils import Cache, DynamicCache
+from ...generation import GenerationMixin
+from ...integrations import use_experts_implementation, use_kernel_forward_from_hub, use_kernel_func_from_hub
+from ...masking_utils import create_causal_mask
+from ...modeling_flash_attention_utils import FlashAttentionKwargs
+from ...modeling_layers import (
+    GenericForSequenceClassification,
+    GenericForTokenClassification,
+    GradientCheckpointingLayer,
+)
+from ...modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast
+from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update
+from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
+from ...processing_utils import Unpack
+from ...utils import TransformersKwargs, auto_docstring, can_return_tuple
+from ...utils.generic import is_flash_attention_requested, maybe_autocast, merge_with_config_defaults
+from ...utils.output_capturing import capture_outputs
+from .configuration_deepseek_v3 import DeepseekV3Config
+
+
+@use_kernel_forward_from_hub("RMSNorm")
+class DeepseekV3RMSNorm(nn.Module):
+    def __init__(self, hidden_size, eps: float = 1e-6) -> None:
+        """
+        DeepseekV3RMSNorm is equivalent to T5LayerNorm
+        """
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        input_dtype = hidden_states.dtype
+        hidden_states = hidden_states.to(torch.float32)
+        variance = hidden_states.pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+        return self.weight * hidden_states.to(input_dtype)
+
+    def extra_repr(self):
+        return f"{tuple(self.weight.shape)}, eps={self.variance_epsilon}"
+
+
+class DeepseekV3RotaryEmbedding(nn.Module):
+    inv_freq: torch.Tensor  # fix linting for `register_buffer`
+
+    def __init__(self, config: DeepseekV3Config, device=None):
+        super().__init__()
+        self.max_seq_len_cached = config.max_position_embeddings
+        self.original_max_seq_len = config.max_position_embeddings
+
+        self.config = config
+
+        self.rope_type = self.config.rope_parameters["rope_type"]
+        rope_init_fn: Callable = self.compute_default_rope_parameters
+        if self.rope_type != "default":
+            rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
+        inv_freq, self.attention_scaling = rope_init_fn(self.config, device)
+
+        self.register_buffer("inv_freq", inv_freq, persistent=False)
+        self.register_buffer("original_inv_freq", inv_freq.clone(), persistent=False)
+
+    @staticmethod
+    def compute_default_rope_parameters(
+        config: DeepseekV3Config | None = None,
+        device: Optional["torch.device"] = None,
+        seq_len: int | None = None,
+    ) -> tuple["torch.Tensor", float]:
+        """
+        Computes the inverse frequencies according to the original RoPE implementation
+        Args:
+            config ([`~transformers.PreTrainedConfig`]):
+                The model configuration.
+            device (`torch.device`):
+                The device to use for initialization of the inverse frequencies.
+            seq_len (`int`, *optional*):
+                The current sequence length. Unused for this type of RoPE.
+        Returns:
+            Tuple of (`torch.Tensor`, `float`), containing the inverse frequencies for the RoPE embeddings and the
+            post-processing scaling factor applied to the computed cos/sin (unused in this type of RoPE).
+        """
+        base = config.rope_parameters["rope_theta"]
+        dim = getattr(config, "head_dim", None) or config.hidden_size // config.num_attention_heads
+
+        attention_factor = 1.0  # Unused in this type of RoPE
+
+        # Compute the inverse frequencies
+        inv_freq = 1.0 / (
+            base ** (torch.arange(0, dim, 2, dtype=torch.int64).to(device=device, dtype=torch.float) / dim)
+        )
+        return inv_freq, attention_factor
+
+    @torch.no_grad()
+    @dynamic_rope_update  # power user: used with advanced RoPE types (e.g. dynamic rope)
+    def forward(self, x, position_ids):
+        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(x.device)
+        position_ids_expanded = position_ids[:, None, :].float()
+
+        device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"
+        with maybe_autocast(device_type=device_type, enabled=False):  # Force float32
+            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
+            emb = torch.cat((freqs, freqs), dim=-1)
+            cos = emb.cos() * self.attention_scaling
+            sin = emb.sin() * self.attention_scaling
+
+        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)
+
+
+class DeepseekV3MLP(nn.Module):
+    def __init__(self, config, intermediate_size=None):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.intermediate_size = config.intermediate_size if intermediate_size is None else intermediate_size
+        self.gate_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
+        self.up_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
+        self.down_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=False)
+        self.act_fn = ACT2FN[config.hidden_act]
+
+    def forward(self, x):
+        down_proj = self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
+        return down_proj
+
+
+class DeepseekV3TopkRouter(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.n_routed_experts = config.n_routed_experts
+
+        self.weight = nn.Parameter(torch.empty((self.n_routed_experts, config.hidden_size)))
+        self.register_buffer("e_score_correction_bias", torch.zeros(self.n_routed_experts))
+
+    def forward(self, hidden_states):
+        hidden_states = hidden_states.view(-1, self.config.hidden_size)
+        router_logits = F.linear(hidden_states.type(torch.float32), self.weight.type(torch.float32))
+        return router_logits
+
+
+@use_experts_implementation
+class DeepseekV3NaiveMoe(nn.Module):
+    """Collection of expert weights stored as 3D tensors."""
+
+    def __init__(self, config):
+        super().__init__()
+        self.num_experts = config.num_local_experts
+        self.hidden_dim = config.hidden_size
+        self.intermediate_dim = config.moe_intermediate_size
+        self.gate_up_proj = nn.Parameter(torch.empty(self.num_experts, 2 * self.intermediate_dim, self.hidden_dim))
+        self.down_proj = nn.Parameter(torch.empty(self.num_experts, self.hidden_dim, self.intermediate_dim))
+        self.act_fn = ACT2FN[config.hidden_act]
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        top_k_index: torch.Tensor,
+        top_k_weights: torch.Tensor,
+    ) -> torch.Tensor:
+        final_hidden_states = torch.zeros_like(hidden_states)
+        with torch.no_grad():
+            expert_mask = torch.nn.functional.one_hot(top_k_index, num_classes=self.num_experts)
+            expert_mask = expert_mask.permute(2, 1, 0)
+            expert_hit = torch.greater(expert_mask.sum(dim=(-1, -2)), 0).nonzero()
+
+        for expert_idx in expert_hit:
+            expert_idx = expert_idx[0]
+            if expert_idx == self.num_experts:
+                continue
+            top_k_pos, token_idx = torch.where(expert_mask[expert_idx])
+            current_state = hidden_states[token_idx]
+            gate, up = nn.functional.linear(current_state, self.gate_up_proj[expert_idx]).chunk(2, dim=-1)
+            current_hidden_states = self.act_fn(gate) * up
+            current_hidden_states = nn.functional.linear(current_hidden_states, self.down_proj[expert_idx])
+            current_hidden_states = current_hidden_states * top_k_weights[token_idx, top_k_pos, None]
+            final_hidden_states.index_add_(0, token_idx, current_hidden_states.to(final_hidden_states.dtype))
+
+        return final_hidden_states
+
+
+class DeepseekV3MoE(nn.Module):
+    """
+    A mixed expert module containing shared experts.
+    """
+
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.experts = DeepseekV3NaiveMoe(config)
+        self.gate = DeepseekV3TopkRouter(config)
+        self.shared_experts = DeepseekV3MLP(
+            config=config, intermediate_size=config.moe_intermediate_size * config.n_shared_experts
+        )
+        self.n_routed_experts = config.n_routed_experts
+        self.n_group = config.n_group
+        self.topk_group = config.topk_group
+        self.norm_topk_prob = config.norm_topk_prob
+        self.routed_scaling_factor = config.routed_scaling_factor
+        self.top_k = config.num_experts_per_tok
+
+    def route_tokens_to_experts(self, router_logits):
+        router_logits = router_logits.sigmoid()
+        router_logits_for_choice = router_logits + self.gate.e_score_correction_bias
+        group_scores = (
+            router_logits_for_choice.view(-1, self.n_group, self.n_routed_experts // self.n_group)
+            .topk(2, dim=-1)[0]
+            .sum(dim=-1)
+        )
+        group_idx = torch.topk(group_scores, k=self.topk_group, dim=-1, sorted=False)[1]
+        group_mask = torch.zeros_like(group_scores)
+        group_mask.scatter_(1, group_idx, 1)
+        score_mask = (
+            group_mask.unsqueeze(-1)
+            .expand(-1, self.n_group, self.n_routed_experts // self.n_group)
+            .reshape(-1, self.n_routed_experts)
+        )
+        scores_for_choice = router_logits_for_choice.masked_fill(~score_mask.bool(), float("-inf"))
+        topk_indices = torch.topk(scores_for_choice, k=self.top_k, dim=-1, sorted=False)[1]
+        topk_weights = router_logits.gather(1, topk_indices)
+        if self.norm_topk_prob:
+            denominator = topk_weights.sum(dim=-1, keepdim=True) + 1e-20
+            topk_weights /= denominator
+        topk_weights = topk_weights * self.routed_scaling_factor
+        return topk_indices, topk_weights
+
+    def forward(self, hidden_states):
+        residuals = hidden_states
+        orig_shape = hidden_states.shape
+        router_logits = self.gate(hidden_states)
+        topk_indices, topk_weights = self.route_tokens_to_experts(router_logits)
+        hidden_states = hidden_states.view(-1, hidden_states.shape[-1])
+        hidden_states = self.experts(hidden_states, topk_indices, topk_weights).view(*orig_shape)
+        hidden_states = hidden_states + self.shared_experts(residuals)
+        return hidden_states
+
+
+def rotate_half(x):
+    """Rotates half the hidden dims of the input."""
+    x1 = x[..., : x.shape[-1] // 2]
+    x2 = x[..., x.shape[-1] // 2 :]
+    return torch.cat((-x2, x1), dim=-1)
+
+
+@use_kernel_func_from_hub("rotary_pos_emb")
+def apply_rotary_pos_emb(q, k, cos, sin, unsqueeze_dim=1):
+    """Applies Rotary Position Embedding to the query and key tensors.
+
+    Args:
+        q (`torch.Tensor`): The query tensor.
+        k (`torch.Tensor`): The key tensor.
+        cos (`torch.Tensor`): The cosine part of the rotary embedding.
+        sin (`torch.Tensor`): The sine part of the rotary embedding.
+        unsqueeze_dim (`int`, *optional*, defaults to 1):
+            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
+            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
+            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
+            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
+            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
+            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
+    Returns:
+        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
+    """
+    cos = cos.unsqueeze(unsqueeze_dim)
+    sin = sin.unsqueeze(unsqueeze_dim)
+    q_embed = (q * cos) + (rotate_half(q) * sin)
+    k_embed = (k * cos) + (rotate_half(k) * sin)
+    return q_embed, k_embed
+
+
+def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
+    """
+    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
+    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
+    """
+    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
+    if n_rep == 1:
+        return hidden_states
+    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
+    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)
+
+
+def eager_attention_forward(
+    module: nn.Module,
+    query: torch.Tensor,
+    key: torch.Tensor,
+    value: torch.Tensor,
+    attention_mask: torch.Tensor | None,
+    scaling: float,
+    dropout: float = 0.0,
+    **kwargs: Unpack[TransformersKwargs],
+):
+    key_states = repeat_kv(key, module.num_key_value_groups)
+    value_states = repeat_kv(value, module.num_key_value_groups)
+
+    attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling
+    if attention_mask is not None:
+        attn_weights = attn_weights + attention_mask
+
+    attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query.dtype)
+    attn_weights = nn.functional.dropout(attn_weights, p=dropout, training=module.training)
+    attn_output = torch.matmul(attn_weights, value_states)
+    attn_output = attn_output.transpose(1, 2).contiguous()
+
+    return attn_output, attn_weights
+
+
+def apply_rotary_pos_emb_interleave(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
+    r"""
+    TODO let's just use the original freqcis computation to not have the view
+    transpose + reshape! This is not optimized!
+    Applies Rotary Position Embedding to the query and key tensors.
+
+    Args:
+        q (`torch.Tensor`): The query tensor.
+        k (`torch.Tensor`): The key tensor.
+        cos (`torch.Tensor`): The cosine part of the rotary embedding.
+        sin (`torch.Tensor`): The sine part of the rotary embedding.
+        position_ids (`torch.Tensor`):
+            The position indices of the tokens corresponding to the query and key tensors. For example, this can be
+            used to pass offsetted position ids when working with a KV-cache.
+        unsqueeze_dim (`int`, *optional*, defaults to 1):
+            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
+            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
+            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
+            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
+            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
+            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
+    Returns:
+        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
+    """
+    cos = cos.unsqueeze(unsqueeze_dim)
+    sin = sin.unsqueeze(unsqueeze_dim)
+
+    b, h, s, d = q.shape
+    q = q.view(b, h, s, d // 2, 2).transpose(4, 3).reshape(b, h, s, d)
+
+    b, h, s, d = k.shape
+    k = k.view(b, h, s, d // 2, 2).transpose(4, 3).reshape(b, h, s, d)
+
+    q_embed = (q * cos) + (rotate_half(q) * sin)
+    k_embed = (k * cos) + (rotate_half(k) * sin)
+    return q_embed, k_embed
+
+
+def yarn_get_mscale(scale=1, mscale=1):
+    if scale <= 1:
+        return 1.0
+    return 0.1 * mscale * math.log(scale) + 1.0
+
+
+class DeepseekV3Attention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: DeepseekV3Config, layer_idx: int):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
+        self.attention_dropout = config.attention_dropout
+        self.num_heads = config.num_attention_heads
+
+        self.q_lora_rank = config.q_lora_rank
+        self.qk_rope_head_dim = config.qk_rope_head_dim
+        self.kv_lora_rank = config.kv_lora_rank
+        self.v_head_dim = config.v_head_dim
+        self.qk_nope_head_dim = config.qk_nope_head_dim
+        self.qk_head_dim = config.qk_head_dim
+
+        self.is_causal = True
+        if self.q_lora_rank is None:
+            self.q_proj = nn.Linear(config.hidden_size, self.num_heads * self.qk_head_dim, bias=False)
+        else:
+            self.q_a_proj = nn.Linear(config.hidden_size, config.q_lora_rank, bias=config.attention_bias)
+            self.q_a_layernorm = DeepseekV3RMSNorm(config.q_lora_rank)
+            self.q_b_proj = nn.Linear(config.q_lora_rank, self.num_heads * self.qk_head_dim, bias=False)
+
+        self.kv_a_proj_with_mqa = nn.Linear(
+            config.hidden_size,
+            self.kv_lora_rank + self.qk_rope_head_dim,
+            bias=config.attention_bias,
+        )
+        self.kv_a_layernorm = DeepseekV3RMSNorm(self.kv_lora_rank)
+        self.kv_b_proj = nn.Linear(
+            self.kv_lora_rank,
+            self.num_heads * (self.qk_nope_head_dim + self.v_head_dim),
+            bias=False,
+        )
+
+        self.o_proj = nn.Linear(
+            self.num_heads * self.v_head_dim,
+            config.hidden_size,
+            bias=config.attention_bias,
+        )
+
+        self.scaling = self.qk_head_dim ** (-0.5)
+        if self.config.rope_parameters.get("rope_type", "default") != "default":
+            mscale_all_dim = self.config.rope_parameters.get("mscale_all_dim", 0)
+            scaling_factor = self.config.rope_parameters["factor"]
+            if mscale_all_dim:
+                mscale = yarn_get_mscale(scaling_factor, mscale_all_dim)
+                self.scaling = self.scaling * mscale * mscale
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        position_embeddings: tuple[torch.Tensor, torch.Tensor],
+        attention_mask: torch.Tensor | None,
+        past_key_values: Cache | None = None,
+        **kwargs: Unpack[FlashAttentionKwargs],
+    ) -> tuple[torch.Tensor, torch.Tensor | None, tuple[torch.Tensor] | None]:
+        batch_size, seq_length = hidden_states.shape[:-1]
+        query_shape = (batch_size, seq_length, -1, self.qk_head_dim)
+        key_shape = (batch_size, seq_length, -1, self.qk_nope_head_dim + self.v_head_dim)
+
+        if self.q_lora_rank is None:
+            q_states = self.q_proj(hidden_states)
+        else:
+            q_states = self.q_b_proj(self.q_a_layernorm(self.q_a_proj(hidden_states)))
+        q_states = q_states.view(query_shape).transpose(1, 2)
+        q_pass, q_rot = torch.split(q_states, [self.qk_nope_head_dim, self.qk_rope_head_dim], dim=-1)
+
+        compressed_kv = self.kv_a_proj_with_mqa(hidden_states)
+        k_pass, k_rot = torch.split(compressed_kv, [self.kv_lora_rank, self.qk_rope_head_dim], dim=-1)
+
+        k_pass = self.kv_b_proj(self.kv_a_layernorm(k_pass)).view(key_shape).transpose(1, 2)
+        k_pass, value_states = torch.split(k_pass, [self.qk_nope_head_dim, self.v_head_dim], dim=-1)
+
+        k_rot = k_rot.view(batch_size, 1, seq_length, self.qk_rope_head_dim)
+
+        cos, sin = position_embeddings
+        if self.config.rope_interleave:  # support using interleaved weights for efficiency
+            q_rot, k_rot = apply_rotary_pos_emb_interleave(q_rot, k_rot, cos, sin)
+        else:
+            q_rot, k_rot = apply_rotary_pos_emb(q_rot, k_rot, cos, sin)
+        k_rot = k_rot.expand(*k_pass.shape[:-1], -1)
+
+        query_states = torch.cat((q_pass, q_rot), dim=-1)
+        key_states = torch.cat((k_pass, k_rot), dim=-1)
+
+        if past_key_values is not None:
+            key_states, value_states = past_key_values.update(key_states, value_states, self.layer_idx)
+
+        if is_flash_attention_requested(self.config) and self.qk_head_dim != self.v_head_dim:
+            value_states = F.pad(value_states, [0, self.qk_head_dim - self.v_head_dim])
+
+        attention_interface: Callable = ALL_ATTENTION_FUNCTIONS.get_interface(
+            self.config._attn_implementation, eager_attention_forward
+        )
+
+        attn_output, attn_weights = attention_interface(
+            self,
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            dropout=0.0 if not self.training else self.attention_dropout,
+            scaling=self.scaling,
+            **kwargs,
+        )
+
+        if is_flash_attention_requested(self.config) and self.qk_head_dim != self.v_head_dim:
+            attn_output = attn_output[:, :, :, : self.v_head_dim]
+
+        attn_output = attn_output.reshape(batch_size, seq_length, -1).contiguous()
+        attn_output = self.o_proj(attn_output)
+        return attn_output, attn_weights
+
+
+class DeepseekV3DecoderLayer(GradientCheckpointingLayer):
+    def __init__(self, config: DeepseekV3Config, layer_idx: int):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+
+        self.self_attn = DeepseekV3Attention(config=config, layer_idx=layer_idx)
+
+        if layer_idx >= config.first_k_dense_replace:
+            self.mlp = DeepseekV3MoE(config)
+        else:
+            self.mlp = DeepseekV3MLP(config)
+
+        self.input_layernorm = DeepseekV3RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.post_attention_layernorm = DeepseekV3RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: torch.Tensor | None = None,
+        position_ids: torch.LongTensor | None = None,
+        past_key_values: Cache | None = None,
+        use_cache: bool | None = False,
+        position_embeddings: tuple[torch.Tensor, torch.Tensor] | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> torch.Tensor:
+        residual = hidden_states
+        hidden_states = self.input_layernorm(hidden_states)
+        # Self Attention
+        hidden_states, _ = self.self_attn(
+            hidden_states=hidden_states,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            use_cache=use_cache,
+            position_embeddings=position_embeddings,
+            **kwargs,
+        )
+        hidden_states = residual + hidden_states
+
+        # Fully Connected
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = residual + hidden_states
+        return hidden_states
+
+
+@auto_docstring
+class DeepseekV3PreTrainedModel(PreTrainedModel):
+    config: DeepseekV3Config
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["DeepseekV3DecoderLayer"]
+    _skip_keys_device_placement = ["past_key_values"]
+    _supports_flash_attn = True
+    _supports_sdpa = True
+    _supports_flex_attn = True
+
+    _can_compile_fullgraph = True
+    _supports_attention_backend = True
+    _can_record_outputs = {
+        "hidden_states": DeepseekV3DecoderLayer,
+        "attentions": DeepseekV3Attention,
+    }
+    _keep_in_fp32_modules_strict = ["e_score_correction_bias"]
+    _keys_to_ignore_on_load_unexpected = [r"model\.layers\.61.*"]
+
+    @torch.no_grad()
+    def _init_weights(self, module):
+        super()._init_weights(module)
+        if isinstance(module, DeepseekV3TopkRouter):
+            init.normal_(module.weight, mean=0.0, std=self.config.initializer_range)
+            init.zeros_(module.e_score_correction_bias)
+        elif isinstance(module, DeepseekV3NaiveMoe):
+            init.normal_(module.gate_up_proj, mean=0.0, std=self.config.initializer_range)
+            init.normal_(module.down_proj, mean=0.0, std=self.config.initializer_range)
+
+
+@auto_docstring
+class DeepseekV3Model(DeepseekV3PreTrainedModel):
+    def __init__(self, config: DeepseekV3Config):
+        super().__init__(config)
+        self.padding_idx = config.pad_token_id
+        self.vocab_size = config.vocab_size
+
+        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx)
+        self.layers = nn.ModuleList(
+            [DeepseekV3DecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
+        )
+        self.norm = DeepseekV3RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.rotary_emb = DeepseekV3RotaryEmbedding(config=config)
+        self.gradient_checkpointing = False
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @merge_with_config_defaults
+    @capture_outputs
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: torch.LongTensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        position_ids: torch.LongTensor | None = None,
+        past_key_values: Cache | None = None,
+        inputs_embeds: torch.FloatTensor | None = None,
+        use_cache: bool | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> BaseModelOutputWithPast:
+        if (input_ids is None) ^ (inputs_embeds is not None):
+            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
+
+        if inputs_embeds is None:
+            inputs_embeds: torch.Tensor = self.embed_tokens(input_ids)
+
+        if use_cache and past_key_values is None:
+            past_key_values = DynamicCache(config=self.config)
+
+        if position_ids is None:
+            past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
+            position_ids = torch.arange(inputs_embeds.shape[1], device=inputs_embeds.device) + past_seen_tokens
+            position_ids = position_ids.unsqueeze(0)
+
+        causal_mask = create_causal_mask(
+            config=self.config,
+            inputs_embeds=inputs_embeds,
+            attention_mask=attention_mask,
+            past_key_values=past_key_values,
+            position_ids=position_ids,
+        )
+
+        hidden_states = inputs_embeds
+        position_embeddings = self.rotary_emb(hidden_states, position_ids=position_ids)
+
+        for decoder_layer in self.layers[: self.config.num_hidden_layers]:
+            hidden_states = decoder_layer(
+                hidden_states,
+                attention_mask=causal_mask,
+                position_embeddings=position_embeddings,
+                position_ids=position_ids,
+                past_key_values=past_key_values,
+                use_cache=use_cache,
+                **kwargs,
+            )
+
+        hidden_states = self.norm(hidden_states)
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=past_key_values,
+        )
+
+
+@auto_docstring
+class DeepseekV3ForCausalLM(DeepseekV3PreTrainedModel, GenerationMixin):
+    _tied_weights_keys = {"lm_head.weight": "model.embed_tokens.weight"}
+    _tp_plan = {"lm_head": "colwise_gather_output"}
+    _pp_plan = {"lm_head": (["hidden_states"], ["logits"])}
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.model = DeepseekV3Model(config)
+        self.vocab_size = config.vocab_size
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: torch.LongTensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        position_ids: torch.LongTensor | None = None,
+        past_key_values: Cache | None = None,
+        inputs_embeds: torch.FloatTensor | None = None,
+        labels: torch.LongTensor | None = None,
+        use_cache: bool | None = None,
+        logits_to_keep: int | torch.Tensor = 0,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> CausalLMOutputWithPast:
+        r"""
+        Example:
+
+        ```python
+        >>> from transformers import AutoTokenizer, DeepseekV3ForCausalLM
+
+        >>> model = DeepseekV3ForCausalLM.from_pretrained("meta-deepseek_v3/DeepseekV3-2-7b-hf")
+        >>> tokenizer = AutoTokenizer.from_pretrained("meta-deepseek_v3/DeepseekV3-2-7b-hf")
+
+        >>> prompt = "Hey, are you conscious? Can you talk to me?"
+        >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+        >>> # Generate
+        >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+        >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+        "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+        ```"""
+        outputs: BaseModelOutputWithPast = self.model(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            **kwargs,
+        )
+
+        hidden_states = outputs.last_hidden_state
+        # Only compute necessary logits, and do not upcast them to float if we are not computing the loss
+        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
+        logits = self.lm_head(hidden_states[:, slice_indices, :])
+
+        loss = None
+        if labels is not None:
+            loss = self.loss_function(logits=logits, labels=labels, vocab_size=self.config.vocab_size, **kwargs)
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+
+class DeepseekV3ForSequenceClassification(GenericForSequenceClassification, DeepseekV3PreTrainedModel):
+    pass
+
+
+class DeepseekV3ForTokenClassification(GenericForTokenClassification, DeepseekV3PreTrainedModel):
+    pass
+
+
+__all__ = [
+    "DeepseekV3PreTrainedModel",
+    "DeepseekV3Model",
+    "DeepseekV3ForCausalLM",
+    "DeepseekV3ForSequenceClassification",
+    "DeepseekV3ForTokenClassification",
+]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/modular_deepseek_v3.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/modular_deepseek_v3.py
new file mode 100644
index 0000000000000000000000000000000000000000..2bf7d347e85d27bad36ff7867b9942c2a95e1d3d
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/deepseek_v3/modular_deepseek_v3.py
@@ -0,0 +1,340 @@
+import math
+from collections.abc import Callable
+
+import torch
+import torch.nn.functional as F
+from torch import nn
+
+from ... import initialization as init
+from ...cache_utils import Cache
+from ...modeling_flash_attention_utils import FlashAttentionKwargs
+from ...modeling_layers import GenericForSequenceClassification, GenericForTokenClassification
+from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
+from ...processing_utils import Unpack
+from ...utils import logging
+from ...utils.generic import is_flash_attention_requested
+from ..llama.modeling_llama import (
+    LlamaDecoderLayer,
+    LlamaForCausalLM,
+    LlamaModel,
+    LlamaPreTrainedModel,
+    LlamaRMSNorm,
+    LlamaRotaryEmbedding,
+    apply_rotary_pos_emb,
+    eager_attention_forward,
+    rotate_half,
+)
+from ..mixtral.modeling_mixtral import MixtralExperts
+from ..qwen2_moe.modeling_qwen2_moe import Qwen2MoeMLP
+from .configuration_deepseek_v3 import DeepseekV3Config
+
+
+logger = logging.get_logger(__name__)
+
+
+class DeepseekV3RMSNorm(LlamaRMSNorm):
+    pass
+
+
+class DeepseekV3RotaryEmbedding(LlamaRotaryEmbedding):
+    pass
+
+
+class DeepseekV3MLP(Qwen2MoeMLP):
+    pass
+
+
+def apply_rotary_pos_emb_interleave(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
+    r"""
+    TODO let's just use the original freqcis computation to not have the view
+    transpose + reshape! This is not optimized!
+    Applies Rotary Position Embedding to the query and key tensors.
+
+    Args:
+        q (`torch.Tensor`): The query tensor.
+        k (`torch.Tensor`): The key tensor.
+        cos (`torch.Tensor`): The cosine part of the rotary embedding.
+        sin (`torch.Tensor`): The sine part of the rotary embedding.
+        position_ids (`torch.Tensor`):
+            The position indices of the tokens corresponding to the query and key tensors. For example, this can be
+            used to pass offsetted position ids when working with a KV-cache.
+        unsqueeze_dim (`int`, *optional*, defaults to 1):
+            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
+            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
+            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
+            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
+            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
+            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
+    Returns:
+        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
+    """
+    cos = cos.unsqueeze(unsqueeze_dim)
+    sin = sin.unsqueeze(unsqueeze_dim)
+
+    b, h, s, d = q.shape
+    q = q.view(b, h, s, d // 2, 2).transpose(4, 3).reshape(b, h, s, d)
+
+    b, h, s, d = k.shape
+    k = k.view(b, h, s, d // 2, 2).transpose(4, 3).reshape(b, h, s, d)
+
+    q_embed = (q * cos) + (rotate_half(q) * sin)
+    k_embed = (k * cos) + (rotate_half(k) * sin)
+    return q_embed, k_embed
+
+
+def yarn_get_mscale(scale=1, mscale=1):
+    if scale <= 1:
+        return 1.0
+    return 0.1 * mscale * math.log(scale) + 1.0
+
+
+class DeepseekV3TopkRouter(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.n_routed_experts = config.n_routed_experts
+
+        self.weight = nn.Parameter(torch.empty((self.n_routed_experts, config.hidden_size)))
+        self.register_buffer("e_score_correction_bias", torch.zeros(self.n_routed_experts))
+
+    def forward(self, hidden_states):
+        hidden_states = hidden_states.view(-1, self.config.hidden_size)
+        router_logits = F.linear(hidden_states.type(torch.float32), self.weight.type(torch.float32))
+        return router_logits
+
+
+class DeepseekV3NaiveMoe(MixtralExperts):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_experts = config.num_local_experts
+        self.intermediate_dim = config.moe_intermediate_size
+
+
+class DeepseekV3MoE(nn.Module):
+    """
+    A mixed expert module containing shared experts.
+    """
+
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.experts = DeepseekV3NaiveMoe(config)
+        self.gate = DeepseekV3TopkRouter(config)
+        self.shared_experts = DeepseekV3MLP(
+            config=config, intermediate_size=config.moe_intermediate_size * config.n_shared_experts
+        )
+        self.n_routed_experts = config.n_routed_experts
+        self.n_group = config.n_group
+        self.topk_group = config.topk_group
+        self.norm_topk_prob = config.norm_topk_prob
+        self.routed_scaling_factor = config.routed_scaling_factor
+        self.top_k = config.num_experts_per_tok
+
+    def route_tokens_to_experts(self, router_logits):
+        router_logits = router_logits.sigmoid()
+        router_logits_for_choice = router_logits + self.gate.e_score_correction_bias
+        group_scores = (
+            router_logits_for_choice.view(-1, self.n_group, self.n_routed_experts // self.n_group)
+            .topk(2, dim=-1)[0]
+            .sum(dim=-1)
+        )
+        group_idx = torch.topk(group_scores, k=self.topk_group, dim=-1, sorted=False)[1]
+        group_mask = torch.zeros_like(group_scores)
+        group_mask.scatter_(1, group_idx, 1)
+        score_mask = (
+            group_mask.unsqueeze(-1)
+            .expand(-1, self.n_group, self.n_routed_experts // self.n_group)
+            .reshape(-1, self.n_routed_experts)
+        )
+        scores_for_choice = router_logits_for_choice.masked_fill(~score_mask.bool(), float("-inf"))
+        topk_indices = torch.topk(scores_for_choice, k=self.top_k, dim=-1, sorted=False)[1]
+        topk_weights = router_logits.gather(1, topk_indices)
+        if self.norm_topk_prob:
+            denominator = topk_weights.sum(dim=-1, keepdim=True) + 1e-20
+            topk_weights /= denominator
+        topk_weights = topk_weights * self.routed_scaling_factor
+        return topk_indices, topk_weights
+
+    def forward(self, hidden_states):
+        residuals = hidden_states
+        orig_shape = hidden_states.shape
+        router_logits = self.gate(hidden_states)
+        topk_indices, topk_weights = self.route_tokens_to_experts(router_logits)
+        hidden_states = hidden_states.view(-1, hidden_states.shape[-1])
+        hidden_states = self.experts(hidden_states, topk_indices, topk_weights).view(*orig_shape)
+        hidden_states = hidden_states + self.shared_experts(residuals)
+        return hidden_states
+
+
+class DeepseekV3Attention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: DeepseekV3Config, layer_idx: int):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
+        self.attention_dropout = config.attention_dropout
+        self.num_heads = config.num_attention_heads
+
+        self.q_lora_rank = config.q_lora_rank
+        self.qk_rope_head_dim = config.qk_rope_head_dim
+        self.kv_lora_rank = config.kv_lora_rank
+        self.v_head_dim = config.v_head_dim
+        self.qk_nope_head_dim = config.qk_nope_head_dim
+        self.qk_head_dim = config.qk_head_dim
+
+        self.is_causal = True
+        if self.q_lora_rank is None:
+            self.q_proj = nn.Linear(config.hidden_size, self.num_heads * self.qk_head_dim, bias=False)
+        else:
+            self.q_a_proj = nn.Linear(config.hidden_size, config.q_lora_rank, bias=config.attention_bias)
+            self.q_a_layernorm = DeepseekV3RMSNorm(config.q_lora_rank)
+            self.q_b_proj = nn.Linear(config.q_lora_rank, self.num_heads * self.qk_head_dim, bias=False)
+
+        self.kv_a_proj_with_mqa = nn.Linear(
+            config.hidden_size,
+            self.kv_lora_rank + self.qk_rope_head_dim,
+            bias=config.attention_bias,
+        )
+        self.kv_a_layernorm = DeepseekV3RMSNorm(self.kv_lora_rank)
+        self.kv_b_proj = nn.Linear(
+            self.kv_lora_rank,
+            self.num_heads * (self.qk_nope_head_dim + self.v_head_dim),
+            bias=False,
+        )
+
+        self.o_proj = nn.Linear(
+            self.num_heads * self.v_head_dim,
+            config.hidden_size,
+            bias=config.attention_bias,
+        )
+
+        self.scaling = self.qk_head_dim ** (-0.5)
+        if self.config.rope_parameters.get("rope_type", "default") != "default":
+            mscale_all_dim = self.config.rope_parameters.get("mscale_all_dim", 0)
+            scaling_factor = self.config.rope_parameters["factor"]
+            if mscale_all_dim:
+                mscale = yarn_get_mscale(scaling_factor, mscale_all_dim)
+                self.scaling = self.scaling * mscale * mscale
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        position_embeddings: tuple[torch.Tensor, torch.Tensor],
+        attention_mask: torch.Tensor | None,
+        past_key_values: Cache | None = None,
+        **kwargs: Unpack[FlashAttentionKwargs],
+    ) -> tuple[torch.Tensor, torch.Tensor | None, tuple[torch.Tensor] | None]:
+        batch_size, seq_length = hidden_states.shape[:-1]
+        query_shape = (batch_size, seq_length, -1, self.qk_head_dim)
+        key_shape = (batch_size, seq_length, -1, self.qk_nope_head_dim + self.v_head_dim)
+
+        if self.q_lora_rank is None:
+            q_states = self.q_proj(hidden_states)
+        else:
+            q_states = self.q_b_proj(self.q_a_layernorm(self.q_a_proj(hidden_states)))
+        q_states = q_states.view(query_shape).transpose(1, 2)
+        q_pass, q_rot = torch.split(q_states, [self.qk_nope_head_dim, self.qk_rope_head_dim], dim=-1)
+
+        compressed_kv = self.kv_a_proj_with_mqa(hidden_states)
+        k_pass, k_rot = torch.split(compressed_kv, [self.kv_lora_rank, self.qk_rope_head_dim], dim=-1)
+
+        k_pass = self.kv_b_proj(self.kv_a_layernorm(k_pass)).view(key_shape).transpose(1, 2)
+        k_pass, value_states = torch.split(k_pass, [self.qk_nope_head_dim, self.v_head_dim], dim=-1)
+
+        k_rot = k_rot.view(batch_size, 1, seq_length, self.qk_rope_head_dim)
+
+        cos, sin = position_embeddings
+        if self.config.rope_interleave:  # support using interleaved weights for efficiency
+            q_rot, k_rot = apply_rotary_pos_emb_interleave(q_rot, k_rot, cos, sin)
+        else:
+            q_rot, k_rot = apply_rotary_pos_emb(q_rot, k_rot, cos, sin)
+        k_rot = k_rot.expand(*k_pass.shape[:-1], -1)
+
+        query_states = torch.cat((q_pass, q_rot), dim=-1)
+        key_states = torch.cat((k_pass, k_rot), dim=-1)
+
+        if past_key_values is not None:
+            key_states, value_states = past_key_values.update(key_states, value_states, self.layer_idx)
+
+        if is_flash_attention_requested(self.config) and self.qk_head_dim != self.v_head_dim:
+            value_states = F.pad(value_states, [0, self.qk_head_dim - self.v_head_dim])
+
+        attention_interface: Callable = ALL_ATTENTION_FUNCTIONS.get_interface(
+            self.config._attn_implementation, eager_attention_forward
+        )
+
+        attn_output, attn_weights = attention_interface(
+            self,
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            dropout=0.0 if not self.training else self.attention_dropout,
+            scaling=self.scaling,
+            **kwargs,
+        )
+
+        if is_flash_attention_requested(self.config) and self.qk_head_dim != self.v_head_dim:
+            attn_output = attn_output[:, :, :, : self.v_head_dim]
+
+        attn_output = attn_output.reshape(batch_size, seq_length, -1).contiguous()
+        attn_output = self.o_proj(attn_output)
+        return attn_output, attn_weights
+
+
+class DeepseekV3DecoderLayer(LlamaDecoderLayer):
+    def __init__(self, config: DeepseekV3Config, layer_idx: int):
+        nn.Module.__init__(self)
+        self.hidden_size = config.hidden_size
+
+        self.self_attn = DeepseekV3Attention(config=config, layer_idx=layer_idx)
+
+        if layer_idx >= config.first_k_dense_replace:
+            self.mlp = DeepseekV3MoE(config)
+        else:
+            self.mlp = DeepseekV3MLP(config)
+
+        self.input_layernorm = DeepseekV3RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.post_attention_layernorm = DeepseekV3RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+
+class DeepseekV3PreTrainedModel(LlamaPreTrainedModel):
+    _keep_in_fp32_modules_strict = ["e_score_correction_bias"]
+    _keys_to_ignore_on_load_unexpected = [r"model\.layers\.61.*"]
+
+    @torch.no_grad()
+    def _init_weights(self, module):
+        PreTrainedModel._init_weights(self, module)
+        if isinstance(module, DeepseekV3TopkRouter):
+            init.normal_(module.weight, mean=0.0, std=self.config.initializer_range)
+            init.zeros_(module.e_score_correction_bias)
+        elif isinstance(module, DeepseekV3NaiveMoe):
+            init.normal_(module.gate_up_proj, mean=0.0, std=self.config.initializer_range)
+            init.normal_(module.down_proj, mean=0.0, std=self.config.initializer_range)
+
+
+class DeepseekV3Model(LlamaModel):
+    pass
+
+
+class DeepseekV3ForCausalLM(LlamaForCausalLM):
+    pass
+
+
+class DeepseekV3ForSequenceClassification(GenericForSequenceClassification, DeepseekV3PreTrainedModel):
+    pass
+
+
+class DeepseekV3ForTokenClassification(GenericForTokenClassification, DeepseekV3PreTrainedModel):
+    pass
+
+
+__all__ = [
+    "DeepseekV3PreTrainedModel",
+    "DeepseekV3Model",
+    "DeepseekV3ForCausalLM",
+    "DeepseekV3ForSequenceClassification",
+    "DeepseekV3ForTokenClassification",
+]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/__init__.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..ee436e5c05b7fbc1e399dce607f6b539210e3daf
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/__init__.py
@@ -0,0 +1,31 @@
+# Copyright 2024 The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import TYPE_CHECKING
+
+from ...utils import _LazyModule
+from ...utils.import_utils import define_import_structure
+
+
+if TYPE_CHECKING:
+    from .configuration_llava_onevision import *
+    from .image_processing_llava_onevision import *
+    from .image_processing_pil_llava_onevision import *
+    from .modeling_llava_onevision import *
+    from .processing_llava_onevision import *
+    from .video_processing_llava_onevision import *
+else:
+    import sys
+
+    _file = globals()["__file__"]
+    sys.modules[__name__] = _LazyModule(__name__, _file, define_import_structure(_file), module_spec=__spec__)
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/configuration_llava_onevision.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/configuration_llava_onevision.py
new file mode 100644
index 0000000000000000000000000000000000000000..29e5f69b87a0e0c7667d5cb217f81d1039af053e
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/configuration_llava_onevision.py
@@ -0,0 +1,147 @@
+# Copyright 2024 HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from typing import Literal
+
+from huggingface_hub.dataclasses import strict
+
+from ...configuration_utils import PreTrainedConfig
+from ...utils import auto_docstring
+from ..auto import CONFIG_MAPPING, AutoConfig
+
+
+@auto_docstring(checkpoint="llava-hf/llava-onevision-qwen2-7b-ov-hf")
+@strict
+class LlavaOnevisionConfig(PreTrainedConfig):
+    r"""
+    image_grid_pinpoints (`List`, *optional*):
+        A list of possible resolutions to use for processing high resolution images. Each item in the list should be a tuple or list
+        of the form `(height, width)`.
+    vision_aspect_ratio (`str`, *optional*, defaults to `"anyres_max_9"`):
+        Aspect ratio used when processong image features. The default value is "anyres_max_9".
+
+    Example:
+
+    ```python
+    >>> from transformers import LlavaOnevisionForConditionalGeneration, LlavaOnevisionConfig, SiglipVisionConfig, Qwen2Config
+
+    >>> # Initializing a CLIP-vision config
+    >>> vision_config = SiglipVisionConfig()
+
+    >>> # Initializing a Llama config
+    >>> text_config = Qwen2Config()
+
+    >>> # Initializing a Llava-Next llava-hf/llava-onevision-qwen2-7b-ov-hf style configuration
+    >>> configuration = LlavaOnevisionConfig(vision_config, text_config)
+
+    >>> # Initializing a model from the llava-hf/llava-onevision-qwen2-7b-ov-hf style configuration
+    >>> model = LlavaOnevisionForConditionalGeneration(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "llava_onevision"
+    attribute_map = {
+        "image_token_id": "image_token_index",
+        "video_token_id": "video_token_index",
+    }
+    sub_configs = {"text_config": AutoConfig, "vision_config": AutoConfig}
+
+    vision_config: dict | PreTrainedConfig | None = None
+    text_config: dict | PreTrainedConfig | None = None
+    image_token_index: int = 151646
+    video_token_index: int = 151647
+    projector_hidden_act: str = "gelu"
+    vision_feature_select_strategy: Literal["default", "full"] = "full"
+    vision_feature_layer: int | list[int] = -1
+    multimodal_projector_bias: bool = True
+    tie_word_embeddings: bool = False
+    image_grid_pinpoints: list | None = None
+    vision_aspect_ratio: str = "anyres_max_9"
+
+    def __post_init__(self, **kwargs):
+        if isinstance(self.vision_config, dict):
+            self.vision_config["model_type"] = self.vision_config.get("model_type", "siglip_vision_model")
+            self.vision_config = CONFIG_MAPPING[self.vision_config["model_type"]](**self.vision_config)
+        elif self.vision_config is None:
+            self.vision_config = CONFIG_MAPPING["siglip_vision_model"](
+                hidden_size=1152,
+                intermediate_size=4304,
+                patch_size=14,
+                image_size=384,
+                num_hidden_layers=26,
+                num_attention_heads=16,
+                vision_use_head=False,
+            )
+
+        if isinstance(self.text_config, dict):
+            self.text_config["model_type"] = self.text_config.get("model_type", "qwen2")
+            self.text_config = CONFIG_MAPPING[self.text_config["model_type"]](**self.text_config)
+        elif self.text_config is None:
+            self.text_config = CONFIG_MAPPING["qwen2"]()
+
+        self.image_grid_pinpoints = (
+            self.image_grid_pinpoints
+            if self.image_grid_pinpoints is not None
+            else [
+                [384, 384],
+                [384, 768],
+                [384, 1152],
+                [384, 1536],
+                [384, 1920],
+                [384, 2304],
+                [768, 384],
+                [768, 768],
+                [768, 1152],
+                [768, 1536],
+                [768, 1920],
+                [768, 2304],
+                [1152, 384],
+                [1152, 768],
+                [1152, 1152],
+                [1152, 1536],
+                [1152, 1920],
+                [1152, 2304],
+                [1536, 384],
+                [1536, 768],
+                [1536, 1152],
+                [1536, 1536],
+                [1536, 1920],
+                [1536, 2304],
+                [1920, 384],
+                [1920, 768],
+                [1920, 1152],
+                [1920, 1536],
+                [1920, 1920],
+                [1920, 2304],
+                [2304, 384],
+                [2304, 768],
+                [2304, 1152],
+                [2304, 1536],
+                [2304, 1920],
+                [2304, 2304],
+            ]
+        )
+
+        # The default value is `False` but this config is used with many model types
+        # Attr `tie_word_embeddings` was saved in text config for those models, so we
+        # need an ugly workaround and forward-pass the attr from text config
+        if not self.tie_word_embeddings and self.text_config.tie_word_embeddings:
+            self.tie_word_embeddings = self.text_config.tie_word_embeddings
+
+        super().__post_init__(**kwargs)
+
+
+__all__ = ["LlavaOnevisionConfig"]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/image_processing_pil_llava_onevision.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/image_processing_pil_llava_onevision.py
new file mode 100644
index 0000000000000000000000000000000000000000..23534a65d70f11ab6b4eaa96c72210be85930e0a
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/image_processing_pil_llava_onevision.py
@@ -0,0 +1,307 @@
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+#           This file was automatically generated from src/transformers/models/llava_onevision/modular_llava_onevision.py.
+#               Do NOT edit this file manually as any edits will be overwritten by the generation of
+#             the file from the modular. If any change should be done, please apply the change to the
+#                          modular_llava_onevision.py file directly. One of our CI enforces this.
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+# Copyright 2024 the HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+import numpy as np
+
+from ...image_processing_backends import PilBackend
+from ...image_processing_utils import BatchFeature, get_patch_output_size, select_best_resolution
+from ...image_transforms import divide_to_patches
+from ...image_utils import (
+    OPENAI_CLIP_MEAN,
+    OPENAI_CLIP_STD,
+    ChannelDimension,
+    ImageInput,
+    PILImageResampling,
+    SizeDict,
+)
+from ...processing_utils import ImagesKwargs, Unpack
+from ...utils import TensorType, auto_docstring
+
+
+class LlavaOnevisionImageProcessorKwargs(ImagesKwargs, total=False):
+    r"""
+    image_grid_pinpoints (`list[list[int]]`, *optional*):
+        A list of possible resolutions to use for processing high resolution images. The best resolution is selected
+        based on the original size of the image. Can be overridden by `image_grid_pinpoints` in the `preprocess`
+        method.
+    """
+
+    image_grid_pinpoints: list[list[int]]
+
+
+@auto_docstring
+class LlavaOnevisionImageProcessorPil(PilBackend):
+    model_input_names = ["pixel_values", "image_sizes", "batch_num_images"]
+    valid_kwargs = LlavaOnevisionImageProcessorKwargs
+    resample = PILImageResampling.BICUBIC
+    image_mean = OPENAI_CLIP_MEAN
+    image_std = OPENAI_CLIP_STD
+    size = {"height": 384, "width": 384}
+    default_to_square = False
+    crop_size = None
+    do_resize = True
+    do_center_crop = None
+    do_rescale = True
+    do_normalize = True
+    do_convert_rgb = True
+    do_pad = True
+    image_grid_pinpoints = [[384, 384], [384, 768], [384, 1152], [384, 1536], [384, 1920], [384, 2304], [768, 384], [768, 768], [768, 1152], [768, 1536], [768, 1920], [768, 2304], [1152, 384], [1152, 768], [1152, 1152], [1152, 1536], [1152, 1920], [1152, 2304], [1536, 384], [1536, 768], [1536, 1152], [1536, 1536], [1536, 1920], [1536, 2304], [1920, 384], [1920, 768], [1920, 1152], [1920, 1536], [1920, 1920], [1920, 2304], [2304, 384], [2304, 768], [2304, 1152], [2304, 1536], [2304, 1920], [2304, 2304]]  # fmt: skip
+
+    def __init__(self, **kwargs: Unpack[LlavaOnevisionImageProcessorKwargs]):
+        super().__init__(**kwargs)
+
+    @auto_docstring
+    def preprocess(self, images: ImageInput, **kwargs: Unpack[LlavaOnevisionImageProcessorKwargs]) -> BatchFeature:
+        if isinstance(images, (tuple, list)) and isinstance(images[0], (tuple, list)):
+            # if the first element is a list, we assume that all elements are lists
+            images = [x for x in images if x]  # handle text-only case
+            batch_num_images = [len(x) for x in images]
+        elif isinstance(images, (tuple, list)):
+            # treat this as a single-image case for backward compatibility
+            batch_num_images = [1] * len(images)
+        else:
+            batch_num_images = [1]
+        return super().preprocess(images, batch_num_images, **kwargs)
+
+    def _get_padding_size(self, original_resolution: tuple, target_resolution: tuple):
+        """Get padding size for patching (returns tuple format for np.pad)."""
+        original_height, original_width = original_resolution
+        target_height, target_width = target_resolution
+        paste_x, r_x = divmod(target_width - original_width, 2)
+        paste_y, r_y = divmod(target_height - original_height, 2)
+        return (paste_y, paste_y + r_y), (paste_x, paste_x + r_x)
+
+    def _resize_for_patching(
+        self,
+        image: np.ndarray,
+        target_resolution: tuple,
+        resample: PILImageResampling,
+    ) -> np.ndarray:
+        """Resizes an image to a target resolution while maintaining aspect ratio."""
+        new_height, new_width = get_patch_output_size(
+            image, target_resolution, input_data_format=ChannelDimension.FIRST
+        )
+        resized_image = self.resize(image=image, size=SizeDict(height=new_height, width=new_width), resample=resample)
+
+        return resized_image
+
+    def _pad_for_patching(self, image: np.ndarray, target_resolution: tuple) -> np.ndarray:
+        """Pad an image to a target resolution while maintaining aspect ratio."""
+        new_resolution = get_patch_output_size(image, target_resolution, input_data_format=ChannelDimension.FIRST)
+        padding_hw = self._get_padding_size(new_resolution, target_resolution)
+
+        # For channels_first format (C, H, W), add (0, 0) for channel dimension
+        # padding_hw is ((before_h, after_h), (before_w, after_w))
+        # np.pad expects ((before_C, after_C), (before_H, after_H), (before_W, after_W))
+        padding = ((0, 0), padding_hw[0], padding_hw[1])
+
+        # Use np.pad directly for patching padding
+        padded_image = np.pad(image, padding, mode="constant", constant_values=0)
+
+        return padded_image
+
+    def get_image_patches(
+        self,
+        image: np.ndarray,
+        grid_pinpoints: list[list[int]],
+        size: tuple,
+        patch_size: int,
+        resample: PILImageResampling,
+    ) -> list[np.ndarray]:
+        """Process an image with variable resolutions by dividing it into patches."""
+        if not isinstance(grid_pinpoints, list):
+            raise TypeError("grid_pinpoints must be a list of possible resolutions.")
+
+        possible_resolutions = grid_pinpoints
+
+        image_size = image.shape[-2:]
+        best_resolution = select_best_resolution(image_size, possible_resolutions)
+        resized_image = self._resize_for_patching(image, best_resolution, resample=resample)
+        padded_image = self._pad_for_patching(resized_image, best_resolution)
+
+        patches = divide_to_patches(padded_image, patch_size=patch_size)
+
+        size_height, size_width = size
+        resized_original_image = self.resize(
+            image=image,
+            size=SizeDict(height=size_height, width=size_width),
+            resample=resample,
+        )
+
+        image_patches = [resized_original_image] + patches
+
+        return image_patches
+
+    def _pad_for_batching(
+        self,
+        pixel_values: list[np.ndarray],
+    ) -> list[np.ndarray]:
+        """Pads images on the `num_of_patches` dimension with zeros to form a batch of same number of patches."""
+        max_patch = max(len(x) for x in pixel_values)
+        # Use np.pad directly for patch dimension padding
+        padded_values = []
+        for image in pixel_values:
+            # Padding format: ((before_dim0, after_dim0), (before_dim1, after_dim1), ...)
+            padding = ((0, max_patch - image.shape[0]), (0, 0), (0, 0), (0, 0))
+            padded_image = np.pad(image, padding, mode="constant", constant_values=0)
+            padded_values.append(padded_image)
+
+        return padded_values
+
+    def _preprocess(
+        self,
+        images: list[np.ndarray],
+        batch_num_images: list[int],
+        do_resize: bool,
+        size: SizeDict,
+        image_grid_pinpoints: list[list[int]],
+        resample: "PILImageResampling | None",
+        do_center_crop: bool,
+        crop_size: SizeDict,
+        do_rescale: bool,
+        rescale_factor: float,
+        do_normalize: bool,
+        image_mean: float | list[float] | None,
+        image_std: float | list[float] | None,
+        do_pad: bool | None,
+        pad_size: SizeDict | None,
+        disable_grouping: bool | None,
+        return_tensors: str | TensorType | None,
+        **kwargs,
+    ) -> BatchFeature:
+        """Custom preprocessing for LLaVA-NeXT with patch processing."""
+        processed_images = []
+        image_sizes = []
+        # only single image patching is supported
+        need_patching = [n == 1 for n in batch_num_images for _ in range(n)]
+
+        # Backend's resize method handles resample conversion, so we can pass it directly
+        # Determine the size tuple
+        if size and size.height and size.width:
+            size_tuple = (size.height, size.width)
+        else:
+            size_tuple = (size.shortest_edge, size.shortest_edge)
+
+        # Determine the patch size
+        if crop_size and crop_size.height:
+            patch_size = crop_size.height
+        elif size and size.height:
+            patch_size = size.height
+        else:
+            patch_size = size.shortest_edge
+
+        for i, image in enumerate(images):
+            if need_patching[i]:
+                # convert image into a list of patches
+                # we intentionally use the same data format as the input data format
+                image_patches = self.get_image_patches(
+                    image,
+                    image_grid_pinpoints,
+                    size=size_tuple,
+                    patch_size=patch_size,
+                    resample=resample,
+                )
+            else:
+                padded_image = self.pad_to_square(
+                    image=image, background_color=tuple(int(x * 255) for x in self.image_mean)
+                )
+                image_patches = [padded_image]
+
+            # preprocess patches
+            pixel_values = []
+            for patch in image_patches:
+                if do_resize:
+                    patch = self.resize(image=patch, size=size, resample=resample)
+
+                if do_center_crop:
+                    patch = self.center_crop(image=patch, size=crop_size)
+
+                if do_rescale:
+                    patch = self.rescale(image=patch, scale=rescale_factor)
+
+                if do_normalize:
+                    patch = self.normalize(image=patch, mean=image_mean, std=image_std)
+
+                pixel_values.append(patch)
+
+            pixel_values = np.array(pixel_values)
+            processed_images.append(pixel_values)
+            image_sizes.append(image.shape[-2:])
+
+        if do_pad:
+            processed_images = self._pad_for_batching(processed_images)
+
+        return BatchFeature(
+            data={
+                "pixel_values": processed_images,
+                "image_sizes": image_sizes,
+                "batch_num_images": batch_num_images,
+            },
+            tensor_type=return_tensors,
+        )
+
+    def pad_to_square(
+        self,
+        image: np.ndarray,
+        background_color: int | tuple[int, int, int] = 0,
+    ) -> np.ndarray:
+        """
+        Pads an image to a square based on the longest edge.
+
+        Args:
+            image (`np.ndarray`):
+                The image to pad. Shape: (num_channels, height, width) - always channels_first in backend.
+            background_color (`int` or `tuple[int, int, int]`, *optional*, defaults to 0):
+                The color to use for the padding.
+
+        Returns:
+            `np.ndarray`: The padded image.
+        """
+        # Backend always uses channels_first format: (num_channels, height, width)
+        num_channels, height, width = image.shape
+
+        if height == width:
+            return image
+
+        max_dim = max(height, width)
+
+        # Ensure background_color is the correct shape
+        if isinstance(background_color, int):
+            background_color = [background_color]
+        elif len(background_color) != num_channels:
+            raise ValueError(
+                f"background_color must have no more than {num_channels} elements to match the number of channels"
+            )
+
+        result = np.zeros((num_channels, max_dim, max_dim), dtype=image.dtype)
+        for i, color in enumerate(background_color):
+            result[i, :, :] = color
+        if width > height:
+            start = (max_dim - height) // 2
+            result[:, start : start + height, :] = image
+        else:
+            start = (max_dim - width) // 2
+            result[:, :, start : start + width] = image
+
+        return result
+
+
+__all__ = ["LlavaOnevisionImageProcessorPil"]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/modeling_llava_onevision.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/modeling_llava_onevision.py
new file mode 100644
index 0000000000000000000000000000000000000000..335ebd004455e80802976955bb232a222b52d661
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/modeling_llava_onevision.py
@@ -0,0 +1,848 @@
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+#           This file was automatically generated from src/transformers/models/llava_onevision/modular_llava_onevision.py.
+#               Do NOT edit this file manually as any edits will be overwritten by the generation of
+#             the file from the modular. If any change should be done, please apply the change to the
+#                          modular_llava_onevision.py file directly. One of our CI enforces this.
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+# Copyright 2024 the HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import math
+from dataclasses import dataclass
+
+import numpy as np
+import torch
+from torch import nn
+
+from ... import initialization as init
+from ...activations import ACT2FN
+from ...cache_utils import Cache
+from ...generation import GenerationMixin
+from ...image_processing_utils import select_best_resolution
+from ...modeling_flash_attention_utils import FlashAttentionKwargs
+from ...modeling_outputs import BaseModelOutputWithPast, BaseModelOutputWithPooling, ModelOutput
+from ...modeling_utils import PreTrainedModel
+from ...processing_utils import Unpack
+from ...utils import TransformersKwargs, auto_docstring, torch_compilable_check
+from ...utils.generic import can_return_tuple, merge_with_config_defaults
+from ..auto import AutoModel
+from .configuration_llava_onevision import LlavaOnevisionConfig
+
+
+@auto_docstring(
+    custom_intro="""
+    Base class for Llava outputs, with hidden states and attentions.
+    """
+)
+@dataclass
+class LlavaOnevisionModelOutputWithPast(BaseModelOutputWithPast):
+    r"""
+    past_key_values (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
+        It is a [`~cache_utils.Cache`] instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache).
+
+        Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see
+        `past_key_values` input) to speed up sequential decoding.
+    image_hidden_states (`torch.FloatTensor`, *optional*):
+        A `torch.FloatTensor` of size `(batch_size, num_images, sequence_length, hidden_size)`.
+        image_hidden_states of the model produced by the vision encoder and after projecting the last hidden state.
+    video_hidden_states (`torch.FloatTensor`, *optional*):
+        A `torch.FloatTensor`  of size `(batch_size * num_frames, num_videos, sequence_length, hidden_size)`.
+        video_hidden_states of the model produced by the vision encoder and after projecting the last hidden state.
+    """
+
+    image_hidden_states: torch.FloatTensor | None = None
+
+    video_hidden_states: torch.FloatTensor | None = None
+
+
+@auto_docstring(
+    custom_intro="""
+    Base class for LlavaOnevision causal language model (or autoregressive) outputs.
+    """
+)
+@dataclass
+class LlavaOnevisionCausalLMOutputWithPast(ModelOutput):
+    r"""
+    loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
+        Language modeling loss (for next-token prediction).
+    logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
+        Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
+    past_key_values (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
+        It is a [`~cache_utils.Cache`] instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache).
+
+        Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see
+        `past_key_values` input) to speed up sequential decoding.
+    image_hidden_states (`torch.FloatTensor`, *optional*):
+        A `torch.FloatTensor` of size (batch_size * num_patches, num_images, sequence_length, hidden_size)`.
+        image_hidden_states of the model produced by the vision encoder and after projecting the last hidden state.
+    video_hidden_states (`torch.FloatTensor`, *optional*):
+        A `torch.FloatTensor`  of size `(batch_size * num_frames, num_videos, sequence_length, hidden_size)`.
+        video_hidden_states of the model produced by the vision encoder and after projecting the last hidden state.
+    """
+
+    loss: torch.FloatTensor | None = None
+    logits: torch.FloatTensor | None = None
+    past_key_values: Cache | None = None
+    hidden_states: tuple[torch.FloatTensor] | None = None
+    attentions: tuple[torch.FloatTensor] | None = None
+    image_hidden_states: torch.FloatTensor | None = None
+
+    video_hidden_states: torch.FloatTensor | None = None
+
+
+@auto_docstring
+class LlavaOnevisionPreTrainedModel(PreTrainedModel):
+    config: LlavaOnevisionConfig
+    base_model_prefix = "model"
+    input_modalities = ("image", "video", "text")
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["LlamaDecoderLayer"]
+    _skip_keys_device_placement = ["past_key_values"]
+
+    _supports_flash_attn = True
+    _supports_sdpa = True
+
+    _can_compile_fullgraph = True
+    _supports_flex_attn = True
+    _supports_attention_backend = True
+    _can_record_outputs = {
+        "hidden_states": "LlamaDecoderLayer",
+        "attentions": "LlamaAttention",
+    }
+
+    @torch.no_grad()
+    def _init_weights(self, module):
+        super()._init_weights(module)
+        if isinstance(module, LlavaOnevisionModel):
+            embed_std = 1 / math.sqrt(self.config.text_config.hidden_size)
+            init.normal_(module.image_newline, mean=0.0, std=embed_std)
+
+
+class LlavaOnevisionMultiModalProjector(nn.Module):
+    def __init__(self, config: LlavaOnevisionConfig):
+        super().__init__()
+        # We have hidden_size * the number of vision feature layers
+        num_feature_layers = 1 if isinstance(config.vision_feature_layer, int) else len(config.vision_feature_layer)
+        self.linear_1 = nn.Linear(
+            config.vision_config.hidden_size * num_feature_layers,
+            config.text_config.hidden_size,
+            bias=config.multimodal_projector_bias,
+        )
+        self.act = ACT2FN[config.projector_hidden_act]
+        self.linear_2 = nn.Linear(
+            config.text_config.hidden_size, config.text_config.hidden_size, bias=config.multimodal_projector_bias
+        )
+
+    def forward(self, image_features):
+        hidden_states = self.linear_1(image_features)
+        hidden_states = self.act(hidden_states)
+        hidden_states = self.linear_2(hidden_states)
+        return hidden_states
+
+
+def get_anyres_image_grid_shape(image_size, grid_pinpoints, patch_size):
+    """
+    Calculate the shape of the image patch grid after the preprocessing for images of any resolution.
+
+    Args:
+        image_size (`tuple`):
+            The size of the input image in the format (width, height).
+        grid_pinpoints (`List`):
+            A list containing possible resolutions. Each item in the list should be a tuple or list
+            of the form `(height, width)`.
+        patch_size (`int`):
+            The size of each image patch.
+
+    Returns:
+        tuple: The shape of the image patch grid in the format (width, height).
+    """
+    if not isinstance(grid_pinpoints, list):
+        raise TypeError("grid_pinpoints should be a list of tuples or lists")
+
+    # ! VERY IMPORTANT if image_size is tensor, must convert to into tuple, otherwise it will cause wrong calculate
+    if not isinstance(image_size, (list, tuple)):
+        if not isinstance(image_size, (torch.Tensor, np.ndarray)):
+            raise TypeError(
+                f"image_size invalid type: {type(image_size)} not valid, should be either list, tuple, np.ndarray or tensor"
+            )
+        image_size = image_size.tolist()
+
+    height, width = select_best_resolution(image_size, grid_pinpoints)
+    return height // patch_size, width // patch_size
+
+
+def image_size_to_num_patches(image_size, grid_pinpoints, patch_size: int):
+    """
+    Calculate the number of patches after the preprocessing for images of any resolution.
+
+    Args:
+        image_size (`torch.LongTensor` or `np.ndarray` or `tuple[int, int]`):
+            The size of the input image in the format (height, width). ?
+        grid_pinpoints (`List`):
+            A list containing possible resolutions. Each item in the list should be a tuple or list
+            of the form `(height, width)`.
+        patch_size (`int`):
+            The size of each image patch.
+
+    Returns:
+        int: the number of patches
+    """
+    if not isinstance(grid_pinpoints, list):
+        raise TypeError("grid_pinpoints should be a list of tuples or lists")
+
+    # ! VERY IMPORTANT if image_size is tensor, must convert to into tuple, otherwise it will cause wrong calculate
+    if not isinstance(image_size, (list, tuple)):
+        if not isinstance(image_size, (torch.Tensor, np.ndarray)):
+            raise TypeError(f"image_size invalid type {type(image_size)} with value {image_size}")
+        image_size = image_size.tolist()
+
+    best_resolution = select_best_resolution(image_size, grid_pinpoints)
+    height, width = best_resolution
+    num_patches = 0
+    # consider change to ceil(height/patch_size)*ceil(width/patch_size) + 1
+    for i in range(0, height, patch_size):
+        for j in range(0, width, patch_size):
+            num_patches += 1
+    # add the base patch
+    num_patches += 1
+    return num_patches
+
+
+def unpad_image(tensor, original_size):
+    """
+    Unpads a PyTorch tensor of a padded and resized image.
+
+    Args:
+        tensor (`torch.Tensor`):
+            The image tensor, assumed to be of shape (num_channels, height, width).
+        original_size (`tuple`):
+            The original size of the image (height, width).
+
+    Returns:
+        `torch.Tensor`: The unpadded image tensor.
+    """
+    if not isinstance(original_size, (list, tuple)):
+        if not isinstance(original_size, (torch.Tensor, np.ndarray)):
+            raise TypeError(
+                f"image_size invalid type: {type(original_size)} not valid, should be either list, tuple, np.ndarray or tensor"
+            )
+        original_size = original_size.tolist()
+    original_height, original_width = original_size
+    current_height, current_width = tensor.shape[1:]
+
+    original_aspect_ratio = original_width / original_height
+    current_aspect_ratio = current_width / current_height
+
+    if original_aspect_ratio > current_aspect_ratio:
+        scale_factor = current_width / original_width
+        new_height = int(round(original_height * scale_factor, 7))
+        padding = (current_height - new_height) // 2
+        unpadded_tensor = tensor[:, padding : current_height - padding, :]
+    else:
+        scale_factor = current_height / original_height
+        new_width = int(round(original_width * scale_factor, 7))
+        padding = (current_width - new_width) // 2
+        unpadded_tensor = tensor[:, :, padding : current_width - padding]
+
+    return unpadded_tensor
+
+
+@auto_docstring(
+    custom_intro="""
+    The Llava-Next model which consists of a vision backbone and a language model without language modeling head.
+    """
+)
+class LlavaOnevisionModel(LlavaOnevisionPreTrainedModel):
+    base_model_prefix = "model"
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.vision_tower = AutoModel.from_config(config.vision_config)
+
+        self.multi_modal_projector = LlavaOnevisionMultiModalProjector(config)
+        embed_std = 1 / math.sqrt(config.text_config.hidden_size)
+        self.image_newline = nn.Parameter(torch.randn(config.text_config.hidden_size, dtype=self.dtype) * embed_std)
+
+        self.vocab_size = config.text_config.vocab_size
+        self.language_model = AutoModel.from_config(config.text_config)
+        self.post_init()
+
+    def pack_image_features(self, image_features, image_sizes, image_newline=None, vision_aspect_ratio="anyres_max_9"):
+        """
+        Reshape, unpad and then pack each image_feature into a single image_features tensor containing all visual vectors.
+
+        Args:
+            image_features (`list[torch.Tensor]` of length num_images, each of shape `(num_patches, image_length, embed_dim)`)
+                List of image feature tensor, each contains all the visual feature of all patches.
+            image_sizes (`torch.Tensor` of shape `(num_images, 2)`)
+                Actual image size of each images (H, W).
+            image_newline (`torch.Tensor` of shape `(embed_dim)`)
+                New line embedding vector.
+            vision_aspect_ratio (`str`, *optional*, "anyres_max_9"):
+                Aspect ratio used when processing image features. The default value is "anyres_max_9".
+        Returns:
+            image_features (`torch.Tensor` of shape `(all_feat_len, embed_dim)`)
+            feature_lens (`list[int]`)
+                token length of each image in image_features
+        """
+        new_image_features = []
+        feature_lens = []
+        for image_idx, image_feature in enumerate(image_features):
+            if image_feature.shape[0] > 1:
+                base_image_feature = image_feature[0]
+                image_feature = image_feature[1:]
+                height = width = self.config.vision_config.image_size // self.config.vision_config.patch_size
+                if height * width != base_image_feature.shape[0]:
+                    raise ValueError("The number of patches is not consistent with the image size.")
+                num_patch_height, num_patch_width = get_anyres_image_grid_shape(
+                    image_sizes[image_idx],
+                    self.config.image_grid_pinpoints,
+                    self.config.vision_config.image_size,
+                )
+                image_feature = image_feature.view(num_patch_height, num_patch_width, height, width, -1)
+                image_feature = image_feature.permute(4, 0, 2, 1, 3).contiguous()
+                image_feature = image_feature.flatten(1, 2).flatten(2, 3)
+                image_feature = unpad_image(image_feature, image_sizes[image_idx])
+                max_num_patches = int(vision_aspect_ratio.strip("anyres_max_"))
+                channels, curr_height, curr_width = image_feature.shape
+                ratio = math.sqrt(curr_height * curr_width / (max_num_patches * height**2))
+                if ratio > 1.1:
+                    image_feature = image_feature[None]
+                    image_feature = nn.functional.interpolate(
+                        image_feature, [int(curr_height // ratio), int(curr_width // ratio)], mode="bilinear"
+                    )[0]
+                if image_newline is not None:
+                    image_feature = torch.cat(
+                        (
+                            image_feature,
+                            image_newline[:, None, None]
+                            .expand(*image_feature.shape[:-1], 1)
+                            .to(image_feature.device, image_feature.dtype),
+                        ),
+                        dim=-1,
+                    )
+                image_feature = image_feature.flatten(1, 2).transpose(0, 1)
+                image_feature = torch.cat((base_image_feature, image_feature), dim=0)
+            else:
+                image_feature = image_feature[0]
+                if image_newline is not None:
+                    image_feature = torch.cat((image_feature, image_newline[None].to(image_feature)), dim=0)
+            new_image_features.append(image_feature)
+            feature_lens.append(image_feature.size(0))
+        feature_lens = torch.tensor(feature_lens, dtype=torch.long, device=image_features[0].device)
+        return new_image_features, feature_lens
+
+    @merge_with_config_defaults
+    @can_return_tuple
+    @auto_docstring(
+        custom_intro="Obtains image last hidden states from the vision tower and apply multimodal projection."
+    )
+    def get_image_features(
+        self,
+        pixel_values: torch.FloatTensor,
+        image_sizes: torch.Tensor,
+        vision_feature_layer: int | list[int] | list[int] | None = None,
+        vision_feature_select_strategy: str | None = None,
+        vision_aspect_ratio: str | None = None,
+        batch_num_images: torch.LongTensor | None = None,
+        output_hidden_states: bool | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple | BaseModelOutputWithPooling:
+        r"""
+        image_sizes (`torch.Tensor` of shape `(num_images, 2)`):
+            Actual image size of each images (H, W).
+        vision_aspect_ratio (`str`, *optional*, defaults to `"anyres_max_9"`):
+            Aspect ratio used when processing image features. The default value is "anyres_max_9".
+        batch_num_images (`torch.LongTensor`, *optional*):
+            Number of images in each sample.
+        """
+        # ! infer image_num_patches from image_sizes
+        if batch_num_images is None:
+            # treat this as a single-image case for backward compatibility
+            need_patching = [True] * len(image_sizes)
+        else:
+            need_patching = [n == 1 for n in batch_num_images for _ in range(n)]
+        image_num_patches = [
+            image_size_to_num_patches(
+                image_size=imsize,
+                grid_pinpoints=self.config.image_grid_pinpoints,
+                patch_size=self.config.vision_config.image_size,
+            )
+            if should_patch
+            else 1
+            for imsize, should_patch in zip(image_sizes, need_patching)
+        ]
+        if pixel_values.dim() == 5:
+            # stacked if input is (batch_size, num_patches, num_channels, height, width)
+            _pixel_values_list = [pix_val[:num_patch] for pix_val, num_patch in zip(pixel_values, image_num_patches)]
+            pixel_values = torch.cat(_pixel_values_list, dim=0)
+        elif pixel_values.dim() != 4:
+            # otherwise has to be stacked from list of (num_patches, num_channels, height, width)
+            raise ValueError(f"pixel_values of shape {pixel_values.shape}, expect to be of 4 or 5 dimensions")
+
+        image_outputs = self.vision_tower(
+            pixel_values,
+            output_hidden_states=True,  # Ignore arg on purpose
+            return_dict=True,
+            **kwargs,
+        )
+        # If we have one vision feature layer, return the corresponding hidden states,
+        # otherwise, select the hidden states of each feature layer and concatenate them
+        if isinstance(vision_feature_layer, int):
+            selected_image_feature = image_outputs.hidden_states[vision_feature_layer]
+        else:
+            hs_pool = [image_outputs.hidden_states[layer_idx] for layer_idx in vision_feature_layer]
+            selected_image_feature = torch.cat(hs_pool, dim=-1)
+
+        if vision_feature_select_strategy == "default":
+            selected_image_feature = selected_image_feature[:, 1:]
+        image_features = self.multi_modal_projector(selected_image_feature)
+        image_features = torch.split(image_features, image_num_patches, dim=0)
+
+        image_features, feature_lens = self.pack_image_features(
+            image_features,
+            image_sizes,
+            image_newline=self.image_newline,
+            vision_aspect_ratio=vision_aspect_ratio,
+        )
+        image_outputs.pooler_output = image_features
+
+        return image_outputs
+
+    def get_placeholder_mask(
+        self,
+        input_ids: torch.LongTensor,
+        inputs_embeds: torch.FloatTensor,
+        image_features: torch.FloatTensor | None = None,
+        video_features: torch.FloatTensor | None = None,
+    ):
+        """
+        Obtains multimodal placeholder mask from `input_ids` or `inputs_embeds`, and checks that the placeholder token count is
+        equal to the length of multimodal features. If the lengths are different, an error is raised.
+        """
+        if input_ids is None:
+            special_image_mask = inputs_embeds == self.get_input_embeddings()(
+                torch.tensor(self.config.image_token_id, dtype=torch.long, device=inputs_embeds.device)
+            )
+            special_image_mask = special_image_mask.all(-1)
+            special_video_mask = inputs_embeds == self.get_input_embeddings()(
+                torch.tensor(self.config.video_token_id, dtype=torch.long, device=inputs_embeds.device)
+            )
+            special_video_mask = special_video_mask.all(-1)
+        else:
+            special_image_mask = input_ids == self.config.image_token_id
+            special_video_mask = input_ids == self.config.video_token_id
+
+        n_image_tokens = special_image_mask.sum()
+        special_image_mask = special_image_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
+        if image_features is not None:
+            torch_compilable_check(
+                inputs_embeds[special_image_mask].numel() == image_features.numel(),
+                f"Image features and image tokens do not match, tokens: {n_image_tokens}, features: {image_features.shape[0]}",
+            )
+
+        n_video_tokens = special_video_mask.sum()
+        special_video_mask = special_video_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
+        if video_features is not None:
+            torch_compilable_check(
+                inputs_embeds[special_video_mask].numel() == video_features.numel(),
+                f"Video features and video tokens do not match, tokens: {n_video_tokens}, features: {video_features.shape[0]}",
+            )
+        return special_image_mask, special_video_mask
+
+    @merge_with_config_defaults
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: torch.LongTensor | None = None,
+        pixel_values: torch.FloatTensor | None = None,
+        image_sizes: torch.LongTensor | None = None,
+        pixel_values_videos: torch.FloatTensor | None = None,
+        image_sizes_videos: torch.LongTensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        position_ids: torch.LongTensor | None = None,
+        past_key_values: Cache | None = None,
+        inputs_embeds: torch.FloatTensor | None = None,
+        vision_feature_layer: int | list[int] | list[int] | None = None,
+        vision_feature_select_strategy: str | None = None,
+        vision_aspect_ratio: str | None = None,
+        batch_num_images: torch.LongTensor | None = None,
+        use_cache: bool | None = None,
+        **kwargs: Unpack[FlashAttentionKwargs],
+    ) -> tuple | LlavaOnevisionModelOutputWithPast:
+        r"""
+        image_sizes_videos (`torch.LongTensor` of shape `(batch_size, frames, 2)`, *optional*):
+            The sizes of the videos in the batch, being (height, width) for each frame in the video.
+        vision_aspect_ratio (`str`, *optional*, defaults to `"anyres_max_9"`):
+            Aspect ratio used when processing image features. The default value is "anyres_max_9".
+        batch_num_images (`torch.LongTensor`, *optional*):
+            Number of images in each sample.
+        """
+        if (input_ids is None) ^ (inputs_embeds is not None):
+            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
+
+        if inputs_embeds is None:
+            inputs_embeds = self.get_input_embeddings()(input_ids)
+
+        image_features = None
+        if pixel_values is not None:
+            image_features = self.get_image_features(
+                pixel_values,
+                image_sizes,
+                vision_feature_layer=vision_feature_layer,
+                vision_feature_select_strategy=vision_feature_select_strategy,
+                batch_num_images=batch_num_images,
+                return_dict=True,
+            ).pooler_output
+            image_features = torch.cat(image_features, dim=0)
+            image_features = image_features.to(inputs_embeds.device, inputs_embeds.dtype)
+            special_image_mask, _ = self.get_placeholder_mask(
+                input_ids, inputs_embeds=inputs_embeds, image_features=image_features
+            )
+            inputs_embeds = inputs_embeds.masked_scatter(special_image_mask, image_features)
+
+        video_features = None
+        if pixel_values_videos is not None:
+            video_features = self.get_video_features(
+                pixel_values_videos,
+                vision_feature_layer=vision_feature_layer,
+                vision_feature_select_strategy=vision_feature_select_strategy,
+                return_dict=True,
+            ).pooler_output
+            image_newline = (
+                self.image_newline[None, None, :].repeat(video_features.shape[0], 1, 1).to(video_features.device)
+            )
+            video_features = torch.cat((video_features, image_newline), dim=1)
+            video_features = video_features.flatten(0, 1).to(inputs_embeds.device, inputs_embeds.dtype)
+            _, special_video_mask = self.get_placeholder_mask(
+                input_ids, inputs_embeds=inputs_embeds, video_features=video_features
+            )
+            inputs_embeds = inputs_embeds.masked_scatter(special_video_mask, video_features)
+
+        outputs = self.language_model(
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            **kwargs,
+        )
+
+        return LlavaOnevisionModelOutputWithPast(
+            last_hidden_state=outputs.last_hidden_state,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+            image_hidden_states=image_features,
+            video_hidden_states=video_features,
+        )
+
+    @merge_with_config_defaults
+    @can_return_tuple
+    @auto_docstring(
+        custom_intro="Obtains video last hidden states from the vision tower, apply multimodal projection and pooling."
+    )
+    def get_video_features(
+        self,
+        pixel_values: torch.FloatTensor,
+        vision_feature_layer: int | list[int] | list[int] | None = None,
+        vision_feature_select_strategy: str | None = None,
+        output_hidden_states: bool | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple | BaseModelOutputWithPooling:
+        r"""
+        pixel_values (`torch.FloatTensor]` of shape `(batch_size, num_frames, channels, height, width)`)
+            The tensors corresponding to the input video.
+        vision_feature_layer (`Union[int, list[int]], *optional*, defaults to -2`):
+            The index of the layer to select the vision feature. If multiple indices are provided,
+            the vision feature of the corresponding indices will be concatenated to form the
+            vision features.
+        vision_feature_select_strategy (`str`):
+            The feature selection strategy used to select the vision feature from the vision backbone.
+            Can be one of `"default"` or `"full"`
+        """
+        batch_size, frames, channels, height, width = pixel_values.shape
+        pixel_values = pixel_values.view(batch_size * frames, channels, height, width)
+        vision_outputs = self.vision_tower(
+            pixel_values,
+            output_hidden_states=True,  # Ignore arg on purpose
+            return_dict=True,
+            **kwargs,
+        )
+
+        # If we have one vision feature layer, return the corresponding hidden states,
+        # otherwise, select the hidden states of each feature layer and concatenate them
+        if isinstance(vision_feature_layer, int):
+            selected_video_feature = vision_outputs.hidden_states[vision_feature_layer]
+        else:
+            hs_pool = [vision_outputs.hidden_states[layer_idx] for layer_idx in vision_feature_layer]
+            selected_video_feature = torch.cat(hs_pool, dim=-1)
+
+        if vision_feature_select_strategy == "default":
+            selected_video_feature = selected_video_feature[:, 1:]
+        video_features = self.multi_modal_projector(selected_video_feature)
+
+        video_features = self.apply_pooling(video_features)
+        video_features = video_features.reshape(batch_size, frames * video_features.shape[1], -1)
+        vision_outputs.pooler_output = video_features
+
+        return vision_outputs
+
+    def apply_pooling(self, image_features):
+        height = width = self.config.vision_config.image_size // self.config.vision_config.patch_size
+        batch_frames, seq_len, dim = image_features.shape
+        image_features = image_features.view(batch_frames, height, width, -1)
+        image_features = image_features.permute(0, 3, 1, 2).contiguous()
+
+        height, width = image_features.shape[2:]
+        scaled_shape = [math.ceil(height / 2), math.ceil(width / 2)]
+        image_features = nn.functional.interpolate(image_features, size=scaled_shape, mode="bilinear")
+
+        image_features = image_features.permute(0, 2, 3, 1)
+        image_features = image_features.view(batch_frames, -1, dim)
+        return image_features
+
+
+@auto_docstring(
+    custom_intro="""
+    The LLAVA-NeXT model which consists of a vision backbone and a language model.
+    """
+)
+class LlavaOnevisionForConditionalGeneration(LlavaOnevisionPreTrainedModel, GenerationMixin):
+    _tied_weights_keys = {"lm_head.weight": "model.language_model.embed_tokens.weight"}
+
+    def __init__(self, config: LlavaOnevisionConfig):
+        super().__init__(config)
+        self.model = LlavaOnevisionModel(config)
+        self.lm_head = nn.Linear(config.text_config.hidden_size, config.text_config.vocab_size, bias=False)
+        self.post_init()
+
+    def get_output_embeddings(self) -> nn.Module:
+        return self.lm_head
+
+    def pack_image_features(self, image_features, image_sizes, vision_feature_select_strategy, image_newline=None):
+        return self.model.pack_image_features(
+            image_features=image_features,
+            image_sizes=image_sizes,
+            vision_feature_select_strategy=vision_feature_select_strategy,
+            image_newline=image_newline,
+        )
+
+    @merge_with_config_defaults
+    @can_return_tuple
+    @auto_docstring
+    def get_image_features(
+        self,
+        pixel_values: torch.FloatTensor,
+        image_sizes: torch.Tensor,
+        vision_feature_layer: int | list[int] | list[int] | None = None,
+        vision_feature_select_strategy: str | None = None,
+        vision_aspect_ratio: str | None = None,
+        batch_num_images: torch.LongTensor | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple | BaseModelOutputWithPooling:
+        r"""
+        image_sizes (`torch.Tensor` of shape `(num_images, 2)`):
+            Actual image size of each images (H, W).
+        vision_aspect_ratio (`str`, *optional*, defaults to `"anyres_max_9"`):
+            Aspect ratio used when processing image features. The default value is "anyres_max_9".
+        batch_num_images (`torch.LongTensor`, *optional*):
+            Number of images in each sample.
+        """
+        return self.model.get_image_features(
+            pixel_values=pixel_values,
+            image_sizes=image_sizes,
+            vision_feature_layer=vision_feature_layer,
+            vision_feature_select_strategy=vision_feature_select_strategy,
+            vision_aspect_ratio=vision_aspect_ratio,
+            batch_num_images=batch_num_images,
+            **kwargs,
+        )
+
+    @merge_with_config_defaults
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: torch.LongTensor | None = None,
+        pixel_values: torch.FloatTensor | None = None,
+        image_sizes: torch.LongTensor | None = None,
+        pixel_values_videos: torch.FloatTensor | None = None,
+        image_sizes_videos: torch.LongTensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        position_ids: torch.LongTensor | None = None,
+        past_key_values: Cache | None = None,
+        inputs_embeds: torch.FloatTensor | None = None,
+        vision_feature_layer: int | list[int] | list[int] | None = None,
+        vision_feature_select_strategy: str | None = None,
+        vision_aspect_ratio: str | None = None,
+        batch_num_images: torch.LongTensor | None = None,
+        labels: torch.LongTensor | None = None,
+        use_cache: bool | None = None,
+        logits_to_keep: int | torch.Tensor = 0,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple | LlavaOnevisionCausalLMOutputWithPast:
+        r"""
+        image_sizes_videos (`torch.LongTensor` of shape `(batch_size, frames, 2)`, *optional*):
+            The sizes of the videos in the batch, being (height, width) for each frame in the video.
+        vision_aspect_ratio (`str`, *optional*, defaults to `"anyres_max_9"`):
+            Aspect ratio used when processing image features. The default value is "anyres_max_9".
+        batch_num_images (`torch.LongTensor`, *optional*):
+            Number of images in each sample.
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        Example:
+
+        ```python
+        >>> from PIL import Image
+        >>> import httpx
+        >>> from io import BytesIO
+        >>> import torch
+        >>> from transformers import LlavaOnevisionProcessor, LlavaOnevisionForConditionalGeneration
+
+        >>> model = LlavaOnevisionForConditionalGeneration.from_pretrained("llava-hf/llava-onevision-qwen2-7b-ov-hf", dtype="float16", device_map="cuda:0")
+        >>> processor = LlavaOnevisionProcessor.from_pretrained("llava-hf/llava-onevision-qwen2-7b-ov-hf")
+
+        >>> conversation = [
+        ...     {
+        ...       "role": "user",
+        ...       "content": [
+        ...           {"type": "text", "text": "What is shown in this image?"},
+        ...           {"type": "image"},
+        ...         ],
+        ...     },
+        ... ]
+        >>> prompt = processor.apply_chat_template(conversation, add_generation_prompt=True)
+
+        >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+        >>> with httpx.stream("GET", url) as response:
+        ...     image = Image.open(BytesIO(response.read()))
+        >>> inputs = processor(text=prompt, images=image, return_tensors='pt').to(0, torch.float16)
+
+        >>> output = model.generate(**inputs, max_new_tokens=20, do_sample=False)
+        >>> processor.batch_decode(output, skip_special_tokens=True)[0]
+        "user\n\nWhat is shown in this image?\nassistant\ncat"
+        ```"""
+        outputs = self.model(
+            input_ids=input_ids,
+            pixel_values=pixel_values,
+            pixel_values_videos=pixel_values_videos,
+            image_sizes=image_sizes,
+            image_sizes_videos=image_sizes_videos,
+            vision_aspect_ratio=vision_aspect_ratio,
+            vision_feature_layer=vision_feature_layer,
+            vision_feature_select_strategy=vision_feature_select_strategy,
+            batch_num_images=batch_num_images,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            logits_to_keep=logits_to_keep,
+            **kwargs,
+        )
+
+        hidden_states = outputs[0]
+        # Only compute necessary logits, and do not upcast them to float if we are not computing the loss
+        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
+        logits = self.lm_head(hidden_states[:, slice_indices, :])
+
+        loss = None
+        if labels is not None:
+            loss = self.loss_function(
+                logits=logits, labels=labels, vocab_size=self.config.text_config.vocab_size, **kwargs
+            )
+
+        return LlavaOnevisionCausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+            image_hidden_states=outputs.image_hidden_states,
+            video_hidden_states=outputs.video_hidden_states,
+        )
+
+    def prepare_inputs_for_generation(
+        self,
+        input_ids,
+        past_key_values=None,
+        inputs_embeds=None,
+        pixel_values=None,
+        image_sizes=None,
+        pixel_values_videos=None,
+        image_sizes_videos=None,
+        attention_mask=None,
+        logits_to_keep=None,
+        is_first_iteration=False,
+        **kwargs,
+    ):
+        # Overwritten -- in specific circumstances we don't want to forward image inputs to the model
+
+        model_inputs = super().prepare_inputs_for_generation(
+            input_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            attention_mask=attention_mask,
+            logits_to_keep=logits_to_keep,
+            is_first_iteration=is_first_iteration,
+            **kwargs,
+        )
+
+        if is_first_iteration or not kwargs.get("use_cache", True):
+            # Pixel values are used only in the first iteration if available
+            # In subsequent iterations, they are already merged with text and cached
+            # NOTE: first iteration doesn't have to be prefill, it can be the first
+            # iteration with a question and cached system prompt (continue generate from cache)
+            model_inputs["pixel_values"] = pixel_values
+            model_inputs["image_sizes"] = image_sizes
+            model_inputs["pixel_values_videos"] = pixel_values_videos
+            model_inputs["image_sizes_videos"] = image_sizes_videos
+
+        return model_inputs
+
+    @merge_with_config_defaults
+    @can_return_tuple
+    @auto_docstring
+    def get_video_features(
+        self,
+        pixel_values: torch.FloatTensor,
+        vision_feature_layer: int | list[int] | list[int] | None = None,
+        vision_feature_select_strategy: str | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple | BaseModelOutputWithPooling:
+        r"""
+        pixel_values (`torch.FloatTensor]` of shape `(batch_size, num_frames, channels, height, width)`)
+            The tensors corresponding to the input video.
+        vision_feature_layer (`Union[int, list[int]]`, *optional;*):
+            The index of the layer to select the vision feature. If multiple indices are provided,
+            the vision feature of the corresponding indices will be concatenated to form the
+            vision features.
+        vision_feature_select_strategy (`str`, *optional*):
+            The feature selection strategy used to select the vision feature from the vision backbone.
+            Can be one of `"default"` or `"full"`
+        """
+        return self.model.get_video_features(
+            pixel_values=pixel_values,
+            vision_feature_layer=vision_feature_layer,
+            vision_feature_select_strategy=vision_feature_select_strategy,
+            **kwargs,
+        )
+
+
+__all__ = ["LlavaOnevisionModel", "LlavaOnevisionForConditionalGeneration", "LlavaOnevisionPreTrainedModel"]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/processing_llava_onevision.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/processing_llava_onevision.py
new file mode 100644
index 0000000000000000000000000000000000000000..ca5f6e3a5bd1c4338aa425438f25cdba7bea61c5
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/processing_llava_onevision.py
@@ -0,0 +1,290 @@
+# Copyright 2024 The HuggingFace Inc. team.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+Processor class for LLaVa-Onevision.
+"""
+
+import math
+from collections.abc import Iterable
+
+import numpy as np
+
+from ...feature_extraction_utils import BatchFeature
+from ...image_processing_utils import select_best_resolution
+from ...image_utils import ImageInput, get_image_size, to_numpy_array
+from ...processing_utils import MultiModalData, ProcessingKwargs, ProcessorMixin, Unpack
+from ...tokenization_utils_base import PreTokenizedInput, TextInput
+from ...utils import auto_docstring, logging
+from ...video_utils import VideoInput
+
+
+logger = logging.get_logger(__name__)
+
+
+class LlavaOnevisionProcessorKwargs(ProcessingKwargs, total=False):
+    # see processing_utils.ProcessingKwargs documentation for usage.
+    _defaults = {
+        "text_kwargs": {
+            "padding": False,
+            "return_mm_token_type_ids": False,
+        },
+        "image_kwargs": {},
+    }
+
+
+@auto_docstring
+class LlavaOnevisionProcessor(ProcessorMixin):
+    def __init__(
+        self,
+        image_processor=None,
+        tokenizer=None,
+        video_processor=None,
+        num_image_tokens=None,
+        vision_feature_select_strategy=None,
+        chat_template=None,
+        image_token="<image>",
+        video_token="<video>",
+        vision_aspect_ratio="anyres_max_9",
+        **kwargs,
+    ):
+        r"""
+        num_image_tokens (`int`, *optional*):
+            Number of image tokens for one imagethat will be returned by vision tower.
+        vision_feature_select_strategy (`str`, *optional*):
+            The feature selection strategy used to select the vision feature from the vision backbone.
+            Should be same as in model's config
+        image_token (`str`, *optional*, defaults to `"<image>"`):
+            Special token used to denote image location.
+        video_token (`str`, *optional*, defaults to `"<video>"`):
+            Special token used to denote video location.
+        vision_aspect_ratio (`str`, *optional*, defaults to `"anyres_max_9"`):
+            Aspect ratio used when processong image features. The default value is "anyres_max_9".
+        """
+        self.num_image_tokens = num_image_tokens
+        self.vision_feature_select_strategy = vision_feature_select_strategy
+        self.image_token = tokenizer.image_token if hasattr(tokenizer, "image_token") else image_token
+        self.video_token = tokenizer.video_token if hasattr(tokenizer, "video_token") else video_token
+        self.image_token_id = (
+            tokenizer.image_token_id
+            if getattr(tokenizer, "image_token_id", None)
+            else tokenizer.convert_tokens_to_ids(self.image_token)
+        )
+        self.video_token_id = (
+            tokenizer.video_token_id
+            if getattr(tokenizer, "video_token_id", None)
+            else tokenizer.convert_tokens_to_ids(self.video_token)
+        )
+        self.vision_aspect_ratio = vision_aspect_ratio
+        super().__init__(image_processor, tokenizer, video_processor, chat_template=chat_template)
+
+    @auto_docstring
+    def __call__(
+        self,
+        images: ImageInput | None = None,
+        text: TextInput | PreTokenizedInput | list[TextInput] | list[PreTokenizedInput] = None,
+        videos: VideoInput | None = None,
+        **kwargs: Unpack[LlavaOnevisionProcessorKwargs],
+    ) -> BatchFeature:
+        r"""
+        Returns:
+            [`BatchFeature`]: A [`BatchFeature`] with the following fields:
+
+            - **input_ids** -- List of token ids to be fed to a model. Returned when `text` is not `None`.
+            - **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when
+              `return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not
+              `None`).
+            - **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`.
+            - **pixel_values_videos** -- Pixel values of a video input to be fed to a model. Returned when `videos` is not `None`.
+            - **image_sizes** -- Size of each image that will be used to unpad an image. Returned when `images` is not `None`.
+        """
+
+        output_kwargs = self._merge_kwargs(
+            LlavaOnevisionProcessorKwargs,
+            tokenizer_init_kwargs=self.tokenizer.init_kwargs,
+            **kwargs,
+        )
+
+        if isinstance(text, str):
+            text = [text]
+        elif not isinstance(text, list) and not isinstance(text[0], str):
+            raise TypeError("Invalid input text. Please provide a string, or a list of strings")
+
+        image_inputs = video_inputs = {}
+
+        if images is not None:
+            image_inputs = self.image_processor(images, **output_kwargs["images_kwargs"])
+
+            batch_num_images = iter(image_inputs["batch_num_images"])
+            image_sizes = iter(image_inputs["image_sizes"])
+            height, width = get_image_size(
+                to_numpy_array(image_inputs["pixel_values"][0][0]),
+                channel_dim=output_kwargs["images_kwargs"].get("data_format"),
+            )
+            text, num_image_tokens = self._expand_image_tokens(
+                text, image_sizes, height, width, self.image_token, batch_num_images
+            )
+
+        if videos is not None:
+            video_inputs = self.video_processor(videos, **output_kwargs["videos_kwargs"])
+
+            one_video = video_inputs.get("pixel_values_videos")[0]
+            if isinstance(video_inputs.get("pixel_values_videos")[0], (list, tuple)):
+                one_video = np.array(one_video)
+            else:
+                one_video = to_numpy_array(one_video)
+            height, width = get_image_size(one_video[0], channel_dim=output_kwargs["images_kwargs"].get("data_format"))
+            num_frames = one_video.shape[0]  # frame dim is always after batch dim
+            patches_height_width = int(math.sqrt(self.num_image_tokens))
+            pooled_height_width = math.ceil(patches_height_width / 2)
+            num_video_tokens = (num_frames * pooled_height_width * pooled_height_width) + 1  # +1 for newline token
+            text = [sample.replace(self.video_token, self.video_token * num_video_tokens) for sample in text]
+
+        return_tensors = output_kwargs["text_kwargs"].pop("return_tensors", None)
+        return_mm_token_type_ids = output_kwargs["text_kwargs"].pop("return_mm_token_type_ids", None)
+        text_inputs = self.tokenizer(text, **output_kwargs["text_kwargs"])
+        self._check_special_mm_tokens(text, text_inputs, modalities=["image"])
+
+        if return_mm_token_type_ids:
+            text_inputs["mm_token_type_ids"] = self.create_mm_token_type_ids(text_inputs["input_ids"])
+        return BatchFeature(data={**text_inputs, **image_inputs, **video_inputs}, tensor_type=return_tensors)
+
+    def _expand_image_tokens(
+        self,
+        text: list[TextInput],
+        image_sizes: Iterable[list[int] | int],
+        height: int,
+        width: int,
+        special_token: str,
+        batch_num_images: Iterable[int],
+    ):
+        prompt_strings = []
+        max_num_vision_tokens = 0
+        for sample in text:
+            if special_token in sample:
+                num_images = next(batch_num_images)  # should consume iterable
+                is_multi_image = num_images != 1
+            else:
+                is_multi_image = False
+            while special_token in sample:
+                original_size = next(image_sizes)  # should consume iterable
+                if is_multi_image:
+                    num_image_tokens = self.num_image_tokens + 1  # one for image_newline
+                else:
+                    if not isinstance(original_size, (list, tuple)):
+                        # cast to list to avoid numerical precision errors when calculating unpadding
+                        original_size = original_size.tolist()
+                    orig_height, orig_width = original_size
+                    num_image_tokens = self._get_number_of_features(orig_height, orig_width, height, width)
+                max_num_vision_tokens = max(max_num_vision_tokens, num_image_tokens)
+                if self.vision_feature_select_strategy == "default":
+                    num_image_tokens -= 1
+                sample = sample.replace(special_token, "<placeholder>" * num_image_tokens, 1)
+            prompt_strings.append(sample)
+        text = [sample.replace("<placeholder>", special_token) for sample in prompt_strings]
+        return text, max_num_vision_tokens
+
+    def _get_number_of_features(self, orig_height: int, orig_width: int, height: int, width: int) -> int:
+        image_grid_pinpoints = self.image_processor.image_grid_pinpoints
+
+        height_best_resolution, width_best_resolution = select_best_resolution(
+            [orig_height, orig_width], image_grid_pinpoints
+        )
+        scale_height, scale_width = height_best_resolution // height, width_best_resolution // width
+
+        patches_height = patches_width = int(math.sqrt(self.num_image_tokens))
+        unpadded_features, newline_features = self._get_unpadded_features(
+            orig_height, orig_width, patches_height, patches_width, scale_height, scale_width
+        )
+
+        # The base patch covers the entire image (no CLS for SigLIP)
+        base_features = self.num_image_tokens
+        num_image_tokens = unpadded_features + newline_features + base_features
+        return num_image_tokens
+
+    # Adapted from transformers.models.llava_next.processing_llava_next.LlavaNextProcessor._get_unpadded_features
+    def _get_unpadded_features(self, height, width, patches_height, patches_width, scale_height, scale_width):
+        """
+        Get number of features for a given image with height/width. LLaVA-NeXT is different from LLaVA
+        because it divided each image into patches depending on its resolution. Therefore we need to calculate how many
+        patches an image is divided into and get the number of features from that.
+        """
+        current_height = patches_height * scale_height
+        current_width = patches_width * scale_width
+
+        original_aspect_ratio = width / height
+        current_aspect_ratio = current_width / current_height
+        if original_aspect_ratio > current_aspect_ratio:
+            new_height = int(round(height * (current_width / width), 7))
+            padding = (current_height - new_height) // 2
+            current_height -= padding * 2
+        else:
+            new_width = int(round(width * (current_height / height), 7))
+            padding = (current_width - new_width) // 2
+            current_width -= padding * 2
+
+        unpadded_features = current_height * current_width
+        newline_features = current_height
+
+        max_num_patches = int(self.vision_aspect_ratio.strip("anyres_max_"))
+        ratio = math.sqrt(current_height * current_width / (max_num_patches * patches_height**2))
+        if ratio > 1.1:
+            unpadded_features = int(current_height // ratio) * int(current_width // ratio)
+            newline_features = int(current_height // ratio)
+
+        return (unpadded_features, newline_features)
+
+    def _get_num_multimodal_tokens(self, image_sizes=None, video_sizes=None, **kwargs):
+        """
+        Computes the number of placeholder tokens needed for multimodal inputs with the given sizes.
+        Args:
+            image_sizes (list[list[str]], *optional*):
+                The input sizes formatted as (height, width) per each image.
+            video_sizes (list[list[str]], *optional*):
+                The input sizes formatted as (num_frames, height, width) per each video.
+            audio_lengths (list[int], *optional*):
+                The input length formatted as per each audio.
+        Returns:
+            dict[str, list[int]]: A dictionary mapping each modality ("image", "video", "audio")
+            to a list containing the number of placeholder tokens required. If the model doesn't accept
+            a certain modality or no input sizes are provided, the dict value is set to an empty list.
+        """
+        vision_data = {}
+        if image_sizes is not None:
+            images_kwargs = LlavaOnevisionProcessorKwargs._defaults.get("images_kwargs", {})
+            images_kwargs.update(kwargs)
+
+            size = images_kwargs.get("size", None) or self.image_processor.size
+            size = (
+                (size["shortest_edge"], size["shortest_edge"])
+                if "shortest_edge" in size
+                else (min(size["height"], size["width"]), min(size["height"], size["width"]))
+            )
+            processed_height, processed_width = size
+
+            batch_num_image_tokens = []
+            num_image_patches = [1] * len(image_sizes)  # llava-ov doesn't batch pixels as Idefics, thus `1` patch`
+            for image_size in image_sizes:
+                orig_height, orig_width = image_size
+                num_image_tokens = self._get_number_of_features(
+                    orig_height, orig_width, processed_height, processed_width
+                )
+                if self.vision_feature_select_strategy == "default":
+                    num_image_tokens -= 1
+                batch_num_image_tokens.append(num_image_tokens)
+            vision_data.update({"num_image_tokens": batch_num_image_tokens, "num_image_patches": num_image_patches})
+
+        return MultiModalData(**vision_data)
+
+
+__all__ = ["LlavaOnevisionProcessor"]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/video_processing_llava_onevision.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/video_processing_llava_onevision.py
new file mode 100644
index 0000000000000000000000000000000000000000..6d80dad0cb38ddd66449d8d6ce5edd17c6e182ee
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/llava_onevision/video_processing_llava_onevision.py
@@ -0,0 +1,36 @@
+# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Video processor class for LLaVa-Onevision."""
+
+from ...image_utils import OPENAI_CLIP_MEAN, OPENAI_CLIP_STD, PILImageResampling
+from ...video_processing_utils import BaseVideoProcessor
+
+
+class LlavaOnevisionVideoProcessor(BaseVideoProcessor):
+    resample = PILImageResampling.BICUBIC
+    image_mean = OPENAI_CLIP_MEAN
+    image_std = OPENAI_CLIP_STD
+    size = {"height": 384, "width": 384}
+    rescale_factor = 1 / 255
+    default_to_square = False
+    crop_size = None
+    do_resize = True
+    do_center_crop = None
+    do_rescale = True
+    do_normalize = True
+    do_convert_rgb = True
+    do_sample_frames = False  # Set to False for BC, recommended to set `True` in new models
+
+
+__all__ = ["LlavaOnevisionVideoProcessor"]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/pegasus/configuration_pegasus.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/pegasus/configuration_pegasus.py
new file mode 100644
index 0000000000000000000000000000000000000000..b4a3c87b54ab05fc4a6d51f3dc7b60030718f4bb
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/pegasus/configuration_pegasus.py
@@ -0,0 +1,76 @@
+# Copyright 2021, Google and The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""PEGASUS model configuration"""
+
+from huggingface_hub.dataclasses import strict
+
+from ...configuration_utils import PreTrainedConfig
+from ...utils import auto_docstring
+
+
+@auto_docstring(checkpoint="google/pegasus-large")
+@strict
+class PegasusConfig(PreTrainedConfig):
+    r"""
+    Example:
+
+    ```python
+    >>> from transformers import PegasusConfig, PegasusModel
+
+    >>> # Initializing a PEGASUS google/pegasus-large style configuration
+    >>> configuration = PegasusConfig()
+
+    >>> # Initializing a model (with random weights) from the google/pegasus-large style configuration
+    >>> model = PegasusModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "pegasus"
+    keys_to_ignore_at_inference = ["past_key_values"]
+    attribute_map = {
+        "num_attention_heads": "encoder_attention_heads",
+        "hidden_size": "d_model",
+        "num_hidden_layers": "encoder_layers",
+    }
+
+    vocab_size: int = 50265
+    max_position_embeddings: int = 1024
+    encoder_layers: int = 12
+    encoder_ffn_dim: int = 4096
+    encoder_attention_heads: int = 16
+    decoder_layers: int = 12
+    decoder_ffn_dim: int = 4096
+    decoder_attention_heads: int = 16
+    encoder_layerdrop: float | int = 0.0
+    decoder_layerdrop: float | int = 0.0
+    use_cache: bool = True
+    is_encoder_decoder: bool = True
+    activation_function: str = "gelu"
+    d_model: int = 1024
+    dropout: float | int = 0.1
+    attention_dropout: float | int = 0.0
+    activation_dropout: float | int = 0.0
+    init_std: float = 0.02
+    decoder_start_token_id: int | None = 0
+    scale_embedding: bool = False
+    pad_token_id: int | None = 0
+    eos_token_id: int | list[int] | None = 1
+    forced_eos_token_id: int | list[int] | None = 1
+    is_decoder: bool = False
+    tie_word_embeddings: bool = True
+
+
+__all__ = ["PegasusConfig"]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/pegasus/modeling_pegasus.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/pegasus/modeling_pegasus.py
new file mode 100644
index 0000000000000000000000000000000000000000..1eed3f26e4e1c798632217c05f009cb7703c26cc
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/pegasus/modeling_pegasus.py
@@ -0,0 +1,1132 @@
+# Copyright 2021, Google and The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""PyTorch PEGASUS model."""
+
+import copy
+import math
+from collections.abc import Callable
+
+import numpy as np
+import torch
+from torch import nn
+from torch.nn import CrossEntropyLoss
+
+from ... import initialization as init
+from ...activations import ACT2FN
+from ...cache_utils import Cache, DynamicCache, EncoderDecoderCache
+from ...generation import GenerationMixin
+from ...masking_utils import create_bidirectional_mask, create_causal_mask
+from ...modeling_flash_attention_utils import FlashAttentionKwargs
+from ...modeling_layers import GradientCheckpointingLayer
+from ...modeling_outputs import (
+    BaseModelOutput,
+    BaseModelOutputWithPastAndCrossAttentions,
+    CausalLMOutputWithCrossAttentions,
+    Seq2SeqLMOutput,
+    Seq2SeqModelOutput,
+)
+from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
+from ...processing_utils import Unpack
+from ...utils import (
+    TransformersKwargs,
+    auto_docstring,
+    can_return_tuple,
+    is_torchdynamo_compiling,
+    logging,
+)
+from ...utils.generic import merge_with_config_defaults
+from ...utils.output_capturing import OutputRecorder, capture_outputs
+from .configuration_pegasus import PegasusConfig
+
+
+logger = logging.get_logger(__name__)
+
+
+# Copied from transformers.models.bart.modeling_bart.shift_tokens_right
+def shift_tokens_right(input_ids: torch.Tensor, pad_token_id: int, decoder_start_token_id: int):
+    """
+    Shift input ids one token to the right.
+    """
+    shifted_input_ids = input_ids.new_zeros(input_ids.shape)
+    shifted_input_ids[:, 1:] = input_ids[:, :-1].clone()
+    shifted_input_ids[:, 0] = decoder_start_token_id
+
+    if pad_token_id is None:
+        raise ValueError("self.model.config.pad_token_id has to be defined.")
+    # replace possible -100 values in labels by `pad_token_id`
+    shifted_input_ids.masked_fill_(shifted_input_ids == -100, pad_token_id)
+
+    return shifted_input_ids
+
+
+# Copied from transformers.models.marian.modeling_marian.MarianSinusoidalPositionalEmbedding with Marian->Pegasus
+class PegasusSinusoidalPositionalEmbedding(nn.Embedding):
+    """This module produces sinusoidal positional embeddings of any length."""
+
+    def __init__(self, num_positions: int, embedding_dim: int, padding_idx: int | None = None) -> None:
+        super().__init__(num_positions, embedding_dim, _freeze=True)
+
+    def create_weight(self):
+        """
+        Identical to the XLM create_sinusoidal_embeddings except features are not interleaved. The cos features are in
+        the 2nd half of the vector. [dim // 2:]
+        """
+        n_pos, dim = self.weight.shape
+        position_enc = np.array(
+            [[pos / np.power(10000, 2 * (j // 2) / dim) for j in range(dim)] for pos in range(n_pos)]
+        )
+        out = torch.empty(n_pos, dim, dtype=self.weight.dtype, requires_grad=False)
+        sentinel = dim // 2 if dim % 2 == 0 else (dim // 2) + 1
+        out[:, 0:sentinel] = torch.FloatTensor(np.sin(position_enc[:, 0::2]))
+        out[:, sentinel:] = torch.FloatTensor(np.cos(position_enc[:, 1::2]))
+        return out
+
+    @torch.no_grad()
+    def forward(
+        self, input_ids_shape: torch.Size, past_key_values_length: int = 0, position_ids: torch.Tensor | None = None
+    ) -> torch.Tensor:
+        """`input_ids_shape` is expected to be [bsz x seqlen]."""
+        if position_ids is None:
+            bsz, seq_len = input_ids_shape[:2]
+            position_ids = torch.arange(
+                past_key_values_length, past_key_values_length + seq_len, dtype=torch.long, device=self.weight.device
+            )
+        return super().forward(position_ids)
+
+
+# Copied from transformers.models.bert.modeling_bert.eager_attention_forward
+def eager_attention_forward(
+    module: nn.Module,
+    query: torch.Tensor,
+    key: torch.Tensor,
+    value: torch.Tensor,
+    attention_mask: torch.Tensor | None,
+    scaling: float | None = None,
+    dropout: float = 0.0,
+    **kwargs: Unpack[TransformersKwargs],
+):
+    if scaling is None:
+        scaling = query.size(-1) ** -0.5
+
+    # Take the dot product between "query" and "key" to get the raw attention scores.
+    attn_weights = torch.matmul(query, key.transpose(2, 3)) * scaling
+
+    if attention_mask is not None:
+        attn_weights = attn_weights + attention_mask
+
+    attn_weights = nn.functional.softmax(attn_weights, dim=-1)
+    attn_weights = nn.functional.dropout(attn_weights, p=dropout, training=module.training)
+
+    attn_output = torch.matmul(attn_weights, value)
+    attn_output = attn_output.transpose(1, 2).contiguous()
+
+    return attn_output, attn_weights
+
+
+# Copied from transformers.models.bart.modeling_bart.BartAttention with Bart->Pegasus
+class PegasusAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(
+        self,
+        embed_dim: int,
+        num_heads: int,
+        dropout: float = 0.0,
+        is_decoder: bool = False,
+        bias: bool = True,
+        is_causal: bool = False,
+        config: PegasusConfig | None = None,
+        layer_idx: int | None = None,
+    ):
+        super().__init__()
+        self.embed_dim = embed_dim
+        self.num_heads = num_heads
+        self.dropout = dropout
+        self.head_dim = embed_dim // num_heads
+        self.config = config
+
+        if (self.head_dim * num_heads) != self.embed_dim:
+            raise ValueError(
+                f"embed_dim must be divisible by num_heads (got `embed_dim`: {self.embed_dim}"
+                f" and `num_heads`: {num_heads})."
+            )
+        self.scaling = self.head_dim**-0.5
+        self.is_decoder = is_decoder
+        self.is_causal = is_causal
+        self.layer_idx = layer_idx
+        if layer_idx is None and self.is_decoder:
+            logger.warning_once(
+                f"Instantiating a decoder {self.__class__.__name__} without passing `layer_idx` is not recommended and "
+                "will lead to errors during the forward call, if caching is used. Please make sure to provide a `layer_idx` "
+                "when creating this class."
+            )
+
+        self.k_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
+        self.v_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
+        self.q_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
+        self.out_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        key_value_states: torch.Tensor | None = None,
+        past_key_values: Cache | None = None,
+        attention_mask: torch.Tensor | None = None,
+        # TODO: we need a refactor so that the different attention modules can get their specific kwargs
+        # ATM, we have mixed things encoder, decoder, and encoder-decoder attn
+        **kwargs: Unpack[FlashAttentionKwargs],
+    ) -> tuple[torch.Tensor, torch.Tensor | None]:
+        """Input shape: Batch x Time x Channel"""
+
+        # if key_value_states are provided this layer is used as a cross-attention layer
+        # for the decoder
+        is_cross_attention = key_value_states is not None
+
+        # determine input shapes
+        input_shape = hidden_states.shape[:-1]
+
+        hidden_shape = (*input_shape, -1, self.head_dim)
+
+        # get query proj
+        query_states = self.q_proj(hidden_states).view(hidden_shape).transpose(1, 2)
+
+        is_updated = False
+        if past_key_values is not None:
+            if isinstance(past_key_values, EncoderDecoderCache):
+                is_updated = past_key_values.is_updated.get(self.layer_idx)
+                if is_cross_attention:
+                    # after the first generated id, we can subsequently re-use all key/value_states from cache
+                    curr_past_key_values = past_key_values.cross_attention_cache
+                else:
+                    curr_past_key_values = past_key_values.self_attention_cache
+            else:
+                curr_past_key_values = past_key_values
+
+        current_states = key_value_states if is_cross_attention else hidden_states
+        if is_cross_attention and past_key_values is not None and is_updated:
+            # reuse k,v, cross_attentions
+            key_states = curr_past_key_values.layers[self.layer_idx].keys
+            value_states = curr_past_key_values.layers[self.layer_idx].values
+        else:
+            key_states = self.k_proj(current_states)
+            value_states = self.v_proj(current_states)
+            kv_shape = (*current_states.shape[:-1], -1, self.head_dim)
+            key_states = key_states.view(kv_shape).transpose(1, 2)
+            value_states = value_states.view(kv_shape).transpose(1, 2)
+
+            if past_key_values is not None:
+                key_states, value_states = curr_past_key_values.update(key_states, value_states, self.layer_idx)
+                # set flag that curr layer for cross-attn is already updated so we can re-use in subsequent calls
+                if is_cross_attention and isinstance(past_key_values, EncoderDecoderCache):
+                    past_key_values.is_updated[self.layer_idx] = True
+
+        attention_interface: Callable = ALL_ATTENTION_FUNCTIONS.get_interface(
+            self.config._attn_implementation, eager_attention_forward
+        )
+
+        attn_output, attn_weights = attention_interface(
+            self,
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            dropout=0.0 if not self.training else self.dropout,
+            scaling=self.scaling,
+            **kwargs,
+        )
+
+        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
+        attn_output = self.out_proj(attn_output)
+
+        return attn_output, attn_weights
+
+
+# Copied from transformers.models.mbart.modeling_mbart.MBartEncoderLayer with MBart->Pegasus, MBART->PEGASUS
+class PegasusEncoderLayer(GradientCheckpointingLayer):
+    def __init__(self, config: PegasusConfig):
+        super().__init__()
+        self.embed_dim = config.d_model
+
+        self.self_attn = PegasusAttention(
+            embed_dim=self.embed_dim,
+            num_heads=config.encoder_attention_heads,
+            dropout=config.attention_dropout,
+            config=config,
+        )
+        self.self_attn_layer_norm = nn.LayerNorm(self.embed_dim)
+        self.dropout = config.dropout
+        self.activation_fn = ACT2FN[config.activation_function]
+        self.activation_dropout = config.activation_dropout
+        self.fc1 = nn.Linear(self.embed_dim, config.encoder_ffn_dim)
+        self.fc2 = nn.Linear(config.encoder_ffn_dim, self.embed_dim)
+        self.final_layer_norm = nn.LayerNorm(self.embed_dim)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: torch.Tensor,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> torch.Tensor:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`): attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
+        """
+        residual = hidden_states
+        hidden_states = self.self_attn_layer_norm(hidden_states)
+        hidden_states, _ = self.self_attn(
+            hidden_states=hidden_states,
+            attention_mask=attention_mask,
+            **kwargs,
+        )
+        hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
+        hidden_states = residual + hidden_states
+
+        residual = hidden_states
+        hidden_states = self.final_layer_norm(hidden_states)
+        hidden_states = self.activation_fn(self.fc1(hidden_states))
+        hidden_states = nn.functional.dropout(hidden_states, p=self.activation_dropout, training=self.training)
+        hidden_states = self.fc2(hidden_states)
+        hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
+        hidden_states = residual + hidden_states
+
+        if hidden_states.dtype == torch.float16:
+            clamp_value = torch.finfo(hidden_states.dtype).max - 1000
+            hidden_states = torch.clamp(hidden_states, min=-clamp_value, max=clamp_value)
+
+        return hidden_states
+
+
+# Copied from transformers.models.mbart.modeling_mbart.MBartDecoderLayer with MBart->Pegasus, MBART->PEGASUS
+class PegasusDecoderLayer(GradientCheckpointingLayer):
+    def __init__(self, config: PegasusConfig, layer_idx: int | None = None):
+        super().__init__()
+        self.embed_dim = config.d_model
+
+        self.self_attn = PegasusAttention(
+            embed_dim=self.embed_dim,
+            num_heads=config.decoder_attention_heads,
+            dropout=config.attention_dropout,
+            is_decoder=True,
+            is_causal=True,
+            config=config,
+            layer_idx=layer_idx,
+        )
+        self.dropout = config.dropout
+        self.activation_fn = ACT2FN[config.activation_function]
+        self.activation_dropout = config.activation_dropout
+
+        self.self_attn_layer_norm = nn.LayerNorm(self.embed_dim)
+        self.encoder_attn = PegasusAttention(
+            self.embed_dim,
+            config.decoder_attention_heads,
+            dropout=config.attention_dropout,
+            is_decoder=True,
+            config=config,
+            layer_idx=layer_idx,
+        )
+        self.encoder_attn_layer_norm = nn.LayerNorm(self.embed_dim)
+        self.fc1 = nn.Linear(self.embed_dim, config.decoder_ffn_dim)
+        self.fc2 = nn.Linear(config.decoder_ffn_dim, self.embed_dim)
+        self.final_layer_norm = nn.LayerNorm(self.embed_dim)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: torch.Tensor | None = None,
+        encoder_hidden_states: torch.Tensor | None = None,
+        encoder_attention_mask: torch.Tensor | None = None,
+        past_key_values: Cache | None = None,
+        use_cache: bool | None = True,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> torch.Tensor:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`): attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
+            encoder_hidden_states (`torch.FloatTensor`):
+                cross attention input to the layer of shape `(batch, seq_len, embed_dim)`
+            encoder_attention_mask (`torch.FloatTensor`): encoder attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
+            past_key_values (`Cache`): cached past key and value projection states
+        """
+        residual = hidden_states
+        hidden_states = self.self_attn_layer_norm(hidden_states)
+
+        # Self Attention
+        hidden_states, _ = self.self_attn(
+            hidden_states=hidden_states,
+            past_key_values=past_key_values,
+            attention_mask=attention_mask,
+            **kwargs,
+        )
+        hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
+        hidden_states = residual + hidden_states
+
+        # Cross-Attention Block
+        if encoder_hidden_states is not None:
+            residual = hidden_states
+            hidden_states = self.encoder_attn_layer_norm(hidden_states)
+
+            hidden_states, _ = self.encoder_attn(
+                hidden_states=hidden_states,
+                key_value_states=encoder_hidden_states,
+                attention_mask=encoder_attention_mask,
+                past_key_values=past_key_values,
+                **kwargs,
+            )
+            hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
+            hidden_states = residual + hidden_states
+
+        # Fully Connected
+        residual = hidden_states
+        hidden_states = self.final_layer_norm(hidden_states)
+        hidden_states = self.activation_fn(self.fc1(hidden_states))
+        hidden_states = nn.functional.dropout(hidden_states, p=self.activation_dropout, training=self.training)
+        hidden_states = self.fc2(hidden_states)
+        hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
+        hidden_states = residual + hidden_states
+
+        return hidden_states
+
+
+@auto_docstring
+class PegasusPreTrainedModel(PreTrainedModel):
+    config: PegasusConfig
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _supports_flash_attn = True
+    _supports_sdpa = True
+    _supports_flex_attn = True
+    _can_compile_fullgraph = True
+
+    @torch.no_grad()
+    def _init_weights(self, module):
+        super()._init_weights(module)
+        if isinstance(module, PegasusSinusoidalPositionalEmbedding):
+            init.copy_(module.weight, module.create_weight())
+        elif isinstance(module, PegasusForConditionalGeneration):
+            init.zeros_(module.final_logits_bias)
+
+
+class PegasusEncoder(PegasusPreTrainedModel):
+    """
+    Transformer encoder consisting of *config.encoder_layers* self attention layers. Each layer is a
+    [`PegasusEncoderLayer`].
+
+    Args:
+        config: PegasusConfig
+        embed_tokens (nn.Embedding): output embedding
+    """
+
+    _can_record_outputs = {
+        "hidden_states": PegasusEncoderLayer,
+        "attentions": PegasusAttention,
+    }
+
+    def __init__(self, config: PegasusConfig):
+        super().__init__(config)
+
+        self.dropout = config.dropout
+        self.layerdrop = config.encoder_layerdrop
+
+        embed_dim = config.d_model
+        self.padding_idx = config.pad_token_id
+        self.max_source_positions = config.max_position_embeddings
+        self.embed_scale = math.sqrt(embed_dim) if config.scale_embedding else 1.0
+
+        self.embed_tokens = nn.Embedding(config.vocab_size, embed_dim, self.padding_idx)
+
+        self.embed_positions = PegasusSinusoidalPositionalEmbedding(
+            config.max_position_embeddings,
+            embed_dim,
+            self.padding_idx,
+        )
+        self.layers = nn.ModuleList([PegasusEncoderLayer(config) for _ in range(config.encoder_layers)])
+        self.layer_norm = nn.LayerNorm(config.d_model)
+
+        self.gradient_checkpointing = False
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def resize_position_embeddings(self, new_num_position_embeddings: int):
+        """
+        Resizes position embeddings matrix of the model if `new_num_position_embeddings !=
+        config.max_position_embeddings`.
+
+        Arguments:
+            new_num_position_embeddings (`int`):
+                The number of new position embeddings. If position embeddings are learned, increasing the size will add
+                newly initialized vectors at the end, whereas reducing the size will remove vectors from the end. If
+                position embeddings are not learned (*e.g.* sinusoidal position embeddings), increasing the size will
+                add correct vectors at the end following the position encoding algorithm, whereas reducing the size
+                will remove vectors from the end.
+        """
+        logger.info(f"Setting `config.max_position_embeddings={new_num_position_embeddings}`...")
+        self.config.max_position_embeddings = new_num_position_embeddings
+
+        self.embed_positions = PegasusSinusoidalPositionalEmbedding(
+            self.config.max_position_embeddings,
+            self.config.d_model,
+            self.padding_idx,
+        )
+        init.copy_(self.embed_positions.weight, self.embed_positions.create_weight())
+        self.embed_positions.to(self.device)
+
+    def get_position_embeddings(self) -> nn.Embedding:
+        """
+        Returns the position embeddings matrix
+        """
+        return self.embed_positions
+
+    @merge_with_config_defaults
+    @capture_outputs
+    @auto_docstring
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        inputs_embeds=None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> BaseModelOutput:
+        if (input_ids is None) ^ (inputs_embeds is not None):
+            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale
+
+        input_shape = inputs_embeds.shape[:-1]
+        embed_pos = self.embed_positions(input_shape)
+
+        hidden_states = inputs_embeds + embed_pos
+
+        hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
+
+        attention_mask = create_bidirectional_mask(
+            config=self.config,
+            inputs_embeds=inputs_embeds,
+            attention_mask=attention_mask,
+        )
+
+        for idx, encoder_layer in enumerate(self.layers):
+            # add LayerDrop (see https://huggingface.co/papers/1909.11556 for description)
+            to_drop = False
+            if self.training:
+                dropout_probability = torch.rand([])
+                if dropout_probability < self.layerdrop:  # skip the layer
+                    to_drop = True
+
+            if not to_drop:
+                hidden_states = encoder_layer(
+                    hidden_states,
+                    attention_mask,
+                    **kwargs,
+                )
+
+        hidden_states = self.layer_norm(hidden_states)
+
+        return BaseModelOutput(
+            last_hidden_state=hidden_states,
+        )
+
+
+class PegasusDecoder(PegasusPreTrainedModel):
+    """
+    Transformer decoder consisting of *config.decoder_layers* layers. Each layer is a [`PegasusDecoderLayer`]
+
+    Args:
+        config: PegasusConfig
+        embed_tokens (nn.Embedding): output embedding
+    """
+
+    _can_record_outputs = {
+        "hidden_states": PegasusDecoderLayer,
+        "attentions": OutputRecorder(PegasusAttention, index=1, layer_name="self_attn"),
+        "cross_attentions": OutputRecorder(PegasusAttention, index=1, layer_name="encoder_attn"),
+    }
+
+    def __init__(self, config: PegasusConfig):
+        super().__init__(config)
+        self.dropout = config.dropout
+        self.layerdrop = config.decoder_layerdrop
+        self.padding_idx = config.pad_token_id
+        self.max_target_positions = config.max_position_embeddings
+        self.embed_scale = math.sqrt(config.d_model) if config.scale_embedding else 1.0
+
+        self.embed_tokens = nn.Embedding(config.vocab_size, config.d_model, self.padding_idx)
+
+        self.embed_positions = PegasusSinusoidalPositionalEmbedding(
+            config.max_position_embeddings,
+            config.d_model,
+            self.padding_idx,
+        )
+        self.layers = nn.ModuleList([PegasusDecoderLayer(config, layer_idx=i) for i in range(config.decoder_layers)])
+        self.layer_norm = nn.LayerNorm(config.d_model)
+
+        self.gradient_checkpointing = False
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def resize_position_embeddings(self, new_num_position_embeddings: int):
+        """
+        Resizes position embeddings matrix of the model if `new_num_position_embeddings !=
+        config.max_position_embeddings`.
+
+        Arguments:
+            new_num_position_embeddings (`int`):
+                The number of new position embeddings. If position embeddings are learned, increasing the size will add
+                newly initialized vectors at the end, whereas reducing the size will remove vectors from the end. If
+                position embeddings are not learned (*e.g.* sinusoidal position embeddings), increasing the size will
+                add correct vectors at the end following the position encoding algorithm, whereas reducing the size
+                will remove vectors from the end.
+        """
+        logger.info(f"Setting `config.max_position_embeddings={new_num_position_embeddings}`...")
+        self.config.max_position_embeddings = new_num_position_embeddings
+
+        self.embed_positions = PegasusSinusoidalPositionalEmbedding(
+            self.config.max_position_embeddings,
+            self.config.d_model,
+            self.padding_idx,
+        )
+        init.copy_(self.embed_positions.weight, self.embed_positions.create_weight())
+        self.embed_positions.to(self.device)
+
+    def get_position_embeddings(self) -> nn.Embedding:
+        """
+        Returns the position embeddings matrix
+        """
+        return self.embed_positions
+
+    @merge_with_config_defaults
+    @capture_outputs
+    @auto_docstring
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        past_key_values=None,
+        inputs_embeds=None,
+        use_cache=None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> BaseModelOutputWithPastAndCrossAttentions:
+        if (input_ids is None) ^ (inputs_embeds is not None):
+            raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids)
+
+        # important to apply scale outside of `if` in case users pass `embeds`
+        inputs_embeds = inputs_embeds * self.embed_scale
+
+        # initialize `past_key_values`
+        if use_cache and past_key_values is None:
+            past_key_values = (
+                EncoderDecoderCache(DynamicCache(config=self.config), DynamicCache(config=self.config))
+                if encoder_hidden_states is not None or self.config.is_encoder_decoder
+                else DynamicCache(config=self.config)
+            )
+
+        batch_size, seq_length = inputs_embeds.size()[:-1]
+        past_key_values_length = past_key_values.get_seq_length() if past_key_values is not None else 0
+        position_ids = torch.arange(seq_length, device=inputs_embeds.device) + past_key_values_length
+
+        if attention_mask is None and not is_torchdynamo_compiling():
+            # required mask seq length can be calculated via length of past cache
+            mask_seq_length = past_key_values_length + seq_length
+            attention_mask = torch.ones(batch_size, mask_seq_length, device=inputs_embeds.device)
+
+        self_attn_cache = (
+            past_key_values.self_attention_cache
+            if isinstance(past_key_values, EncoderDecoderCache)
+            else past_key_values
+        )
+
+        causal_mask = create_causal_mask(
+            config=self.config,
+            inputs_embeds=inputs_embeds,
+            attention_mask=attention_mask,
+            past_key_values=self_attn_cache,
+        )
+        encoder_attention_mask = create_bidirectional_mask(
+            config=self.config,
+            inputs_embeds=inputs_embeds,
+            attention_mask=encoder_attention_mask,
+            encoder_hidden_states=encoder_hidden_states,
+        )
+
+        # embed positions
+        positions = self.embed_positions((batch_size, seq_length), past_key_values_length, position_ids=position_ids)
+        hidden_states = inputs_embeds + positions
+        hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
+
+        for idx, decoder_layer in enumerate(self.layers):
+            # add LayerDrop (see https://huggingface.co/papers/1909.11556 for description)
+            if self.training:
+                dropout_probability = torch.rand([])
+                if dropout_probability < self.layerdrop:
+                    continue
+
+            hidden_states = decoder_layer(
+                hidden_states,
+                causal_mask,
+                encoder_hidden_states,  # as a positional argument for gradient checkpointing
+                encoder_attention_mask=encoder_attention_mask,
+                past_key_values=past_key_values,
+                use_cache=use_cache,
+                **kwargs,
+            )
+
+        hidden_states = self.layer_norm(hidden_states)
+
+        return BaseModelOutputWithPastAndCrossAttentions(
+            last_hidden_state=hidden_states,
+            past_key_values=past_key_values,
+        )
+
+
+@auto_docstring
+class PegasusModel(PegasusPreTrainedModel):
+    _tied_weights_keys = {
+        "decoder.embed_tokens.weight": "shared.weight",
+        "encoder.embed_tokens.weight": "shared.weight",
+    }
+
+    def __init__(self, config: PegasusConfig):
+        super().__init__(config)
+
+        padding_idx, vocab_size = config.pad_token_id, config.vocab_size
+        self.shared = nn.Embedding(vocab_size, config.d_model, padding_idx)
+
+        self.encoder = PegasusEncoder(config)
+        self.decoder = PegasusDecoder(config)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.shared
+
+    def set_input_embeddings(self, value):
+        self.shared = value
+        self.encoder.embed_tokens = self.shared
+        self.decoder.embed_tokens = self.shared
+
+    def resize_position_embeddings(self, new_num_position_embeddings: int):
+        """
+        Resizes position embeddings matrix of the model if `new_num_position_embeddings !=
+        config.max_position_embeddings`.
+
+        Arguments:
+            new_num_position_embeddings (`int`):
+                The number of new position embeddings. If position embeddings are learned, increasing the size will add
+                newly initialized vectors at the end, whereas reducing the size will remove vectors from the end. If
+                position embeddings are not learned (*e.g.* sinusoidal position embeddings), increasing the size will
+                add correct vectors at the end following the position encoding algorithm, whereas reducing the size
+                will remove vectors from the end.
+        """
+        self.config.max_position_embeddings = new_num_position_embeddings
+        self.encoder.resize_position_embeddings(new_num_position_embeddings)
+        self.decoder.resize_position_embeddings(new_num_position_embeddings)
+
+    def get_position_embeddings(self) -> tuple[nn.Embedding]:
+        """
+        Returns the position embeddings matrix
+        """
+        return (self.encoder.get_position_embeddings(), self.decoder.get_position_embeddings())
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: torch.Tensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        decoder_input_ids: torch.Tensor | None = None,
+        decoder_attention_mask: torch.Tensor | None = None,
+        encoder_outputs: tuple[torch.FloatTensor] | None = None,
+        past_key_values: Cache | None = None,
+        inputs_embeds: torch.Tensor | None = None,
+        decoder_inputs_embeds: torch.Tensor | None = None,
+        use_cache: bool | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple | Seq2SeqModelOutput:
+        r"""
+        decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
+            Indices of decoder input sequence tokens in the vocabulary.
+
+            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            [What are decoder input IDs?](../glossary#decoder-input-ids)
+
+            Pegasus uses the `pad_token_id` as the starting token for `decoder_input_ids` generation. If
+            `past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see
+            `past_key_values`).
+        decoder_attention_mask (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
+            Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also
+            be used by default.
+
+        Example:
+
+        ```python
+        >>> from transformers import AutoTokenizer, PegasusModel
+
+        >>> tokenizer = AutoTokenizer.from_pretrained("google/pegasus-large")
+        >>> model = PegasusModel.from_pretrained("google/pegasus-large")
+
+        >>> inputs = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt")
+        >>> decoder_inputs = tokenizer("Studies show that", return_tensors="pt")
+        >>> outputs = model(input_ids=inputs.input_ids, decoder_input_ids=decoder_inputs.input_ids)
+
+        >>> last_hidden_states = outputs.last_hidden_state
+        >>> list(last_hidden_states.shape)
+        [1, 4, 1024]
+        ```"""
+
+        if encoder_outputs is None:
+            encoder_outputs = self.encoder(
+                input_ids=input_ids,
+                attention_mask=attention_mask,
+                inputs_embeds=inputs_embeds,
+                **kwargs,
+            )
+        # If the user passed a tuple for encoder_outputs, we wrap it in a BaseModelOutput
+        elif not isinstance(encoder_outputs, BaseModelOutput):
+            encoder_outputs = BaseModelOutput(
+                last_hidden_state=encoder_outputs[0],
+                hidden_states=encoder_outputs[1] if len(encoder_outputs) > 1 else None,
+                attentions=encoder_outputs[2] if len(encoder_outputs) > 2 else None,
+            )
+
+        # decoder outputs consists of (dec_features, past_key_values, dec_hidden, dec_attn)
+        decoder_outputs = self.decoder(
+            input_ids=decoder_input_ids,
+            attention_mask=decoder_attention_mask,
+            encoder_hidden_states=encoder_outputs[0],
+            encoder_attention_mask=attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=decoder_inputs_embeds,
+            use_cache=use_cache,
+            **kwargs,
+        )
+
+        return Seq2SeqModelOutput(
+            last_hidden_state=decoder_outputs.last_hidden_state,
+            past_key_values=decoder_outputs.past_key_values,
+            decoder_hidden_states=decoder_outputs.hidden_states,
+            decoder_attentions=decoder_outputs.attentions,
+            cross_attentions=decoder_outputs.cross_attentions,
+            encoder_last_hidden_state=encoder_outputs.last_hidden_state,
+            encoder_hidden_states=encoder_outputs.hidden_states,
+            encoder_attentions=encoder_outputs.attentions,
+        )
+
+
+@auto_docstring(
+    custom_intro="""
+    The PEGASUS Model with a language modeling head. Can be used for summarization.
+    """
+)
+class PegasusForConditionalGeneration(PegasusPreTrainedModel, GenerationMixin):
+    base_model_prefix = "model"
+    _keys_to_ignore_on_load_missing = ["final_logits_bias"]
+    _tied_weights_keys = {
+        "lm_head.weight": "model.shared.weight",
+    }
+
+    def __init__(self, config: PegasusConfig):
+        super().__init__(config)
+        self.model = PegasusModel(config)
+        self.register_buffer("final_logits_bias", torch.zeros((1, self.model.shared.num_embeddings)))
+        self.lm_head = nn.Linear(config.d_model, self.model.shared.num_embeddings, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def resize_token_embeddings(
+        self, new_num_tokens: int, pad_to_multiple_of: int | None = None, mean_resizing: bool = True
+    ) -> nn.Embedding:
+        new_embeddings = super().resize_token_embeddings(new_num_tokens, pad_to_multiple_of, mean_resizing)
+        self._resize_final_logits_bias(new_embeddings.weight.shape[0])
+        return new_embeddings
+
+    def _resize_final_logits_bias(self, new_num_tokens: int) -> None:
+        old_num_tokens = self.final_logits_bias.shape[-1]
+        if new_num_tokens <= old_num_tokens:
+            new_bias = self.final_logits_bias[:, :new_num_tokens]
+        else:
+            extra_bias = torch.zeros((1, new_num_tokens - old_num_tokens), device=self.final_logits_bias.device)
+            new_bias = torch.cat([self.final_logits_bias, extra_bias], dim=1)
+        self.register_buffer("final_logits_bias", new_bias)
+
+    def resize_position_embeddings(self, new_num_position_embeddings: int):
+        """
+        Resizes position embeddings matrix of the model if `new_num_position_embeddings !=
+        config.max_position_embeddings`.
+
+        Arguments:
+            new_num_position_embeddings (`int`):
+                The number of new position embeddings. If position embeddings are learned, increasing the size will add
+                newly initialized vectors at the end, whereas reducing the size will remove vectors from the end. If
+                position embeddings are not learned (*e.g.* sinusoidal position embeddings), increasing the size will
+                add correct vectors at the end following the position encoding algorithm, whereas reducing the size
+                will remove vectors from the end.
+        """
+        self.config.max_position_embeddings = new_num_position_embeddings
+        self.model.encoder.resize_position_embeddings(new_num_position_embeddings)
+        self.model.decoder.resize_position_embeddings(new_num_position_embeddings)
+
+    def get_position_embeddings(self) -> tuple[nn.Embedding]:
+        """
+        Returns the position embeddings matrix
+        """
+        return (self.model.encoder.get_position_embeddings(), self.model.decoder.get_position_embeddings())
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: torch.Tensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        decoder_input_ids: torch.Tensor | None = None,
+        decoder_attention_mask: torch.Tensor | None = None,
+        encoder_outputs: tuple[torch.FloatTensor] | None = None,
+        past_key_values: Cache | None = None,
+        inputs_embeds: torch.Tensor | None = None,
+        decoder_inputs_embeds: torch.Tensor | None = None,
+        labels: torch.Tensor | None = None,
+        use_cache: bool | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple | Seq2SeqLMOutput:
+        r"""
+        decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
+            Indices of decoder input sequence tokens in the vocabulary.
+
+            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            [What are decoder input IDs?](../glossary#decoder-input-ids)
+
+            Pegasus uses the `pad_token_id` as the starting token for `decoder_input_ids` generation. If
+            `past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see
+            `past_key_values`).
+        decoder_attention_mask (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
+            Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also
+            be used by default.
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        Example Summarization:
+
+        ```python
+        >>> from transformers import AutoTokenizer, PegasusForConditionalGeneration
+
+        >>> model = PegasusForConditionalGeneration.from_pretrained("google/pegasus-xsum")
+        >>> tokenizer = AutoTokenizer.from_pretrained("google/pegasus-xsum")
+
+        >>> ARTICLE_TO_SUMMARIZE = (
+        ...     "PG&E stated it scheduled the blackouts in response to forecasts for high winds "
+        ...     "amid dry conditions. The aim is to reduce the risk of wildfires. Nearly 800 thousand customers were "
+        ...     "scheduled to be affected by the shutoffs which were expected to last through at least midday tomorrow."
+        ... )
+        >>> inputs = tokenizer(ARTICLE_TO_SUMMARIZE, max_length=1024, return_tensors="pt")
+
+        >>> # Generate Summary
+        >>> summary_ids = model.generate(inputs["input_ids"])
+        >>> tokenizer.batch_decode(summary_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+        "California's largest electricity provider has turned off power to hundreds of thousands of customers."
+        ```
+        """
+        if labels is not None:
+            if use_cache:
+                logger.warning("The `use_cache` argument is changed to `False` since `labels` is provided.")
+            use_cache = False
+            if decoder_input_ids is None and decoder_inputs_embeds is None:
+                decoder_input_ids = shift_tokens_right(
+                    labels, self.config.pad_token_id, self.config.decoder_start_token_id
+                )
+
+        outputs: Seq2SeqModelOutput = self.model(
+            input_ids,
+            attention_mask=attention_mask,
+            decoder_input_ids=decoder_input_ids,
+            encoder_outputs=encoder_outputs,
+            decoder_attention_mask=decoder_attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            decoder_inputs_embeds=decoder_inputs_embeds,
+            use_cache=use_cache,
+            **kwargs,
+        )
+        lm_logits = self.lm_head(outputs.last_hidden_state) + self.final_logits_bias
+
+        masked_lm_loss = None
+        if labels is not None:
+            loss_fct = CrossEntropyLoss()
+            masked_lm_loss = loss_fct(lm_logits.view(-1, self.config.vocab_size), labels.view(-1))
+
+        return Seq2SeqLMOutput(
+            loss=masked_lm_loss,
+            logits=lm_logits,
+            past_key_values=outputs.past_key_values,
+            decoder_hidden_states=outputs.decoder_hidden_states,
+            decoder_attentions=outputs.decoder_attentions,
+            cross_attentions=outputs.cross_attentions,
+            encoder_last_hidden_state=outputs.encoder_last_hidden_state,
+            encoder_hidden_states=outputs.encoder_hidden_states,
+            encoder_attentions=outputs.encoder_attentions,
+        )
+
+    def prepare_decoder_input_ids_from_labels(self, labels: torch.Tensor):
+        return shift_tokens_right(labels, self.config.pad_token_id, self.config.decoder_start_token_id)
+
+
+# Copied from transformers.models.bart.modeling_bart.BartDecoderWrapper with Bart->Pegasus
+class PegasusDecoderWrapper(PegasusPreTrainedModel):
+    """
+    This wrapper class is a helper class to correctly load pretrained checkpoints when the causal language model is
+    used in combination with the [`EncoderDecoderModel`] framework.
+    """
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.decoder = PegasusDecoder(config)
+        self.post_init()
+
+    def forward(self, *args, **kwargs):
+        return self.decoder(*args, **kwargs)
+
+
+class PegasusForCausalLM(PegasusPreTrainedModel, GenerationMixin):
+    _tied_weights_keys = {
+        "lm_head.weight": "model.decoder.embed_tokens.weight",
+    }
+
+    def __init__(self, config):
+        config = copy.deepcopy(config)
+        config.is_decoder = True
+        config.is_encoder_decoder = False
+        super().__init__(config)
+        self.model = PegasusDecoderWrapper(config)
+
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.model.decoder.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.decoder.embed_tokens = value
+
+    def get_position_embeddings(self) -> nn.Embedding:
+        """
+        Returns the position embeddings matrix
+        """
+        return self.model.decoder.get_position_embeddings()
+
+    def resize_position_embeddings(self, new_num_position_embeddings: int):
+        """
+        Resizes position embeddings matrix of the model if `new_num_position_embeddings !=
+        config.max_position_embeddings`.
+
+        Arguments:
+            new_num_position_embeddings (`int`):
+                The number of new position embeddings. If position embeddings are learned, increasing the size will add
+                newly initialized vectors at the end, whereas reducing the size will remove vectors from the end. If
+                position embeddings are not learned (*e.g.* sinusoidal position embeddings), increasing the size will
+                add correct vectors at the end following the position encoding algorithm, whereas reducing the size
+                will remove vectors from the end.
+        """
+        self.config.max_position_embeddings = new_num_position_embeddings
+        self.model.decoder.resize_position_embeddings(new_num_position_embeddings)
+
+    @can_return_tuple
+    @auto_docstring
+    # Copied from transformers.models.bart.modeling_bart.BartForCausalLM.forward with Bart->Pegasus, facebook/bart-base->google/pegasus-large
+    def forward(
+        self,
+        input_ids: torch.LongTensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        encoder_hidden_states: torch.FloatTensor | None = None,
+        encoder_attention_mask: torch.FloatTensor | None = None,
+        past_key_values: Cache | None = None,
+        inputs_embeds: torch.FloatTensor | None = None,
+        labels: torch.LongTensor | None = None,
+        use_cache: bool | None = None,
+        logits_to_keep: int | torch.Tensor = 0,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple | CausalLMOutputWithCrossAttentions:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        Example:
+
+        ```python
+        >>> from transformers import AutoTokenizer, PegasusForCausalLM
+
+        >>> tokenizer = AutoTokenizer.from_pretrained("google/pegasus-large")
+        >>> model = PegasusForCausalLM.from_pretrained("google/pegasus-large")
+        >>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
+        >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
+        >>> outputs = model(**inputs)
+
+        >>> logits = outputs.logits
+        >>> expected_shape = [1, inputs.input_ids.shape[-1], model.config.vocab_size]
+        >>> list(logits.shape) == expected_shape
+        True
+        ```"""
+
+        outputs: BaseModelOutputWithPastAndCrossAttentions = self.model.decoder(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            **kwargs,
+        )
+
+        hidden_states = outputs[0]
+        # Only compute necessary logits
+        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
+        logits = self.lm_head(hidden_states[:, slice_indices, :])
+
+        loss = None
+        if labels is not None:
+            labels = labels.to(logits.device)
+            loss_fct = CrossEntropyLoss()
+            loss = loss_fct(logits.view(-1, self.config.vocab_size), labels.view(-1))
+
+        return CausalLMOutputWithCrossAttentions(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+            cross_attentions=outputs.cross_attentions,
+        )
+
+
+__all__ = ["PegasusForCausalLM", "PegasusForConditionalGeneration", "PegasusModel", "PegasusPreTrainedModel"]
diff --git a/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/wavlm/__init__.py b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/wavlm/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..d230bf3f0924b436e1b9cd9ea509a989cf5bdb8b
--- /dev/null
+++ b/LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/transformers/models/wavlm/__init__.py
@@ -0,0 +1,27 @@
+# Copyright 2024 The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import TYPE_CHECKING
+
+from ...utils import _LazyModule
+from ...utils.import_utils import define_import_structure
+
+
+if TYPE_CHECKING:
+    from .configuration_wavlm import *
+    from .modeling_wavlm import *
+else:
+    import sys
+
+    _file = globals()["__file__"]
+    sys.modules[__name__] = _LazyModule(__name__, _file, define_import_structure(_file), module_spec=__spec__)