| {"event": "start", "run": "ali_ali-surface-only_s1", "arch": "ali", "objective": "ratio", "condition": "A", "params": 22785088, "langs": ["en", "tr", "fi"], "premium": {"en": 1.0, "tr": 1.12, "fi": 1.104}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "global", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {}, "weights": {"en": 0.74, "tr": 0.1536, "fi": 0.1064}, "device": "cuda", "ali_params": 268096, "ali_config": {"enabled": true, "budget_B": 65536, "staging_capacity": 262144, "policy": "utility", "candidate_types": ["surface"], "gate": "vector", "retrieval": true, "memory_scope": "persistent", "read_dropout": 0.1, "write_rule": "mean", "eta_min": 0.0, "consolidate_every": 50, "probe_seqs": 2, "probe_reps": 8, "horizon": 32, "predictor": "mlp", "lambda_util": 0.1, "max_span": 64, "byte_classes_path": "/local/scratch/sas245/beetle-hnet/scratch/data/byte_classes.json", "gate_hidden": 0, "type_dim": 16, "gate_bias_init": -2.0}, "t": 2.1} |
| {"event": "probe_schedule", "n": 40, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 200, 256, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 2.1} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2674.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.153652, "cps": 20.890625, "mean_p": 0.478126, "boundary_entropy": 0.685438, "bpb": 8.041379}, "tr": {"chunks": 5645.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.311827, "cps": 44.101562, "mean_p": 0.486063, "boundary_entropy": 0.686375, "bpb": 8.01701}, "fi": {"chunks": 4732.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251555, "cps": 36.96875, "mean_p": 0.48451, "boundary_entropy": 0.686595, "bpb": 8.075294}}, "bpb": {"en": 8.04138, "tr": 8.01701, "fi": 8.07529}, "gini_cpc": 0.147064, "gini_cps": 0.151764, "bpic": 4.346947, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 2.029439514461233, "fi": 1.6371767549398106}, "inventory": {"size": 0, "admitted": 0, "staged": 0, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 0, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 4.2} |
| {"event": "train", "step": 0, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 5.58509, "bpb": 8.05758, "chunk_loss": 0.030141, "F": 0.16724, "G": 0.47713, "chunks": 171.25, "bytes_per_chunk": 5.98, "boundary_entropy": 0.69022, "router_noise": 0.0, "grad_norm": 11.3984, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.012, "probe_sec": 0.0067, "probe_extra_seqs": 0, "n_rows": 0, "extra_fwd_seqs": 0, "wall_s": 0.0, "overhead_frac_est": 0.0, "chunk_mismatch": 0, "n_loo": 0, "mean_delta_loo": NaN, "n_shadow": 0, "mean_delta_shadow": NaN, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "read_hit_rate": {"surface": 0.0}, "mean_gate": {"surface": null}, "n_reads": 0, "n_cand": 5434, "read_suppressed_frac": 0.0, "dropout_reads_logged": 0, "inventory": {"size": 3161, "admitted": 0, "staged": 3161, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 3161, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 0.007, "probe_overhead_seq_frac": 0.0, "util_loss": 0.0, "alpha": 0.03, "t": 5.9} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2659.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.15279, "cps": 20.773438, "mean_p": 0.478105, "boundary_entropy": 0.685439, "bpb": 8.008091}, "tr": {"chunks": 5624.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.310667, "cps": 43.9375, "mean_p": 0.485999, "boundary_entropy": 0.686373, "bpb": 7.993558}, "fi": {"chunks": 4706.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.250173, "cps": 36.765625, "mean_p": 0.484401, "boundary_entropy": 0.686587, "bpb": 8.047735}}, "bpb": {"en": 8.00809, "tr": 7.99356, "fi": 8.04774}, "gini_cpc": 0.147487, "gini_cps": 0.15218, "bpic": 4.367696, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 2.0332957058204215, "fi": 1.6373662046178201}, "inventory": {"size": 3161, "admitted": 0, "staged": 3161, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 3161, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 7.1} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2659.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.15279, "cps": 20.773438, "mean_p": 0.47808, "boundary_entropy": 0.685441, "bpb": 7.943159}, "tr": {"chunks": 5587.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.308623, "cps": 43.648438, "mean_p": 0.4859, "boundary_entropy": 0.686371, "bpb": 7.946053}, "fi": {"chunks": 4666.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.248046, "cps": 36.453125, "mean_p": 0.484226, "boundary_entropy": 0.686575, "bpb": 7.993978}}, "bpb": {"en": 7.94316, "tr": 7.94605, "fi": 7.99398}, "gini_cpc": 0.146434, "gini_cps": 0.151177, "bpic": 4.393742, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 2.0199187603873923, "fi": 1.623448939810189}, "inventory": {"size": 5568, "admitted": 0, "staged": 5568, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 5568, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 11.3} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2615.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.150261, "cps": 20.429688, "mean_p": 0.478043, "boundary_entropy": 0.685455, "bpb": 7.727324}, "tr": {"chunks": 5454.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.301276, "cps": 42.609375, "mean_p": 0.485574, "boundary_entropy": 0.686365, "bpb": 7.790736}, "fi": {"chunks": 4538.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.241242, "cps": 35.453125, "mean_p": 0.483679, "boundary_entropy": 0.686539, "bpb": 7.8144}}, "bpb": {"en": 7.72732, "tr": 7.79074, "fi": 7.8144}, "gini_cpc": 0.145322, "gini_cps": 0.150128, "bpic": 4.50004, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 2.005012152153774, "fi": 1.6054805002524357}, "inventory": {"size": 10489, "admitted": 0, "staged": 10489, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 10489, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 16.4} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2623.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.150721, "cps": 20.492188, "mean_p": 0.478843, "boundary_entropy": 0.685612, "bpb": 7.16975}, "tr": {"chunks": 5152.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.284594, "cps": 40.25, "mean_p": 0.484944, "boundary_entropy": 0.68639, "bpb": 7.398292}, "fi": {"chunks": 4103.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.218117, "cps": 32.054688, "mean_p": 0.482324, "boundary_entropy": 0.686466, "bpb": 7.341237}}, "bpb": {"en": 7.16975, "tr": 7.39829, "fi": 7.34124}, "gini_cpc": 0.136584, "gini_cps": 0.141943, "bpic": 4.776225, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.8882136486372962, "fi": 1.4471563784567854}, "inventory": {"size": 19575, "admitted": 0, "staged": 19575, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 19575, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 22.5} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2595.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.149112, "cps": 20.273438, "mean_p": 0.481094, "boundary_entropy": 0.686031, "bpb": 6.515024}, "tr": {"chunks": 4390.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.242501, "cps": 34.296875, "mean_p": 0.482943, "boundary_entropy": 0.686436, "bpb": 6.898333}, "fi": {"chunks": 2927.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.1556, "cps": 22.867188, "mean_p": 0.478259, "boundary_entropy": 0.686271, "bpb": 6.694145}}, "bpb": {"en": 6.51502, "tr": 6.89833, "fi": 6.69415}, "gini_cpc": 0.113775, "gini_cps": 0.120729, "bpic": 5.723567, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.6263002427662647, "fi": 1.0435123588676285}, "inventory": {"size": 38356, "admitted": 0, "staged": 38356, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 38356, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 28.9} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3674.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.211113, "cps": 28.703125, "mean_p": 0.490425, "boundary_entropy": 0.687168, "bpb": 6.161286}, "tr": {"chunks": 3937.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.217478, "cps": 30.757812, "mean_p": 0.486201, "boundary_entropy": 0.687281, "bpb": 6.577344}, "fi": {"chunks": 3146.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.167243, "cps": 24.578125, "mean_p": 0.485007, "boundary_entropy": 0.687293, "bpb": 6.341681}}, "bpb": {"en": 6.16129, "tr": 6.57734, "fi": 6.34168}, "gini_cpc": 0.056207, "gini_cps": 0.049022, "bpic": 5.273961, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0301484931188678, "fi": 0.792194463871152}, "inventory": {"size": 72570, "admitted": 0, "staged": 72570, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 72570, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 42.7} |
| {"event": "consolidate", "step": 50, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 65536, "swapped": 0, "uncertain_staged": 108293, "t": 53.5} |
| {"event": "train", "step": 50, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 4.09847, "bpb": 5.91285, "chunk_loss": 0.030543, "F": 0.17416, "G": 0.48631, "chunks": 178.34, "bytes_per_chunk": 5.742, "boundary_entropy": 0.69164, "router_noise": 0.0, "grad_norm": 2.362, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.963, "probe_sec": 0.0413, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0337, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.00738, "n_shadow": 4, "mean_delta_shadow": -0.01556, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": -0.18656, "trunc_abs_frac": 3.86936, "mean_delta_full": -0.02165, "read_hit_rate": {"surface": 0.19827}, "mean_gate": {"surface": 0.11941}, "n_reads": 1125, "n_cand": 5674, "read_suppressed_frac": 0.1067, "dropout_reads_logged": 1125, "inventory": {"size": 110326, "admitted": 65536, "staged": 44790, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 44790, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 1.81561279296875, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 6.022, "probe_overhead_seq_frac": 0.4902, "util_loss": 0.002819, "alpha": 0.03, "t": 54.1} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3747.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.215308, "cps": 29.273438, "mean_p": 0.484546, "boundary_entropy": 0.686192, "bpb": 5.428817}, "tr": {"chunks": 4451.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.245871, "cps": 34.773438, "mean_p": 0.481552, "boundary_entropy": 0.686144, "bpb": 5.921963}, "fi": {"chunks": 3972.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.211153, "cps": 31.03125, "mean_p": 0.48266, "boundary_entropy": 0.686565, "bpb": 5.654448}}, "bpb": {"en": 5.42882, "tr": 5.92196, "fi": 5.65445}, "gini_cpc": 0.034425, "gini_cps": 0.038565, "bpic": 4.661627, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1419510021097583, "fi": 0.9807036315352852}, "inventory": {"size": 138604, "admitted": 65536, "staged": 73068, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 73068, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 1.990234375, "n_util_obs": 7, "util_count_spearman": 0.35714285714285715}, "t": 58.9} |
| {"event": "consolidate", "step": 100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 19, "uncertain_staged": 148312, "t": 81.3} |
| {"event": "train", "step": 100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 3.20763, "bpb": 4.62763, "chunk_loss": 0.032931, "F": 0.21246, "G": 0.46167, "chunks": 217.56, "bytes_per_chunk": 4.707, "boundary_entropy": 0.68342, "router_noise": 0.0, "grad_norm": 1.537, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.547, "probe_sec": 0.0287, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02397, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.00283, "n_shadow": 3, "mean_delta_shadow": -0.01832, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 7, "trunc_ratio_sum": -0.09797, "trunc_abs_frac": 0.374, "mean_delta_full": -0.06944, "read_hit_rate": {"surface": 0.06176}, "mean_gate": {"surface": 0.1198}, "n_reads": 428, "n_cand": 6930, "read_suppressed_frac": 0.0888, "dropout_reads_logged": 428, "inventory": {"size": 215626, "admitted": 65536, "staged": 150090, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 150090, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 1.7651519775390625, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 9.307, "probe_overhead_seq_frac": 0.495, "util_loss": 0.000425, "alpha": 0.03, "t": 81.4} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3754.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.21571, "cps": 29.328125, "mean_p": 0.426759, "boundary_entropy": 0.621098, "bpb": 4.113161}, "tr": {"chunks": 4088.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.225819, "cps": 31.9375, "mean_p": 0.40903, "boundary_entropy": 0.640689, "bpb": 4.367252}, "fi": {"chunks": 4216.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.224124, "cps": 32.9375, "mean_p": 0.415711, "boundary_entropy": 0.643646, "bpb": 4.324014}}, "bpb": {"en": 4.11316, "tr": 4.36725, "fi": 4.32401}, "gini_cpc": 0.010124, "gini_cps": 0.025543, "bpic": 4.704926, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0468638125924257, "fi": 1.0390072325029334}, "inventory": {"size": 257431, "admitted": 65536, "staged": 191895, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 191895, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 1.9163818359375, "n_util_obs": 14, "util_count_spearman": 0.01098901098901099}, "t": 93.8} |
| {"event": "consolidate", "step": 150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 18, "uncertain_staged": 219610, "t": 108.3} |
| {"event": "train", "step": 150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.75591, "bpb": 3.97593, "chunk_loss": 0.032645, "F": 0.21838, "G": 0.40102, "chunks": 223.62, "bytes_per_chunk": 4.579, "boundary_entropy": 0.58473, "router_noise": 0.0, "grad_norm": 2.163, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.557, "probe_sec": 0.0492, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03358, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.09041, "n_shadow": 9, "mean_delta_shadow": 0.03767, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.52459, "trunc_abs_frac": 2.4609, "mean_delta_full": 0.03475, "read_hit_rate": {"surface": 0.0546}, "mean_gate": {"surface": 0.11944}, "n_reads": 389, "n_cand": 7124, "read_suppressed_frac": 0.1054, "dropout_reads_logged": 389, "inventory": {"size": 286254, "admitted": 65536, "staged": 220718, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 220718, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 1.722808837890625, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 12.59, "probe_overhead_seq_frac": 0.4967, "util_loss": 0.000453, "alpha": 0.03, "t": 109.3} |
| {"event": "consolidate", "step": 200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 10, "uncertain_staged": 262133, "t": 132.6} |
| {"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3243.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186347, "cps": 25.335938, "mean_p": 0.324529, "boundary_entropy": 0.464747, "bpb": 3.54657}, "tr": {"chunks": 4334.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.239408, "cps": 33.859375, "mean_p": 0.302799, "boundary_entropy": 0.480725, "bpb": 3.780925}, "fi": {"chunks": 3719.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.197703, "cps": 29.054688, "mean_p": 0.327174, "boundary_entropy": 0.520379, "bpb": 3.899712}}, "bpb": {"en": 3.54657, "tr": 3.78092, "fi": 3.89971}, "gini_cpc": 0.056738, "gini_cps": 0.064389, "bpic": 5.022309, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.284740831616064, "fi": 1.0609415702456457}, "inventory": {"size": 327670, "admitted": 65536, "staged": 262134, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262134, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 1.7981414794921875, "n_util_obs": 5, "util_count_spearman": -0.1}, "t": 133.2} |
| {"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.55595, "tr": 3.78987, "fi": 3.89937}, "bytes_per_chunk": {"en": 5.201, "tr": 4.553, "fi": 5.345}, "bpb_mean": 3.74839, "frozen_inventory": true, "t": 134.6} |
| {"event": "train", "step": 200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.48728, "bpb": 3.58839, "chunk_loss": 0.03094, "F": 0.19431, "G": 0.32434, "chunks": 198.97, "bytes_per_chunk": 5.147, "boundary_entropy": 0.47371, "router_noise": 0.0, "grad_norm": 1.6503, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.577, "probe_sec": 0.0487, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02559, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.02717, "n_shadow": 7, "mean_delta_shadow": -0.03272, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 2.34, "trunc_abs_frac": 1.07303, "mean_delta_full": -0.04578, "read_hit_rate": {"surface": 0.03552}, "mean_gate": {"surface": 0.12649}, "n_reads": 225, "n_cand": 6335, "read_suppressed_frac": 0.0711, "dropout_reads_logged": 225, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 1.80157470703125, "n_util_obs": 7, "util_count_spearman": 0.21428571428571427}, "probe_overhead_sec_total": 15.859, "probe_overhead_seq_frac": 0.4975, "util_loss": 0.000897, "alpha": 0.03, "t": 138.2} |
| {"event": "consolidate", "step": 250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262133, "t": 154.9} |
| {"event": "train", "step": 250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.35771, "bpb": 3.40146, "chunk_loss": 0.030567, "F": 0.18765, "G": 0.2879, "chunks": 192.16, "bytes_per_chunk": 5.329, "boundary_entropy": 0.388, "router_noise": 0.0, "grad_norm": 1.3791, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.361, "probe_sec": 0.0597, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04208, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.11211, "n_shadow": 5, "mean_delta_shadow": -0.20359, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.30696, "trunc_abs_frac": 2.65396, "mean_delta_full": 0.08073, "read_hit_rate": {"surface": 0.0976}, "mean_gate": {"surface": 0.13967}, "n_reads": 597, "n_cand": 6117, "read_suppressed_frac": 0.0921, "dropout_reads_logged": 597, "inventory": {"size": 327426, "admitted": 65536, "staged": 261890, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261890, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 2.9429779052734375, "n_util_obs": 13, "util_count_spearman": -0.01098901098901099}, "probe_overhead_sec_total": 19.004, "probe_overhead_seq_frac": 0.498, "util_loss": 0.003619, "alpha": 0.03, "t": 156.2} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3133.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.180026, "cps": 24.476562, "mean_p": 0.278872, "boundary_entropy": 0.368714, "bpb": 3.280862}, "tr": {"chunks": 3875.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214053, "cps": 30.273438, "mean_p": 0.276474, "boundary_entropy": 0.404402, "bpb": 3.5085}, "fi": {"chunks": 3321.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176546, "cps": 25.945312, "mean_p": 0.28119, "boundary_entropy": 0.43625, "bpb": 3.651772}}, "bpb": {"en": 3.28086, "tr": 3.5085, "fi": 3.65177}, "gini_cpc": 0.04382, "gini_cps": 0.047891, "bpic": 5.492497, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.18900828484394, "fi": 0.9806650946147105}, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 2.992401123046875, "n_util_obs": 13, "util_count_spearman": -0.07142857142857142}, "t": 159.0} |
| {"event": "consolidate", "step": 300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262137, "t": 190.2} |
| {"event": "train", "step": 300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.20486, "bpb": 3.18093, "chunk_loss": 0.030509, "F": 0.18658, "G": 0.28515, "chunks": 191.06, "bytes_per_chunk": 5.36, "boundary_entropy": 0.36941, "router_noise": 0.0, "grad_norm": 1.4613, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.686, "probe_sec": 0.0404, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02667, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.36117, "n_shadow": 7, "mean_delta_shadow": -0.02802, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.87896, "trunc_abs_frac": 0.89294, "mean_delta_full": 0.31632, "read_hit_rate": {"surface": 0.06495}, "mean_gate": {"surface": 0.14422}, "n_reads": 395, "n_cand": 6082, "read_suppressed_frac": 0.0962, "dropout_reads_logged": 395, "inventory": {"size": 327416, "admitted": 65536, "staged": 261880, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261880, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 2.866729736328125, "n_util_obs": 13, "util_count_spearman": 0.15934065934065933}, "probe_overhead_sec_total": 22.36, "probe_overhead_seq_frac": 0.4983, "util_loss": 0.004408, "alpha": 0.03, "t": 190.5} |
| {"event": "consolidate", "step": 350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262136, "t": 209.3} |
| {"event": "train", "step": 350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.18896, "bpb": 3.158, "chunk_loss": 0.030548, "F": 0.1889, "G": 0.28228, "chunks": 193.44, "bytes_per_chunk": 5.294, "boundary_entropy": 0.35928, "router_noise": 0.0, "grad_norm": 1.3428, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.381, "probe_sec": 0.0766, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06521, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.14996, "n_shadow": 4, "mean_delta_shadow": -0.04633, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 7, "trunc_ratio_sum": 0.86126, "trunc_abs_frac": 1.01748, "mean_delta_full": 0.14978, "read_hit_rate": {"surface": 0.06431}, "mean_gate": {"surface": 0.15929}, "n_reads": 396, "n_cand": 6158, "read_suppressed_frac": 0.096, "dropout_reads_logged": 396, "inventory": {"size": 327450, "admitted": 65536, "staged": 261914, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261914, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 3.1793212890625, "n_util_obs": 15, "util_count_spearman": 0.3678571428571429}, "probe_overhead_sec_total": 26.159, "probe_overhead_seq_frac": 0.4986, "util_loss": 0.004907, "alpha": 0.03, "t": 209.5} |
| {"event": "consolidate", "step": 400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262125, "t": 234.5} |
| {"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3355.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192783, "cps": 26.210938, "mean_p": 0.256687, "boundary_entropy": 0.286436, "bpb": 2.867804}, "tr": {"chunks": 3441.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.190079, "cps": 26.882812, "mean_p": 0.283416, "boundary_entropy": 0.385213, "bpb": 3.127908}, "fi": {"chunks": 3437.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.182712, "cps": 26.851562, "mean_p": 0.279657, "boundary_entropy": 0.399817, "bpb": 3.349425}}, "bpb": {"en": 2.8678, "tr": 3.12791, "fi": 3.34943}, "gini_cpc": 0.011871, "gini_cps": 0.005603, "bpic": 5.544024, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9859745768397807, "fi": 0.947761896299855}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 3.683441162109375, "n_util_obs": 18, "util_count_spearman": 0.5686274509803921}, "t": 235.1} |
| {"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.8803, "tr": 3.1354, "fi": 3.34106}, "bytes_per_chunk": {"en": 4.918, "tr": 5.632, "fi": 5.642}, "bpb_mean": 3.11892, "frozen_inventory": true, "t": 236.0} |
| {"event": "train", "step": 400, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.02528, "bpb": 2.92186, "chunk_loss": 0.03061, "F": 0.19647, "G": 0.26195, "chunks": 201.19, "bytes_per_chunk": 5.09, "boundary_entropy": 0.29171, "router_noise": 0.0, "grad_norm": 1.2206, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.591, "probe_sec": 0.0332, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02686, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.0962, "n_shadow": 7, "mean_delta_shadow": -0.13535, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -0.54568, "trunc_abs_frac": 0.32882, "mean_delta_full": -0.07599, "read_hit_rate": {"surface": 0.07883}, "mean_gate": {"surface": 0.16554}, "n_reads": 505, "n_cand": 6406, "read_suppressed_frac": 0.105, "dropout_reads_logged": 505, "inventory": {"size": 327242, "admitted": 65536, "staged": 261706, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261706, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 3.6911468505859375, "n_util_obs": 18, "util_count_spearman": 0.5892672858617131}, "probe_overhead_sec_total": 29.814, "probe_overhead_seq_frac": 0.4988, "util_loss": 0.006002, "alpha": 0.03, "t": 239.1} |
| {"event": "consolidate", "step": 450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262120, "t": 264.7} |
| {"event": "train", "step": 450, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.06216, "bpb": 2.97507, "chunk_loss": 0.030645, "F": 0.19662, "G": 0.2626, "chunks": 201.34, "bytes_per_chunk": 5.086, "boundary_entropy": 0.29065, "router_noise": 0.0, "grad_norm": 1.1171, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.526, "probe_sec": 0.0904, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04327, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.12675, "n_shadow": 8, "mean_delta_shadow": 0.10573, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.10401, "trunc_abs_frac": 1.26062, "mean_delta_full": 0.18365, "read_hit_rate": {"surface": 0.09}, "mean_gate": {"surface": 0.1657}, "n_reads": 577, "n_cand": 6411, "read_suppressed_frac": 0.104, "dropout_reads_logged": 577, "inventory": {"size": 327314, "admitted": 65536, "staged": 261778, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261778, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 4.603607177734375, "n_util_obs": 23, "util_count_spearman": 0.5237154150197628}, "probe_overhead_sec_total": 32.889, "probe_overhead_seq_frac": 0.4989, "util_loss": 0.0078, "alpha": 0.03, "t": 265.4} |
| {"event": "consolidate", "step": 500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262117, "t": 293.4} |
| {"event": "train", "step": 500, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.02133, "bpb": 2.91616, "chunk_loss": 0.030453, "F": 0.19165, "G": 0.25403, "chunks": 196.25, "bytes_per_chunk": 5.218, "boundary_entropy": 0.27937, "router_noise": 0.0, "grad_norm": 1.0675, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.575, "probe_sec": 0.1721, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.15362, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.21024, "n_shadow": 6, "mean_delta_shadow": 0.17585, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.77628, "trunc_abs_frac": 0.87487, "mean_delta_full": 0.51379, "read_hit_rate": {"surface": 0.07923}, "mean_gate": {"surface": 0.15276}, "n_reads": 495, "n_cand": 6248, "read_suppressed_frac": 0.0949, "dropout_reads_logged": 495, "inventory": {"size": 327428, "admitted": 65536, "staged": 261892, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261892, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 4.4337921142578125, "n_util_obs": 22, "util_count_spearman": 0.45454545454545453}, "probe_overhead_sec_total": 36.15, "probe_overhead_seq_frac": 0.499, "util_loss": 0.012477, "alpha": 0.03, "t": 294.2} |
| {"event": "consolidate", "step": 550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262115, "t": 317.9} |
| {"event": "train", "step": 550, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.94518, "bpb": 2.8063, "chunk_loss": 0.030837, "F": 0.20587, "G": 0.25931, "chunks": 210.81, "bytes_per_chunk": 4.857, "boundary_entropy": 0.26247, "router_noise": 0.0, "grad_norm": 1.2349, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.498, "probe_sec": 0.0624, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02739, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.63391, "n_shadow": 5, "mean_delta_shadow": 0.15841, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.99586, "trunc_abs_frac": 1.017, "mean_delta_full": 0.25827, "read_hit_rate": {"surface": 0.11498}, "mean_gate": {"surface": 0.1632}, "n_reads": 772, "n_cand": 6714, "read_suppressed_frac": 0.0972, "dropout_reads_logged": 772, "inventory": {"size": 327274, "admitted": 65536, "staged": 261738, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261738, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 5.63525390625, "n_util_obs": 24, "util_count_spearman": 0.3382608695652174}, "probe_overhead_sec_total": 39.401, "probe_overhead_seq_frac": 0.4991, "util_loss": 0.012292, "alpha": 0.03, "t": 319.1} |
| {"event": "consolidate", "step": 600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262114, "t": 342.3} |
| {"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3455.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198529, "cps": 26.992188, "mean_p": 0.234551, "boundary_entropy": 0.216284, "bpb": 2.581601}, "tr": {"chunks": 3365.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185881, "cps": 26.289062, "mean_p": 0.296811, "boundary_entropy": 0.390804, "bpb": 2.887398}, "fi": {"chunks": 3374.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179363, "cps": 26.359375, "mean_p": 0.2645, "boundary_entropy": 0.355117, "bpb": 3.143094}}, "bpb": {"en": 2.5816, "tr": 2.8874, "fi": 3.14309}, "gini_cpc": 0.022664, "gini_cps": 0.005886, "bpic": 5.565234, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9362904326289195, "fi": 0.9034606948962415}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.1844482421875, "n_util_obs": 24, "util_count_spearman": 0.30434782608695654}, "t": 343.0} |
| {"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.59654, "tr": 2.8986, "fi": 3.14016}, "bytes_per_chunk": {"en": 4.819, "tr": 5.728, "fi": 5.789}, "bpb_mean": 2.87843, "frozen_inventory": true, "t": 344.1} |
| {"event": "train", "step": 600, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.92736, "bpb": 2.7806, "chunk_loss": 0.030563, "F": 0.19711, "G": 0.25367, "chunks": 201.84, "bytes_per_chunk": 5.073, "boundary_entropy": 0.25559, "router_noise": 0.0, "grad_norm": 1.3657, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.563, "probe_sec": 0.0451, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02976, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.15684, "n_shadow": 4, "mean_delta_shadow": 0.40046, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.51379, "trunc_abs_frac": 0.53815, "mean_delta_full": 0.36072, "read_hit_rate": {"surface": 0.1058}, "mean_gate": {"surface": 0.15877}, "n_reads": 680, "n_cand": 6427, "read_suppressed_frac": 0.1044, "dropout_reads_logged": 680, "inventory": {"size": 327339, "admitted": 65536, "staged": 261803, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261803, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.19482421875, "n_util_obs": 24, "util_count_spearman": 0.29391304347826086}, "probe_overhead_sec_total": 42.381, "probe_overhead_seq_frac": 0.4992, "util_loss": 0.014572, "alpha": 0.03, "t": 347.2} |
| {"event": "consolidate", "step": 650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262114, "t": 365.4} |
| {"event": "train", "step": 650, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.85401, "bpb": 2.67477, "chunk_loss": 0.030503, "F": 0.19598, "G": 0.25404, "chunks": 200.69, "bytes_per_chunk": 5.102, "boundary_entropy": 0.25285, "router_noise": 0.0, "grad_norm": 1.2035, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.369, "probe_sec": 0.0855, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07251, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.24408, "n_shadow": 9, "mean_delta_shadow": -0.16608, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.13782, "trunc_abs_frac": 0.37045, "mean_delta_full": -0.15072, "read_hit_rate": {"surface": 0.07527}, "mean_gate": {"surface": 0.17925}, "n_reads": 481, "n_cand": 6390, "read_suppressed_frac": 0.0977, "dropout_reads_logged": 481, "inventory": {"size": 327284, "admitted": 65536, "staged": 261748, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261748, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 5.2145538330078125, "n_util_obs": 19, "util_count_spearman": 0.5350877192982456}, "probe_overhead_sec_total": 45.509, "probe_overhead_seq_frac": 0.4992, "util_loss": 0.016334, "alpha": 0.03, "t": 365.7} |
| {"event": "consolidate", "step": 700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262113, "t": 392.7} |
| {"event": "train", "step": 700, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.87684, "bpb": 2.70771, "chunk_loss": 0.030365, "F": 0.18851, "G": 0.26362, "chunks": 193.03, "bytes_per_chunk": 5.305, "boundary_entropy": 0.28287, "router_noise": 0.0, "grad_norm": 1.0689, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.546, "probe_sec": 0.0338, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02585, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.36401, "n_shadow": 10, "mean_delta_shadow": -0.02576, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.40387, "trunc_abs_frac": 3.01074, "mean_delta_full": -0.11153, "read_hit_rate": {"surface": 0.05468}, "mean_gate": {"surface": 0.17918}, "n_reads": 336, "n_cand": 6145, "read_suppressed_frac": 0.125, "dropout_reads_logged": 336, "inventory": {"size": 327396, "admitted": 65536, "staged": 261860, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261860, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 5.162353515625, "n_util_obs": 18, "util_count_spearman": 0.44685242518059853}, "probe_overhead_sec_total": 48.826, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.019189, "alpha": 0.03, "t": 393.0} |
| {"event": "consolidate", "step": 750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262108, "t": 419.3} |
| {"event": "train", "step": 750, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.77785, "bpb": 2.56489, "chunk_loss": 0.03044, "F": 0.19565, "G": 0.24561, "chunks": 200.34, "bytes_per_chunk": 5.111, "boundary_entropy": 0.23094, "router_noise": 0.0, "grad_norm": 0.9966, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.54, "probe_sec": 0.0589, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02646, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.01279, "n_shadow": 9, "mean_delta_shadow": -0.16714, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.049, "trunc_abs_frac": 1.60331, "mean_delta_full": -0.09383, "read_hit_rate": {"surface": 0.05957}, "mean_gate": {"surface": 0.1817}, "n_reads": 380, "n_cand": 6379, "read_suppressed_frac": 0.1211, "dropout_reads_logged": 380, "inventory": {"size": 327293, "admitted": 65536, "staged": 261757, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261757, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 4.88232421875, "n_util_obs": 14, "util_count_spearman": 0.621978021978022}, "probe_overhead_sec_total": 52.157, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.021032, "alpha": 0.03, "t": 420.0} |
| {"event": "consolidate", "step": 800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262105, "t": 445.5} |
| {"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3444.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197897, "cps": 26.90625, "mean_p": 0.224539, "boundary_entropy": 0.185994, "bpb": 2.418795}, "tr": {"chunks": 3240.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178976, "cps": 25.3125, "mean_p": 0.286807, "boundary_entropy": 0.370784, "bpb": 2.70751}, "fi": {"chunks": 3427.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.182181, "cps": 26.773438, "mean_p": 0.259636, "boundary_entropy": 0.322744, "bpb": 2.98155}}, "bpb": {"en": 2.4188, "tr": 2.70751, "fi": 2.98155}, "gini_cpc": 0.022563, "gini_cps": 0.013451, "bpic": 5.610919, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9043893431334942, "fi": 0.9205835250595646}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 5.1596527099609375, "n_util_obs": 14, "util_count_spearman": 0.7450549450549451}, "t": 446.1} |
| {"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.43571, "tr": 2.72009, "fi": 2.98192}, "bytes_per_chunk": {"en": 4.811, "tr": 6.027, "fi": 5.638}, "bpb_mean": 2.71257, "frozen_inventory": true, "t": 447.3} |
| {"event": "train", "step": 800, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.79093, "bpb": 2.58377, "chunk_loss": 0.030435, "F": 0.19376, "G": 0.24997, "chunks": 198.41, "bytes_per_chunk": 5.161, "boundary_entropy": 0.24496, "router_noise": 0.0, "grad_norm": 0.9833, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.613, "probe_sec": 0.1417, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.13357, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.23773, "n_shadow": 5, "mean_delta_shadow": 0.13478, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.62809, "trunc_abs_frac": 2.54126, "mean_delta_full": -0.21709, "read_hit_rate": {"surface": 0.05351}, "mean_gate": {"surface": 0.17807}, "n_reads": 338, "n_cand": 6317, "read_suppressed_frac": 0.1272, "dropout_reads_logged": 338, "inventory": {"size": 327333, "admitted": 65536, "staged": 261797, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261797, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 5.1648101806640625, "n_util_obs": 14, "util_count_spearman": 0.7450549450549451}, "probe_overhead_sec_total": 55.503, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.022394, "alpha": 0.03, "t": 450.7} |
| {"event": "consolidate", "step": 850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262096, "t": 472.9} |
| {"event": "train", "step": 850, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.73036, "bpb": 2.49637, "chunk_loss": 0.030482, "F": 0.20026, "G": 0.23066, "chunks": 205.06, "bytes_per_chunk": 4.994, "boundary_entropy": 0.19202, "router_noise": 0.0, "grad_norm": 1.2604, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.459, "probe_sec": 0.0499, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02534, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": 0.42842, "n_shadow": 11, "mean_delta_shadow": -0.24955, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.0781}, "mean_gate": {"surface": 0.17505}, "n_reads": 510, "n_cand": 6530, "read_suppressed_frac": 0.0961, "dropout_reads_logged": 510, "inventory": {"size": 327158, "admitted": 65536, "staged": 261622, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261622, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 5.421600341796875, "n_util_obs": 12, "util_count_spearman": 0.7622377622377622}, "probe_overhead_sec_total": 58.673, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.022984, "alpha": 0.03, "t": 473.6} |
| {"event": "consolidate", "step": 900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262091, "t": 497.2} |
| {"event": "train", "step": 900, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.7086, "bpb": 2.46499, "chunk_loss": 0.030432, "F": 0.19763, "G": 0.23746, "chunks": 202.38, "bytes_per_chunk": 5.06, "boundary_entropy": 0.20338, "router_noise": 0.0, "grad_norm": 1.1218, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.493, "probe_sec": 0.0545, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0246, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 2, "mean_delta_loo": 0.79095, "n_shadow": 12, "mean_delta_shadow": -0.12218, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.26564, "trunc_abs_frac": 1.16515, "mean_delta_full": -0.02212, "read_hit_rate": {"surface": 0.06518}, "mean_gate": {"surface": 0.16916}, "n_reads": 420, "n_cand": 6444, "read_suppressed_frac": 0.0929, "dropout_reads_logged": 420, "inventory": {"size": 327183, "admitted": 65536, "staged": 261647, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261647, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 5.825714111328125, "n_util_obs": 13, "util_count_spearman": 0.6648351648351648}, "probe_overhead_sec_total": 61.563, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.025617, "alpha": 0.03, "t": 498.3} |
| {"event": "consolidate", "step": 950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262089, "t": 525.5} |
| {"event": "train", "step": 950, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.70805, "bpb": 2.4642, "chunk_loss": 0.030466, "F": 0.19992, "G": 0.23413, "chunks": 204.72, "bytes_per_chunk": 5.002, "boundary_entropy": 0.19111, "router_noise": 0.0, "grad_norm": 1.0084, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.566, "probe_sec": 0.069, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02837, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.8079, "n_shadow": 9, "mean_delta_shadow": -0.10423, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.87219, "trunc_abs_frac": 1.99639, "mean_delta_full": 0.20745, "read_hit_rate": {"surface": 0.07593}, "mean_gate": {"surface": 0.1741}, "n_reads": 495, "n_cand": 6519, "read_suppressed_frac": 0.103, "dropout_reads_logged": 495, "inventory": {"size": 327120, "admitted": 65536, "staged": 261584, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261584, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.1274871826171875, "n_util_obs": 13, "util_count_spearman": 0.6098901098901099}, "probe_overhead_sec_total": 64.817, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.034916, "alpha": 0.03, "t": 526.6} |
| {"event": "consolidate", "step": 1000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262083, "t": 553.2} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3452.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198357, "cps": 26.96875, "mean_p": 0.219173, "boundary_entropy": 0.168181, "bpb": 2.327945}, "tr": {"chunks": 3242.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179086, "cps": 25.328125, "mean_p": 0.274997, "boundary_entropy": 0.344176, "bpb": 2.619347}, "fi": {"chunks": 3386.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.180001, "cps": 26.453125, "mean_p": 0.24282, "boundary_entropy": 0.28623, "bpb": 2.850943}}, "bpb": {"en": 2.32794, "tr": 2.61935, "fi": 2.85094}, "gini_cpc": 0.023046, "gini_cps": 0.013889, "bpic": 5.628175, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9028503946997255, "fi": 0.9074619070114606}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.516845703125, "n_util_obs": 14, "util_count_spearman": 0.6571428571428571}, "t": 553.8} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.3416, "tr": 2.62778, "fi": 2.85412}, "bytes_per_chunk": {"en": 4.796, "tr": 6.01, "fi": 5.754}, "bpb_mean": 2.60783, "frozen_inventory": true, "t": 554.7} |
| {"event": "train", "step": 1000, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.71716, "bpb": 2.47733, "chunk_loss": 0.030426, "F": 0.19827, "G": 0.23605, "chunks": 203.03, "bytes_per_chunk": 5.044, "boundary_entropy": 0.19946, "router_noise": 0.0, "grad_norm": 0.9076, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.658, "probe_sec": 0.0414, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0265, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.7085, "n_shadow": 8, "mean_delta_shadow": -0.26203, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.94801, "trunc_abs_frac": 1.32825, "mean_delta_full": 0.71183, "read_hit_rate": {"surface": 0.07162}, "mean_gate": {"surface": 0.1712}, "n_reads": 463, "n_cand": 6465, "read_suppressed_frac": 0.1037, "dropout_reads_logged": 463, "inventory": {"size": 327227, "admitted": 65536, "staged": 261691, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261691, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.5239105224609375, "n_util_obs": 14, "util_count_spearman": 0.6483516483516484}, "probe_overhead_sec_total": 67.741, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.035182, "alpha": 0.03, "t": 559.5} |
| {"event": "consolidate", "step": 1050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262079, "t": 581.6} |
| {"event": "train", "step": 1050, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.73688, "bpb": 2.50579, "chunk_loss": 0.030376, "F": 0.19479, "G": 0.23509, "chunks": 199.47, "bytes_per_chunk": 5.134, "boundary_entropy": 0.20462, "router_noise": 0.0, "grad_norm": 0.9682, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.461, "probe_sec": 0.0604, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02828, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 1.07238, "n_shadow": 7, "mean_delta_shadow": -0.50218, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.91994, "trunc_abs_frac": 0.89901, "mean_delta_full": 1.4259, "read_hit_rate": {"surface": 0.06897}, "mean_gate": {"surface": 0.16481}, "n_reads": 438, "n_cand": 6351, "read_suppressed_frac": 0.0776, "dropout_reads_logged": 438, "inventory": {"size": 327275, "admitted": 65536, "staged": 261739, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261739, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 7.1179656982421875, "n_util_obs": 17, "util_count_spearman": 0.7205882352941176}, "probe_overhead_sec_total": 71.375, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.036602, "alpha": 0.03, "t": 582.5} |
| {"event": "consolidate", "step": 1100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262073, "t": 607.6} |
| {"event": "train", "step": 1100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.7163, "bpb": 2.47609, "chunk_loss": 0.030503, "F": 0.20267, "G": 0.23308, "chunks": 207.53, "bytes_per_chunk": 4.934, "boundary_entropy": 0.18539, "router_noise": 0.0, "grad_norm": 0.9398, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.516, "probe_sec": 0.0351, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02378, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.90172, "n_shadow": 4, "mean_delta_shadow": 0.24707, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.38722, "trunc_abs_frac": 0.93479, "mean_delta_full": 0.21594, "read_hit_rate": {"surface": 0.06446}, "mean_gate": {"surface": 0.16821}, "n_reads": 426, "n_cand": 6609, "read_suppressed_frac": 0.0962, "dropout_reads_logged": 426, "inventory": {"size": 327223, "admitted": 65536, "staged": 261687, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261687, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 7.2953338623046875, "n_util_obs": 15, "util_count_spearman": 0.65}, "probe_overhead_sec_total": 74.437, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.037327, "alpha": 0.03, "t": 608.3} |
| {"event": "consolidate", "step": 1150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262067, "t": 630.1} |
| {"event": "train", "step": 1150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.72797, "bpb": 2.49293, "chunk_loss": 0.03055, "F": 0.19916, "G": 0.24676, "chunks": 203.94, "bytes_per_chunk": 5.021, "boundary_entropy": 0.2172, "router_noise": 0.0, "grad_norm": 0.9292, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.454, "probe_sec": 0.0361, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02864, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 1.43521, "n_shadow": 6, "mean_delta_shadow": -0.28606, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.2686, "trunc_abs_frac": 0.79724, "mean_delta_full": -0.20577, "read_hit_rate": {"surface": 0.05112}, "mean_gate": {"surface": 0.1738}, "n_reads": 332, "n_cand": 6494, "read_suppressed_frac": 0.1145, "dropout_reads_logged": 332, "inventory": {"size": 327318, "admitted": 65536, "staged": 261782, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261782, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 7.622833251953125, "n_util_obs": 15, "util_count_spearman": 0.6357142857142857}, "probe_overhead_sec_total": 77.595, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.036484, "alpha": 0.03, "t": 631.0} |
| {"event": "consolidate", "step": 1200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262057, "t": 656.8} |
| {"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3446.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198012, "cps": 26.921875, "mean_p": 0.215807, "boundary_entropy": 0.155118, "bpb": 2.263514}, "tr": {"chunks": 3291.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181793, "cps": 25.710938, "mean_p": 0.262067, "boundary_entropy": 0.319968, "bpb": 2.53584}, "fi": {"chunks": 3302.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.175536, "cps": 25.796875, "mean_p": 0.229091, "boundary_entropy": 0.260498, "bpb": 2.744471}}, "bpb": {"en": 2.26351, "tr": 2.53584, "fi": 2.74447}, "gini_cpc": 0.026982, "gini_cps": 0.010293, "bpic": 5.65116, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9180919468717551, "fi": 0.8864903911910126}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 8.007720947265625, "n_util_obs": 16, "util_count_spearman": 0.5852941176470589}, "t": 657.4} |
| {"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.27728, "tr": 2.54376, "fi": 2.7496}, "bytes_per_chunk": {"en": 4.813, "tr": 5.928, "fi": 5.801}, "bpb_mean": 2.52355, "frozen_inventory": true, "t": 658.4} |
| {"event": "train", "step": 1200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.62797, "bpb": 2.34866, "chunk_loss": 0.030413, "F": 0.19693, "G": 0.23588, "chunks": 201.66, "bytes_per_chunk": 5.078, "boundary_entropy": 0.19365, "router_noise": 0.0, "grad_norm": 0.9107, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.638, "probe_sec": 0.0678, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03525, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": 1.50008, "n_shadow": 11, "mean_delta_shadow": -0.04625, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -1.67831, "trunc_abs_frac": 0.71189, "mean_delta_full": 0.06555, "read_hit_rate": {"surface": 0.06806}, "mean_gate": {"surface": 0.16119}, "n_reads": 437, "n_cand": 6421, "read_suppressed_frac": 0.087, "dropout_reads_logged": 437, "inventory": {"size": 327184, "admitted": 65536, "staged": 261648, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261648, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 8.014389038085938, "n_util_obs": 16, "util_count_spearman": 0.5852941176470589}, "probe_overhead_sec_total": 80.636, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.038178, "alpha": 0.03, "t": 662.9} |
| {"event": "consolidate", "step": 1250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262047, "t": 685.0} |
| {"event": "train", "step": 1250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.66274, "bpb": 2.39883, "chunk_loss": 0.030397, "F": 0.19485, "G": 0.23586, "chunks": 199.53, "bytes_per_chunk": 5.132, "boundary_entropy": 0.19678, "router_noise": 0.0, "grad_norm": 0.9322, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.463, "probe_sec": 0.0583, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02895, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.96814, "n_shadow": 10, "mean_delta_shadow": -0.26488, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.98641, "trunc_abs_frac": 0.78908, "mean_delta_full": 0.42664, "read_hit_rate": {"surface": 0.05777}, "mean_gate": {"surface": 0.15982}, "n_reads": 367, "n_cand": 6353, "read_suppressed_frac": 0.1199, "dropout_reads_logged": 367, "inventory": {"size": 327285, "admitted": 65536, "staged": 261749, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261749, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 8.3377685546875, "n_util_obs": 16, "util_count_spearman": 0.5264705882352941}, "probe_overhead_sec_total": 84.124, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.03879, "alpha": 0.03, "t": 686.0} |
| {"event": "consolidate", "step": 1300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262042, "t": 716.5} |
| {"event": "train", "step": 1300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.58729, "bpb": 2.28998, "chunk_loss": 0.030445, "F": 0.20071, "G": 0.23111, "chunks": 205.53, "bytes_per_chunk": 4.982, "boundary_entropy": 0.16704, "router_noise": 0.0, "grad_norm": 0.8661, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.624, "probe_sec": 0.0348, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02798, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 1.1917, "n_shadow": 4, "mean_delta_shadow": -1.2519, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.84001, "trunc_abs_frac": 0.84588, "mean_delta_full": 1.05098, "read_hit_rate": {"surface": 0.07762}, "mean_gate": {"surface": 0.17065}, "n_reads": 508, "n_cand": 6545, "read_suppressed_frac": 0.0984, "dropout_reads_logged": 508, "inventory": {"size": 327094, "admitted": 65536, "staged": 261558, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261558, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 8.668289184570312, "n_util_obs": 16, "util_count_spearman": 0.3764705882352941}, "probe_overhead_sec_total": 87.602, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.038605, "alpha": 0.03, "t": 717.2} |
| {"event": "consolidate", "step": 1350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262034, "t": 741.9} |
| {"event": "train", "step": 1350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.61432, "bpb": 2.32897, "chunk_loss": 0.030392, "F": 0.19925, "G": 0.21841, "chunks": 204.03, "bytes_per_chunk": 5.019, "boundary_entropy": 0.14441, "router_noise": 0.0, "grad_norm": 0.8856, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.506, "probe_sec": 0.0705, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04377, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.44911, "n_shadow": 7, "mean_delta_shadow": -0.9395, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.92574, "trunc_abs_frac": 1.46406, "mean_delta_full": -0.82071, "read_hit_rate": {"surface": 0.08019}, "mean_gate": {"surface": 0.1614}, "n_reads": 521, "n_cand": 6497, "read_suppressed_frac": 0.1056, "dropout_reads_logged": 521, "inventory": {"size": 327129, "admitted": 65536, "staged": 261593, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261593, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.239822387695312, "n_util_obs": 19, "util_count_spearman": 0.519298245614035}, "probe_overhead_sec_total": 90.523, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.038252, "alpha": 0.03, "t": 742.5} |
| {"event": "consolidate", "step": 1400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262031, "t": 768.6} |
| {"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3454.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198472, "cps": 26.984375, "mean_p": 0.214227, "boundary_entropy": 0.138097, "bpb": 2.205381}, "tr": {"chunks": 3229.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178368, "cps": 25.226562, "mean_p": 0.252741, "boundary_entropy": 0.299153, "bpb": 2.476139}, "fi": {"chunks": 3426.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.182127, "cps": 26.765625, "mean_p": 0.226252, "boundary_entropy": 0.242539, "bpb": 2.676086}}, "bpb": {"en": 2.20538, "tr": 2.47614, "fi": 2.67609}, "gini_cpc": 0.023977, "gini_cps": 0.014838, "bpic": 5.612029, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8987093924775368, "fi": 0.9176504082591354}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.4830322265625, "n_util_obs": 18, "util_count_spearman": 0.31475748194014447}, "t": 769.2} |
| {"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.22018, "tr": 2.48292, "fi": 2.68312}, "bytes_per_chunk": {"en": 4.792, "tr": 5.946, "fi": 5.622}, "bpb_mean": 2.46207, "frozen_inventory": true, "t": 770.1} |
| {"event": "train", "step": 1400, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.58509, "bpb": 2.2868, "chunk_loss": 0.030375, "F": 0.19788, "G": 0.22613, "chunks": 202.62, "bytes_per_chunk": 5.054, "boundary_entropy": 0.16586, "router_noise": 0.0, "grad_norm": 0.9423, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.616, "probe_sec": 0.1346, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.11417, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": 0.26129, "n_shadow": 11, "mean_delta_shadow": -0.00171, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.03048, "trunc_abs_frac": 0.84405, "mean_delta_full": 0.30508, "read_hit_rate": {"surface": 0.0682}, "mean_gate": {"surface": 0.15937}, "n_reads": 440, "n_cand": 6452, "read_suppressed_frac": 0.1091, "dropout_reads_logged": 440, "inventory": {"size": 327122, "admitted": 65536, "staged": 261586, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261586, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.48974609375, "n_util_obs": 18, "util_count_spearman": 0.31475748194014447}, "probe_overhead_sec_total": 93.508, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.039121, "alpha": 0.03, "t": 773.3} |
| {"event": "consolidate", "step": 1450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262024, "t": 793.9} |
| {"event": "train", "step": 1450, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.55898, "bpb": 2.24914, "chunk_loss": 0.030446, "F": 0.19992, "G": 0.23171, "chunks": 204.72, "bytes_per_chunk": 5.002, "boundary_entropy": 0.17314, "router_noise": 0.0, "grad_norm": 0.9079, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.427, "probe_sec": 0.0442, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02622, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": 0.67789, "n_shadow": 11, "mean_delta_shadow": -0.15378, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.93837, "trunc_abs_frac": 0.70808, "mean_delta_full": 0.22729, "read_hit_rate": {"surface": 0.06934}, "mean_gate": {"surface": 0.15638}, "n_reads": 452, "n_cand": 6519, "read_suppressed_frac": 0.0996, "dropout_reads_logged": 452, "inventory": {"size": 327161, "admitted": 65536, "staged": 261625, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261625, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.789337158203125, "n_util_obs": 17, "util_count_spearman": 0.32107843137254904}, "probe_overhead_sec_total": 97.048, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.03947, "alpha": 0.03, "t": 794.7} |
| {"event": "consolidate", "step": 1500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262020, "t": 819.1} |
| {"event": "train", "step": 1500, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.55055, "bpb": 2.23697, "chunk_loss": 0.030365, "F": 0.19788, "G": 0.22144, "chunks": 202.62, "bytes_per_chunk": 5.054, "boundary_entropy": 0.1536, "router_noise": 0.0, "grad_norm": 0.9516, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.508, "probe_sec": 0.0531, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02408, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 1.02608, "n_shadow": 10, "mean_delta_shadow": 0.14708, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -7.58016, "trunc_abs_frac": 7.58016, "mean_delta_full": -0.04321, "read_hit_rate": {"surface": 0.06138}, "mean_gate": {"surface": 0.15956}, "n_reads": 396, "n_cand": 6452, "read_suppressed_frac": 0.096, "dropout_reads_logged": 396, "inventory": {"size": 327079, "admitted": 65536, "staged": 261543, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261543, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 10.122756958007812, "n_util_obs": 17, "util_count_spearman": 0.41911764705882354}, "probe_overhead_sec_total": 99.991, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.040203, "alpha": 0.03, "t": 820.1} |
| {"event": "consolidate", "step": 1550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262016, "t": 845.9} |
| {"event": "train", "step": 1550, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.54448, "bpb": 2.22822, "chunk_loss": 0.030438, "F": 0.20062, "G": 0.22453, "chunks": 205.44, "bytes_per_chunk": 4.984, "boundary_entropy": 0.15163, "router_noise": 0.0, "grad_norm": 0.8075, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.522, "probe_sec": 0.0355, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0246, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.00852, "n_shadow": 3, "mean_delta_shadow": -0.40858, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.43376, "trunc_abs_frac": 0.74458, "mean_delta_full": 0.53031, "read_hit_rate": {"surface": 0.06619}, "mean_gate": {"surface": 0.15808}, "n_reads": 433, "n_cand": 6542, "read_suppressed_frac": 0.0762, "dropout_reads_logged": 433, "inventory": {"size": 327148, "admitted": 65536, "staged": 261612, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261612, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 10.580108642578125, "n_util_obs": 18, "util_count_spearman": 0.49019607843137253}, "probe_overhead_sec_total": 103.323, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.041448, "alpha": 0.03, "t": 846.2} |
| {"event": "consolidate", "step": 1600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262010, "t": 876.6} |
| {"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3410.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195943, "cps": 26.640625, "mean_p": 0.207198, "boundary_entropy": 0.130064, "bpb": 2.171664}, "tr": {"chunks": 3250.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179528, "cps": 25.390625, "mean_p": 0.243059, "boundary_entropy": 0.281753, "bpb": 2.429204}, "fi": {"chunks": 3343.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177715, "cps": 26.117188, "mean_p": 0.226822, "boundary_entropy": 0.232104, "bpb": 2.603391}}, "bpb": {"en": 2.17166, "tr": 2.4292, "fi": 2.60339}, "gini_cpc": 0.021967, "gini_cps": 0.010663, "bpic": 5.671499, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9162258714106296, "fi": 0.9069727405706183}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 11.204315185546875, "n_util_obs": 19, "util_count_spearman": 0.4614035087719298}, "t": 877.2} |
| {"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.18922, "tr": 2.43981, "fi": 2.61266}, "bytes_per_chunk": {"en": 4.855, "tr": 5.972, "fi": 5.727}, "bpb_mean": 2.4139, "frozen_inventory": true, "t": 878.2} |
| {"event": "train", "step": 1600, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.52661, "bpb": 2.20243, "chunk_loss": 0.030309, "F": 0.19321, "G": 0.22365, "chunks": 197.84, "bytes_per_chunk": 5.176, "boundary_entropy": 0.16919, "router_noise": 0.0, "grad_norm": 0.8842, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.723, "probe_sec": 0.0611, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04908, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.01727, "n_shadow": 10, "mean_delta_shadow": -0.23393, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.09072, "trunc_abs_frac": 2.03585, "mean_delta_full": 0.38676, "read_hit_rate": {"surface": 0.07144}, "mean_gate": {"surface": 0.16072}, "n_reads": 450, "n_cand": 6299, "read_suppressed_frac": 0.12, "dropout_reads_logged": 450, "inventory": {"size": 327145, "admitted": 65536, "staged": 261609, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261609, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 11.211181640625, "n_util_obs": 19, "util_count_spearman": 0.4614035087719298}, "probe_overhead_sec_total": 106.19, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.040162, "alpha": 0.03, "t": 882.3} |
| {"event": "consolidate", "step": 1650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262007, "t": 906.6} |
| {"event": "train", "step": 1650, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5764, "bpb": 2.27426, "chunk_loss": 0.030393, "F": 0.20053, "G": 0.22249, "chunks": 205.34, "bytes_per_chunk": 4.987, "boundary_entropy": 0.15008, "router_noise": 0.0, "grad_norm": 0.7853, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.489, "probe_sec": 0.0448, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0298, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.23781, "n_shadow": 9, "mean_delta_shadow": -0.03334, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -1.30974, "trunc_abs_frac": 1.00255, "mean_delta_full": 0.04614, "read_hit_rate": {"surface": 0.07845}, "mean_gate": {"surface": 0.1591}, "n_reads": 513, "n_cand": 6539, "read_suppressed_frac": 0.1014, "dropout_reads_logged": 513, "inventory": {"size": 327181, "admitted": 65536, "staged": 261645, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261645, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 11.944198608398438, "n_util_obs": 20, "util_count_spearman": 0.3699248120300752}, "probe_overhead_sec_total": 109.368, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.038842, "alpha": 0.03, "t": 906.7} |
| {"event": "consolidate", "step": 1700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262003, "t": 930.8} |
| {"event": "train", "step": 1700, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.57266, "bpb": 2.26886, "chunk_loss": 0.030311, "F": 0.19357, "G": 0.22301, "chunks": 198.22, "bytes_per_chunk": 5.166, "boundary_entropy": 0.1576, "router_noise": 0.0, "grad_norm": 0.8724, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.485, "probe_sec": 0.0369, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02812, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.1058, "n_shadow": 6, "mean_delta_shadow": 0.00213, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.53876, "trunc_abs_frac": 0.8523, "mean_delta_full": 0.38489, "read_hit_rate": {"surface": 0.07019}, "mean_gate": {"surface": 0.15481}, "n_reads": 443, "n_cand": 6311, "read_suppressed_frac": 0.0971, "dropout_reads_logged": 443, "inventory": {"size": 327183, "admitted": 65536, "staged": 261647, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261647, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 12.5771484375, "n_util_obs": 23, "util_count_spearman": 0.5247035573122529}, "probe_overhead_sec_total": 112.025, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.041402, "alpha": 0.03, "t": 931.0} |
| {"event": "consolidate", "step": 1750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262002, "t": 955.1} |
| {"event": "train", "step": 1750, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.53812, "bpb": 2.21903, "chunk_loss": 0.030427, "F": 0.20084, "G": 0.22201, "chunks": 205.66, "bytes_per_chunk": 4.979, "boundary_entropy": 0.13794, "router_noise": 0.0, "grad_norm": 0.8044, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.485, "probe_sec": 0.0292, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02296, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.44426, "n_shadow": 5, "mean_delta_shadow": 0.14808, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 8, "trunc_ratio_sum": 0.71477, "trunc_abs_frac": 0.91951, "mean_delta_full": 0.37852, "read_hit_rate": {"surface": 0.07635}, "mean_gate": {"surface": 0.15718}, "n_reads": 500, "n_cand": 6549, "read_suppressed_frac": 0.138, "dropout_reads_logged": 500, "inventory": {"size": 327129, "admitted": 65536, "staged": 261593, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261593, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 13.627822875976562, "n_util_obs": 26, "util_count_spearman": 0.5391452991452992}, "probe_overhead_sec_total": 115.236, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.042084, "alpha": 0.03, "t": 955.2} |
| {"event": "consolidate", "step": 1800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262001, "t": 985.1} |
| {"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3412.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196058, "cps": 26.65625, "mean_p": 0.20724, "boundary_entropy": 0.121759, "bpb": 2.134981}, "tr": {"chunks": 3515.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194167, "cps": 27.460938, "mean_p": 0.248912, "boundary_entropy": 0.268462, "bpb": 2.370837}, "fi": {"chunks": 3441.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.182925, "cps": 26.882812, "mean_p": 0.220164, "boundary_entropy": 0.214279, "bpb": 2.57665}}, "bpb": {"en": 2.13498, "tr": 2.37084, "fi": 2.57665}, "gini_cpc": 0.015276, "gini_cps": 0.006623, "bpic": 5.471836, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9903526673828585, "fi": 0.933013412309016}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 14.400634765625, "n_util_obs": 27, "util_count_spearman": 0.47741147741147744}, "t": 985.5} |
| {"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.14968, "tr": 2.37903, "fi": 2.58426}, "bytes_per_chunk": {"en": 4.854, "tr": 5.534, "fi": 5.64}, "bpb_mean": 2.37099, "frozen_inventory": true, "t": 986.6} |
| {"event": "train", "step": 1800, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.59165, "bpb": 2.29626, "chunk_loss": 0.030419, "F": 0.19907, "G": 0.22685, "chunks": 203.84, "bytes_per_chunk": 5.023, "boundary_entropy": 0.15917, "router_noise": 0.0, "grad_norm": 0.8836, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.694, "probe_sec": 0.0542, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04747, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.53348, "n_shadow": 5, "mean_delta_shadow": -0.6564, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.07611}, "mean_gate": {"surface": 0.15785}, "n_reads": 494, "n_cand": 6491, "read_suppressed_frac": 0.087, "dropout_reads_logged": 494, "inventory": {"size": 327279, "admitted": 65536, "staged": 261743, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261743, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 14.408172607421875, "n_util_obs": 27, "util_count_spearman": 0.47741147741147744}, "probe_overhead_sec_total": 118.37, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.042563, "alpha": 0.03, "t": 989.9} |
| {"event": "consolidate", "step": 1850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 460, "uncertain_staged": 261994, "t": 1016.7} |
| {"event": "train", "step": 1850, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5373, "bpb": 2.21786, "chunk_loss": 0.030416, "F": 0.20013, "G": 0.2225, "chunks": 204.94, "bytes_per_chunk": 4.997, "boundary_entropy": 0.15007, "router_noise": 0.0, "grad_norm": 0.7796, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.552, "probe_sec": 0.0486, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02429, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.71785, "n_shadow": 5, "mean_delta_shadow": -0.06423, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.00176, "trunc_abs_frac": 1.00176, "mean_delta_full": 1.21143, "read_hit_rate": {"surface": 0.10451}, "mean_gate": {"surface": 0.15319}, "n_reads": 682, "n_cand": 6526, "read_suppressed_frac": 0.1056, "dropout_reads_logged": 682, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 18.625396728515625, "n_util_obs": 47, "util_count_spearman": 0.47895467160037003}, "probe_overhead_sec_total": 121.547, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.0411, "alpha": 0.03, "t": 1017.5} |
| {"event": "consolidate", "step": 1900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262006, "t": 1036.7} |
| {"event": "train", "step": 1900, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.58203, "bpb": 2.28239, "chunk_loss": 0.030364, "F": 0.19754, "G": 0.21938, "chunks": 202.28, "bytes_per_chunk": 5.062, "boundary_entropy": 0.13641, "router_noise": 0.0, "grad_norm": 0.764, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.387, "probe_sec": 0.0425, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03469, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.11501, "n_shadow": 4, "mean_delta_shadow": -0.38257, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.54329, "trunc_abs_frac": 0.95211, "mean_delta_full": 0.26619, "read_hit_rate": {"surface": 0.10092}, "mean_gate": {"surface": 0.15111}, "n_reads": 650, "n_cand": 6441, "read_suppressed_frac": 0.1123, "dropout_reads_logged": 650, "inventory": {"size": 327192, "admitted": 65536, "staged": 261656, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261656, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 19.205612182617188, "n_util_obs": 47, "util_count_spearman": 0.343432007400555}, "probe_overhead_sec_total": 124.199, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.040678, "alpha": 0.03, "t": 1036.9} |
| {"event": "consolidate", "step": 1950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262002, "t": 1066.1} |
| {"event": "train", "step": 1950, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.53835, "bpb": 2.21937, "chunk_loss": 0.030427, "F": 0.19986, "G": 0.22156, "chunks": 204.66, "bytes_per_chunk": 5.004, "boundary_entropy": 0.14129, "router_noise": 0.0, "grad_norm": 0.791, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.606, "probe_sec": 0.0592, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02419, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.09728, "n_shadow": 4, "mean_delta_shadow": -0.67575, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.91365, "trunc_abs_frac": 1.12952, "mean_delta_full": -0.81494, "read_hit_rate": {"surface": 0.11063}, "mean_gate": {"surface": 0.14534}, "n_reads": 721, "n_cand": 6517, "read_suppressed_frac": 0.0985, "dropout_reads_logged": 721, "inventory": {"size": 327189, "admitted": 65536, "staged": 261653, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261653, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 19.615005493164062, "n_util_obs": 47, "util_count_spearman": 0.3319842738205365}, "probe_overhead_sec_total": 126.935, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.043934, "alpha": 0.03, "t": 1067.2} |
| {"event": "consolidate", "step": 2000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261998, "t": 1109.6} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3395.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195081, "cps": 26.523438, "mean_p": 0.204662, "boundary_entropy": 0.112682, "bpb": 2.101972}, "tr": {"chunks": 3300.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18229, "cps": 25.78125, "mean_p": 0.239307, "boundary_entropy": 0.258007, "bpb": 2.331479}, "fi": {"chunks": 3316.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17628, "cps": 25.90625, "mean_p": 0.213017, "boundary_entropy": 0.207197, "bpb": 2.507814}}, "bpb": {"en": 2.10197, "tr": 2.33148, "fi": 2.50781}, "gini_cpc": 0.022639, "gini_cps": 0.006326, "bpic": 5.666966, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9344320589993262, "fi": 0.9036223830743598}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 22.043869018554688, "n_util_obs": 55, "util_count_spearman": 0.3549486823932837}, "t": 1110.1} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.11751, "tr": 2.33843, "fi": 2.51659}, "bytes_per_chunk": {"en": 4.862, "tr": 5.915, "fi": 5.748}, "bpb_mean": 2.32418, "frozen_inventory": true, "t": 1110.9} |
| {"event": "train", "step": 2000, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.55717, "bpb": 2.24652, "chunk_loss": 0.030319, "F": 0.19476, "G": 0.21816, "chunks": 199.44, "bytes_per_chunk": 5.134, "boundary_entropy": 0.1443, "router_noise": 0.0, "grad_norm": 0.8296, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.959, "probe_sec": 0.0773, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04103, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.21234, "n_shadow": 5, "mean_delta_shadow": 0.01187, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.73899, "trunc_abs_frac": 0.55317, "mean_delta_full": 1.18909, "read_hit_rate": {"surface": 0.11071}, "mean_gate": {"surface": 0.14134}, "n_reads": 703, "n_cand": 6350, "read_suppressed_frac": 0.1038, "dropout_reads_logged": 703, "inventory": {"size": 327272, "admitted": 65536, "staged": 261736, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261736, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 22.054595947265625, "n_util_obs": 55, "util_count_spearman": 0.3577200577200577}, "probe_overhead_sec_total": 129.616, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.040674, "alpha": 0.03, "t": 1115.1} |
| {"event": "consolidate", "step": 2050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262003, "t": 1137.8} |
| {"event": "train", "step": 2050, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.61912, "bpb": 2.3359, "chunk_loss": 0.030334, "F": 0.19391, "G": 0.22161, "chunks": 198.56, "bytes_per_chunk": 5.157, "boundary_entropy": 0.14902, "router_noise": 0.0, "grad_norm": 0.9141, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.458, "probe_sec": 0.1115, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0995, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.25445, "n_shadow": 5, "mean_delta_shadow": -0.0165, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -1.71463, "trunc_abs_frac": 1.71463, "mean_delta_full": 0.0509, "read_hit_rate": {"surface": 0.10898}, "mean_gate": {"surface": 0.14049}, "n_reads": 689, "n_cand": 6322, "read_suppressed_frac": 0.0914, "dropout_reads_logged": 689, "inventory": {"size": 327253, "admitted": 65536, "staged": 261717, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261717, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 23.47344970703125, "n_util_obs": 56, "util_count_spearman": 0.4779220779220779}, "probe_overhead_sec_total": 132.681, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.03964, "alpha": 0.03, "t": 1138.0} |
| {"event": "consolidate", "step": 2100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261999, "t": 1170.8} |
| {"event": "train", "step": 2100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.50226, "bpb": 2.1673, "chunk_loss": 0.030353, "F": 0.19696, "G": 0.2206, "chunks": 201.69, "bytes_per_chunk": 5.077, "boundary_entropy": 0.13713, "router_noise": 0.0, "grad_norm": 0.8029, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.66, "probe_sec": 0.0851, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07211, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.16041, "n_shadow": 4, "mean_delta_shadow": -0.04459, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.67805, "trunc_abs_frac": 1.63446, "mean_delta_full": -0.35883, "read_hit_rate": {"surface": 0.12224}, "mean_gate": {"surface": 0.14027}, "n_reads": 785, "n_cand": 6422, "read_suppressed_frac": 0.1096, "dropout_reads_logged": 785, "inventory": {"size": 327142, "admitted": 65536, "staged": 261606, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261606, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 24.667877197265625, "n_util_obs": 57, "util_count_spearman": 0.4342105263157895}, "probe_overhead_sec_total": 135.721, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.04273, "alpha": 0.03, "t": 1171.0} |
| {"event": "consolidate", "step": 2150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 408, "uncertain_staged": 261992, "t": 1196.1} |
| {"event": "train", "step": 2150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.50885, "bpb": 2.17682, "chunk_loss": 0.03032, "F": 0.19504, "G": 0.2184, "chunks": 199.72, "bytes_per_chunk": 5.127, "boundary_entropy": 0.13349, "router_noise": 0.0, "grad_norm": 0.6903, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.516, "probe_sec": 0.0461, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0242, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.17825, "n_shadow": 7, "mean_delta_shadow": 0.09945, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.53077, "trunc_abs_frac": 1.15379, "mean_delta_full": -0.24792, "read_hit_rate": {"surface": 0.11669}, "mean_gate": {"surface": 0.13587}, "n_reads": 742, "n_cand": 6359, "read_suppressed_frac": 0.0889, "dropout_reads_logged": 742, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 26.259811401367188, "n_util_obs": 58, "util_count_spearman": 0.474053339075333}, "probe_overhead_sec_total": 138.841, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.03915, "alpha": 0.03, "t": 1196.8} |
| {"event": "consolidate", "step": 2200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 17, "uncertain_staged": 261993, "t": 1226.7} |
| {"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3393.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194966, "cps": 26.507812, "mean_p": 0.203609, "boundary_entropy": 0.108046, "bpb": 2.072766}, "tr": {"chunks": 3182.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175772, "cps": 24.859375, "mean_p": 0.234811, "boundary_entropy": 0.250688, "bpb": 2.297614}, "fi": {"chunks": 3259.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17325, "cps": 25.460938, "mean_p": 0.204342, "boundary_entropy": 0.188398, "bpb": 2.472604}}, "bpb": {"en": 2.07277, "tr": 2.29761, "fi": 2.4726}, "gini_cpc": 0.026614, "gini_cps": 0.014304, "bpic": 5.768965, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9015501385064821, "fi": 0.8886131536653333}, "inventory": {"size": 327663, "admitted": 65536, "staged": 262127, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262127, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 26.846405029296875, "n_util_obs": 58, "util_count_spearman": 0.47589898182041895}, "t": 1227.3} |
| {"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.09134, "tr": 2.3035, "fi": 2.48033}, "bytes_per_chunk": {"en": 4.868, "tr": 5.969, "fi": 5.806}, "bpb_mean": 2.29173, "frozen_inventory": true, "t": 1228.3} |
| {"event": "train", "step": 2200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.48453, "bpb": 2.14173, "chunk_loss": 0.030366, "F": 0.19907, "G": 0.2169, "chunks": 203.84, "bytes_per_chunk": 5.023, "boundary_entropy": 0.12489, "router_noise": 0.0, "grad_norm": 0.7346, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.732, "probe_sec": 0.0509, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02744, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.3777, "n_shadow": 8, "mean_delta_shadow": -0.38628, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.84084, "trunc_abs_frac": 1.13691, "mean_delta_full": -0.22415, "read_hit_rate": {"surface": 0.13989}, "mean_gate": {"surface": 0.13765}, "n_reads": 908, "n_cand": 6491, "read_suppressed_frac": 0.0925, "dropout_reads_logged": 908, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 26.860260009765625, "n_util_obs": 58, "util_count_spearman": 0.47620658894459994}, "probe_overhead_sec_total": 141.94, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.038559, "alpha": 0.03, "t": 1233.5} |
| {"event": "consolidate", "step": 2250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 4, "uncertain_staged": 261988, "t": 1263.6} |
| {"event": "train", "step": 2250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5179, "bpb": 2.18986, "chunk_loss": 0.030303, "F": 0.1944, "G": 0.21689, "chunks": 199.06, "bytes_per_chunk": 5.144, "boundary_entropy": 0.13426, "router_noise": 0.0, "grad_norm": 0.75, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.616, "probe_sec": 0.096, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0805, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.0408, "n_shadow": 9, "mean_delta_shadow": -0.63233, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.92369, "trunc_abs_frac": 0.85643, "mean_delta_full": -0.08691, "read_hit_rate": {"surface": 0.12291}, "mean_gate": {"surface": 0.13897}, "n_reads": 779, "n_cand": 6338, "read_suppressed_frac": 0.086, "dropout_reads_logged": 779, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.550262451171875, "n_util_obs": 59, "util_count_spearman": 0.44786674459380477}, "probe_overhead_sec_total": 145.164, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.039266, "alpha": 0.03, "t": 1264.3} |
| {"event": "consolidate", "step": 2300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 34, "uncertain_staged": 261984, "t": 1294.4} |
| {"event": "train", "step": 2300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.50097, "bpb": 2.16544, "chunk_loss": 0.030382, "F": 0.19968, "G": 0.2179, "chunks": 204.47, "bytes_per_chunk": 5.008, "boundary_entropy": 0.12274, "router_noise": 0.0, "grad_norm": 0.7525, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.608, "probe_sec": 0.0283, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02154, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.29762, "n_shadow": 7, "mean_delta_shadow": 0.08123, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 3.50368, "trunc_abs_frac": 15.53102, "mean_delta_full": 0.05308, "read_hit_rate": {"surface": 0.13101}, "mean_gate": {"surface": 0.14206}, "n_reads": 853, "n_cand": 6511, "read_suppressed_frac": 0.1079, "dropout_reads_logged": 853, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 28.114181518554688, "n_util_obs": 61, "util_count_spearman": 0.4251718667371761}, "probe_overhead_sec_total": 148.04, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.037321, "alpha": 0.03, "t": 1294.7} |
| {"event": "consolidate", "step": 2350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 9, "uncertain_staged": 261980, "t": 1318.5} |
| {"event": "train", "step": 2350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.50221, "bpb": 2.16723, "chunk_loss": 0.030252, "F": 0.18848, "G": 0.2227, "chunks": 193.0, "bytes_per_chunk": 5.306, "boundary_entropy": 0.15814, "router_noise": 0.0, "grad_norm": 0.812, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.501, "probe_sec": 0.1182, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07605, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.03086, "n_shadow": 5, "mean_delta_shadow": 0.03236, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -0.21005, "trunc_abs_frac": 0.21005, "mean_delta_full": 0.09473, "read_hit_rate": {"surface": 0.125}, "mean_gate": {"surface": 0.13667}, "n_reads": 768, "n_cand": 6144, "read_suppressed_frac": 0.0898, "dropout_reads_logged": 768, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 29.230438232421875, "n_util_obs": 65, "util_count_spearman": 0.39606643356643356}, "probe_overhead_sec_total": 151.779, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.037367, "alpha": 0.03, "t": 1319.7} |
| {"event": "consolidate", "step": 2400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 227, "uncertain_staged": 261979, "t": 1370.1} |
| {"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3387.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194622, "cps": 26.460938, "mean_p": 0.202436, "boundary_entropy": 0.101574, "bpb": 2.05165}, "tr": {"chunks": 3206.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177098, "cps": 25.046875, "mean_p": 0.231484, "boundary_entropy": 0.242314, "bpb": 2.266655}, "fi": {"chunks": 3282.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.174472, "cps": 25.640625, "mean_p": 0.209951, "boundary_entropy": 0.192369, "bpb": 2.435448}}, "bpb": {"en": 2.05165, "tr": 2.26665, "fi": 2.43545}, "gini_cpc": 0.024594, "gini_cps": 0.012219, "bpic": 5.745013, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9099591369863824, "fi": 0.8964697031244415}, "inventory": {"size": 327453, "admitted": 65536, "staged": 261917, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261917, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 31.3809814453125, "n_util_obs": 65, "util_count_spearman": 0.3814248251748252}, "t": 1370.8} |
| {"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.06699, "tr": 2.27035, "fi": 2.44719}, "bytes_per_chunk": {"en": 4.868, "tr": 5.975, "fi": 5.742}, "bpb_mean": 2.26151, "frozen_inventory": true, "t": 1371.7} |
| {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1376.8} |
| {"event": "train", "step": 2400, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.45467, "bpb": 2.09865, "chunk_loss": 0.03044, "F": 0.20303, "G": 0.21966, "chunks": 207.91, "bytes_per_chunk": 4.925, "boundary_entropy": 0.12223, "router_noise": 0.0, "grad_norm": 0.7467, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 1.176, "probe_sec": 0.0808, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05079, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.10508, "n_shadow": 7, "mean_delta_shadow": -0.14005, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.04888, "trunc_abs_frac": 0.76929, "mean_delta_full": -0.48531, "read_hit_rate": {"surface": 0.13593}, "mean_gate": {"surface": 0.13401}, "n_reads": 900, "n_cand": 6621, "read_suppressed_frac": 0.0933, "dropout_reads_logged": 900, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 31.39471435546875, "n_util_obs": 65, "util_count_spearman": 0.3814248251748252}, "probe_overhead_sec_total": 155.448, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.041233, "alpha": 0.03, "t": 1378.5} |
| {"event": "consolidate", "step": 2450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 12, "uncertain_staged": 261977, "t": 1432.3} |
| {"event": "train", "step": 2450, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.64129, "bpb": 2.36788, "chunk_loss": 0.030388, "F": 0.1991, "G": 0.22029, "chunks": 203.88, "bytes_per_chunk": 5.023, "boundary_entropy": 0.12635, "router_noise": 0.0, "grad_norm": 0.7134, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 1.101, "probe_sec": 0.0924, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07007, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.0154, "n_shadow": 7, "mean_delta_shadow": 0.02569, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.66846, "trunc_abs_frac": 0.91295, "mean_delta_full": -0.11808, "read_hit_rate": {"surface": 0.13262}, "mean_gate": {"surface": 0.13649}, "n_reads": 861, "n_cand": 6492, "read_suppressed_frac": 0.1022, "dropout_reads_logged": 861, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 32.03564453125, "n_util_obs": 66, "util_count_spearman": 0.43077372691759735}, "probe_overhead_sec_total": 159.156, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.035977, "alpha": 0.03, "t": 1433.5} |
| {"event": "consolidate", "step": 2500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 19, "uncertain_staged": 261975, "t": 1490.1} |
| {"event": "train", "step": 2500, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.48916, "bpb": 2.1484, "chunk_loss": 0.030361, "F": 0.19867, "G": 0.21683, "chunks": 203.44, "bytes_per_chunk": 5.033, "boundary_entropy": 0.12114, "router_noise": 0.0, "grad_norm": 0.7271, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 1.152, "probe_sec": 0.1191, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08472, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.12186, "n_shadow": 7, "mean_delta_shadow": 0.08689, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.37471, "trunc_abs_frac": 6.60061, "mean_delta_full": 0.02629, "read_hit_rate": {"surface": 0.14202}, "mean_gate": {"surface": 0.14175}, "n_reads": 920, "n_cand": 6478, "read_suppressed_frac": 0.1054, "dropout_reads_logged": 920, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 32.76896667480469, "n_util_obs": 67, "util_count_spearman": 0.4490781387181738}, "probe_overhead_sec_total": 163.266, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.038406, "alpha": 0.03, "t": 1491.1} |
| {"event": "consolidate", "step": 2550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 145, "uncertain_staged": 261973, "t": 1544.8} |
| {"event": "train", "step": 2550, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.48858, "bpb": 2.14757, "chunk_loss": 0.030388, "F": 0.19855, "G": 0.22131, "chunks": 203.31, "bytes_per_chunk": 5.037, "boundary_entropy": 0.13767, "router_noise": 0.0, "grad_norm": 0.7219, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 1.091, "probe_sec": 0.0833, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04378, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.02307, "n_shadow": 4, "mean_delta_shadow": -0.30329, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.27208, "trunc_abs_frac": 8.8334, "mean_delta_full": -0.03225, "read_hit_rate": {"surface": 0.14473}, "mean_gate": {"surface": 0.13955}, "n_reads": 937, "n_cand": 6474, "read_suppressed_frac": 0.1067, "dropout_reads_logged": 937, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 33.88804626464844, "n_util_obs": 69, "util_count_spearman": 0.46558275484106687}, "probe_overhead_sec_total": 167.055, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.036272, "alpha": 0.03, "t": 1545.7} |
| {"event": "consolidate", "step": 2600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 8, "uncertain_staged": 261966, "t": 1599.5} |
| {"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3365.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193357, "cps": 26.289062, "mean_p": 0.200456, "boundary_entropy": 0.094895, "bpb": 2.031464}, "tr": {"chunks": 3178.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175551, "cps": 24.828125, "mean_p": 0.231552, "boundary_entropy": 0.241161, "bpb": 2.249068}, "fi": {"chunks": 3278.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17426, "cps": 25.609375, "mean_p": 0.208797, "boundary_entropy": 0.186985, "bpb": 2.412541}}, "bpb": {"en": 2.03146, "tr": 2.24907, "fi": 2.41254}, "gini_cpc": 0.02344, "gini_cps": 0.012694, "bpic": 5.776601, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9079091502077554, "fi": 0.9012309907192079}, "inventory": {"size": 327672, "admitted": 65536, "staged": 262136, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262136, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 34.926422119140625, "n_util_obs": 75, "util_count_spearman": 0.47613086770981505}, "t": 1600.1} |
| {"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.04718, "tr": 2.25297, "fi": 2.42329}, "bytes_per_chunk": {"en": 4.905, "tr": 6.013, "fi": 5.755}, "bpb_mean": 2.24115, "frozen_inventory": true, "t": 1601.0} |
| {"event": "train", "step": 2600, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.51285, "bpb": 2.18259, "chunk_loss": 0.030268, "F": 0.19028, "G": 0.21328, "chunks": 194.84, "bytes_per_chunk": 5.255, "boundary_entropy": 0.12926, "router_noise": 0.0, "grad_norm": 0.7679, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 1.238, "probe_sec": 0.0791, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03849, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.26566, "n_shadow": 8, "mean_delta_shadow": -0.07704, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -6.92902, "trunc_abs_frac": 3.4668, "mean_delta_full": 0.0129, "read_hit_rate": {"surface": 0.13897}, "mean_gate": {"surface": 0.13581}, "n_reads": 862, "n_cand": 6203, "read_suppressed_frac": 0.0974, "dropout_reads_logged": 862, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 34.9395751953125, "n_util_obs": 75, "util_count_spearman": 0.4784352773826458}, "probe_overhead_sec_total": 170.618, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.035717, "alpha": 0.03, "t": 1607.6} |
| {"event": "consolidate", "step": 2650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 899, "uncertain_staged": 261967, "t": 1665.0} |
| {"event": "train", "step": 2650, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47018, "bpb": 2.12102, "chunk_loss": 0.030293, "F": 0.19119, "G": 0.219, "chunks": 195.78, "bytes_per_chunk": 5.23, "boundary_entropy": 0.13915, "router_noise": 0.0, "grad_norm": 0.7822, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 1.179, "probe_sec": 0.0552, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03309, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.07417, "n_shadow": 9, "mean_delta_shadow": -0.4775, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.4301, "trunc_abs_frac": 2.03713, "mean_delta_full": -0.07832, "read_hit_rate": {"surface": 0.15434}, "mean_gate": {"surface": 0.13178}, "n_reads": 962, "n_cand": 6233, "read_suppressed_frac": 0.0894, "dropout_reads_logged": 962, "inventory": {"size": 327363, "admitted": 65536, "staged": 261827, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261827, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 36.80601501464844, "n_util_obs": 76, "util_count_spearman": 0.5140123034859877}, "probe_overhead_sec_total": 174.281, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.034055, "alpha": 0.03, "t": 1666.6} |
| {"event": "consolidate", "step": 2700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 11, "uncertain_staged": 261966, "t": 1743.7} |
| {"event": "train", "step": 2700, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.54783, "bpb": 2.23305, "chunk_loss": 0.030375, "F": 0.19708, "G": 0.22089, "chunks": 201.81, "bytes_per_chunk": 5.074, "boundary_entropy": 0.12844, "router_noise": 0.0, "grad_norm": 0.8217, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 1.575, "probe_sec": 0.061, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02236, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.01033, "n_shadow": 6, "mean_delta_shadow": -0.05275, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.61695, "trunc_abs_frac": 0.88996, "mean_delta_full": -0.0665, "read_hit_rate": {"surface": 0.13943}, "mean_gate": {"surface": 0.13068}, "n_reads": 896, "n_cand": 6426, "read_suppressed_frac": 0.0926, "dropout_reads_logged": 896, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 37.64225769042969, "n_util_obs": 78, "util_count_spearman": 0.4748193251104984}, "probe_overhead_sec_total": 177.372, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.037094, "alpha": 0.03, "t": 1745.3} |
| {"event": "consolidate", "step": 2750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 918, "uncertain_staged": 261964, "t": 1794.7} |
| {"event": "train", "step": 2750, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.48293, "bpb": 2.13942, "chunk_loss": 0.030295, "F": 0.19373, "G": 0.21765, "chunks": 198.38, "bytes_per_chunk": 5.162, "boundary_entropy": 0.12979, "router_noise": 0.0, "grad_norm": 0.7411, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.992, "probe_sec": 0.0518, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04461, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.34272, "n_shadow": 6, "mean_delta_shadow": -0.01135, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.29004, "trunc_abs_frac": 0.84132, "mean_delta_full": 0.25077, "read_hit_rate": {"surface": 0.14693}, "mean_gate": {"surface": 0.13224}, "n_reads": 928, "n_cand": 6316, "read_suppressed_frac": 0.1175, "dropout_reads_logged": 928, "inventory": {"size": 327283, "admitted": 65536, "staged": 261747, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261747, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 38.33915710449219, "n_util_obs": 75, "util_count_spearman": 0.4145945945945946}, "probe_overhead_sec_total": 180.71, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.037104, "alpha": 0.03, "t": 1794.9} |
| {"event": "consolidate", "step": 2800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261961, "t": 1844.2} |
| {"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3390.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194794, "cps": 26.484375, "mean_p": 0.20308, "boundary_entropy": 0.095678, "bpb": 2.011265}, "tr": {"chunks": 3182.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175772, "cps": 24.859375, "mean_p": 0.229611, "boundary_entropy": 0.232936, "bpb": 2.218086}, "fi": {"chunks": 3343.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177715, "cps": 26.117188, "mean_p": 0.206364, "boundary_entropy": 0.181095, "bpb": 2.372617}}, "bpb": {"en": 2.01127, "tr": 2.21809, "fi": 2.37262}, "gini_cpc": 0.023129, "gini_cps": 0.013986, "bpic": 5.721836, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9023479704874613, "fi": 0.9123236121964036}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 35.36517333984375, "n_util_obs": 73, "util_count_spearman": 0.46569171911637663}, "t": 1844.7} |
| {"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.02767, "tr": 2.22451, "fi": 2.38245}, "bytes_per_chunk": {"en": 4.865, "tr": 5.979, "fi": 5.651}, "bpb_mean": 2.21155, "frozen_inventory": true, "t": 1845.5} |
| {"event": "train", "step": 2800, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46294, "bpb": 2.11058, "chunk_loss": 0.030384, "F": 0.20044, "G": 0.21622, "chunks": 205.25, "bytes_per_chunk": 4.989, "boundary_entropy": 0.11279, "router_noise": 0.0, "grad_norm": 0.6896, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 1.106, "probe_sec": 0.0487, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03183, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.43953, "n_shadow": 9, "mean_delta_shadow": 0.18161, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.2573, "trunc_abs_frac": 0.9597, "mean_delta_full": -0.18542, "read_hit_rate": {"surface": 0.1302}, "mean_gate": {"surface": 0.13843}, "n_reads": 851, "n_cand": 6536, "read_suppressed_frac": 0.0917, "dropout_reads_logged": 851, "inventory": {"size": 327179, "admitted": 65536, "staged": 261643, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261643, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 35.37815856933594, "n_util_obs": 73, "util_count_spearman": 0.4664321856102678}, "probe_overhead_sec_total": 183.6, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.03509, "alpha": 0.03, "t": 1850.2} |
| {"event": "consolidate", "step": 2850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261955, "t": 1896.6} |
| {"event": "train", "step": 2850, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.54816, "bpb": 2.23353, "chunk_loss": 0.030384, "F": 0.19894, "G": 0.21601, "chunks": 203.72, "bytes_per_chunk": 5.027, "boundary_entropy": 0.11589, "router_noise": 0.0, "grad_norm": 0.727, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.935, "probe_sec": 0.0399, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02226, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.23896, "n_shadow": 5, "mean_delta_shadow": 0.02281, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.14483, "trunc_abs_frac": 0.75166, "mean_delta_full": 0.09773, "read_hit_rate": {"surface": 0.12517}, "mean_gate": {"surface": 0.13988}, "n_reads": 812, "n_cand": 6487, "read_suppressed_frac": 0.0961, "dropout_reads_logged": 812, "inventory": {"size": 327284, "admitted": 65536, "staged": 261748, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261748, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 34.64015197753906, "n_util_obs": 73, "util_count_spearman": 0.46593854128100703}, "probe_overhead_sec_total": 186.216, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.033491, "alpha": 0.03, "t": 1897.0} |
| {"event": "consolidate", "step": 2900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261953, "t": 1944.9} |
| {"event": "train", "step": 2900, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42168, "bpb": 2.05105, "chunk_loss": 0.030304, "F": 0.19412, "G": 0.21751, "chunks": 198.78, "bytes_per_chunk": 5.151, "boundary_entropy": 0.12795, "router_noise": 0.0, "grad_norm": 0.701, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.997, "probe_sec": 0.0728, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02448, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.01812, "n_shadow": 6, "mean_delta_shadow": -0.30574, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -2.42823, "trunc_abs_frac": 0.83779, "mean_delta_full": 0.03561, "read_hit_rate": {"surface": 0.11518}, "mean_gate": {"surface": 0.14046}, "n_reads": 729, "n_cand": 6329, "read_suppressed_frac": 0.0919, "dropout_reads_logged": 729, "inventory": {"size": 327205, "admitted": 65536, "staged": 261669, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261669, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 33.30628967285156, "n_util_obs": 75, "util_count_spearman": 0.43563609984103174}, "probe_overhead_sec_total": 188.961, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.038767, "alpha": 0.03, "t": 1946.8} |
| {"event": "consolidate", "step": 2950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261950, "t": 1986.5} |
| {"event": "train", "step": 2950, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.36515, "bpb": 1.9695, "chunk_loss": 0.030464, "F": 0.20563, "G": 0.21889, "chunks": 210.56, "bytes_per_chunk": 4.863, "boundary_entropy": 0.10248, "router_noise": 0.0, "grad_norm": 0.6679, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.82, "probe_sec": 0.0533, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01917, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.00836, "n_shadow": 7, "mean_delta_shadow": 0.01015, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.42838, "trunc_abs_frac": 0.42838, "mean_delta_full": 0.22668, "read_hit_rate": {"surface": 0.1266}, "mean_gate": {"surface": 0.14535}, "n_reads": 849, "n_cand": 6706, "read_suppressed_frac": 0.1025, "dropout_reads_logged": 849, "inventory": {"size": 327059, "admitted": 65536, "staged": 261523, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261523, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 32.652679443359375, "n_util_obs": 73, "util_count_spearman": 0.4128717758854745}, "probe_overhead_sec_total": 191.585, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.036059, "alpha": 0.03, "t": 1987.8} |
| {"event": "consolidate", "step": 3000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261948, "t": 2036.5} |
| {"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3385.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194507, "cps": 26.445312, "mean_p": 0.201165, "boundary_entropy": 0.089452, "bpb": 1.997944}, "tr": {"chunks": 3199.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176711, "cps": 24.992188, "mean_p": 0.229718, "boundary_entropy": 0.234001, "bpb": 2.197864}, "fi": {"chunks": 3302.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.175536, "cps": 25.796875, "mean_p": 0.206922, "boundary_entropy": 0.176493, "bpb": 2.349014}}, "bpb": {"en": 1.99794, "tr": 2.19786, "fi": 2.34901}, "gini_cpc": 0.023132, "gini_cps": 0.012543, "bpic": 5.73862, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9085087947834364, "fi": 0.9024655503823424}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 31.801589965820312, "n_util_obs": 74, "util_count_spearman": 0.4183783783783784}, "t": 2037.1} |
| {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.01241, "tr": 2.20432, "fi": 2.35759}, "bytes_per_chunk": {"en": 4.875, "tr": 5.965, "fi": 5.684}, "bpb_mean": 2.19144, "frozen_inventory": true, "t": 2037.9} |
| {"event": "train", "step": 3000, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.45859, "bpb": 2.1043, "chunk_loss": 0.030317, "F": 0.19525, "G": 0.21609, "chunks": 199.94, "bytes_per_chunk": 5.122, "boundary_entropy": 0.12322, "router_noise": 0.0, "grad_norm": 0.8049, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 1.161, "probe_sec": 0.1151, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05429, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.32038, "n_shadow": 7, "mean_delta_shadow": 0.01328, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.26956, "trunc_abs_frac": 0.34649, "mean_delta_full": 0.11194, "read_hit_rate": {"surface": 0.10776}, "mean_gate": {"surface": 0.14028}, "n_reads": 686, "n_cand": 6366, "read_suppressed_frac": 0.0948, "dropout_reads_logged": 686, "inventory": {"size": 327215, "admitted": 65536, "staged": 261679, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261679, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 31.812057495117188, "n_util_obs": 74, "util_count_spearman": 0.4178748611625324}, "probe_overhead_sec_total": 194.479, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.037775, "alpha": 0.03, "t": 2045.8} |
| {"event": "consolidate", "step": 3050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261942, "t": 2084.6} |
| {"event": "train", "step": 3050, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42074, "bpb": 2.04969, "chunk_loss": 0.03034, "F": 0.1983, "G": 0.21505, "chunks": 203.06, "bytes_per_chunk": 5.043, "boundary_entropy": 0.10753, "router_noise": 0.0, "grad_norm": 0.7078, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.823, "probe_sec": 0.1163, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07006, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.16589, "n_shadow": 7, "mean_delta_shadow": -0.17902, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.37946, "trunc_abs_frac": 0.8872, "mean_delta_full": 0.2059, "read_hit_rate": {"surface": 0.09604}, "mean_gate": {"surface": 0.14787}, "n_reads": 621, "n_cand": 6466, "read_suppressed_frac": 0.1256, "dropout_reads_logged": 621, "inventory": {"size": 327127, "admitted": 65536, "staged": 261591, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261591, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.577804565429688, "n_util_obs": 60, "util_count_spearman": 0.5683245345929425}, "probe_overhead_sec_total": 196.991, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.033621, "alpha": 0.03, "t": 2087.0} |
| {"event": "consolidate", "step": 3100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261936, "t": 2129.7} |
| {"event": "train", "step": 3100, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.5127, "bpb": 2.18236, "chunk_loss": 0.030238, "F": 0.19016, "G": 0.21089, "chunks": 194.72, "bytes_per_chunk": 5.259, "boundary_entropy": 0.11624, "router_noise": 0.0, "grad_norm": 0.8515, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.864, "probe_sec": 0.032, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01989, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.09128, "n_shadow": 10, "mean_delta_shadow": 0.0211, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.61346, "trunc_abs_frac": 1.03163, "mean_delta_full": -0.04788, "read_hit_rate": {"surface": 0.08808}, "mean_gate": {"surface": 0.15247}, "n_reads": 546, "n_cand": 6199, "read_suppressed_frac": 0.0934, "dropout_reads_logged": 546, "inventory": {"size": 327241, "admitted": 65536, "staged": 261705, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261705, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 26.140426635742188, "n_util_obs": 53, "util_count_spearman": 0.6705370101596516}, "probe_overhead_sec_total": 199.535, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.039881, "alpha": 0.03, "t": 2130.2} |
| {"event": "consolidate", "step": 3150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261932, "t": 2167.5} |
| {"event": "train", "step": 3150, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41825, "bpb": 2.04611, "chunk_loss": 0.030261, "F": 0.19293, "G": 0.20777, "chunks": 197.56, "bytes_per_chunk": 5.183, "boundary_entropy": 0.10574, "router_noise": 0.0, "grad_norm": 0.7267, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.75, "probe_sec": 0.0254, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01933, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.60209, "n_shadow": 6, "mean_delta_shadow": 0.05794, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.3077, "trunc_abs_frac": 1.74588, "mean_delta_full": 0.20872, "read_hit_rate": {"surface": 0.0965}, "mean_gate": {"surface": 0.1487}, "n_reads": 607, "n_cand": 6290, "read_suppressed_frac": 0.084, "dropout_reads_logged": 607, "inventory": {"size": 327134, "admitted": 65536, "staged": 261598, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261598, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.026351928710938, "n_util_obs": 58, "util_count_spearman": 0.6663077916884556}, "probe_overhead_sec_total": 202.386, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.036711, "alpha": 0.03, "t": 2167.7} |
| {"event": "consolidate", "step": 3200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261934, "t": 2220.7} |
| {"event": "probe", "step": 3200, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3372.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19376, "cps": 26.34375, "mean_p": 0.201333, "boundary_entropy": 0.085456, "bpb": 1.978385}, "tr": {"chunks": 3260.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180081, "cps": 25.46875, "mean_p": 0.230283, "boundary_entropy": 0.229524, "bpb": 2.167353}, "fi": {"chunks": 3340.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177556, "cps": 26.09375, "mean_p": 0.206438, "boundary_entropy": 0.171971, "bpb": 2.327972}}, "bpb": {"en": 1.97838, "tr": 2.16735, "fi": 2.32797}, "gini_cpc": 0.019591, "gini_cps": 0.007488, "bpic": 5.68913, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9294020003762574, "fi": 0.9163705797187267}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.064102172851562, "n_util_obs": 56, "util_count_spearman": 0.6380724538619276}, "t": 2221.2} |
| {"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 1.99394, "tr": 2.17539, "fi": 2.33878}, "bytes_per_chunk": {"en": 4.887, "tr": 5.85, "fi": 5.656}, "bpb_mean": 2.16937, "frozen_inventory": true, "t": 2222.0} |
| {"event": "train", "step": 3200, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.3464, "bpb": 1.94244, "chunk_loss": 0.030349, "F": 0.2002, "G": 0.2118, "chunks": 205.0, "bytes_per_chunk": 4.995, "boundary_entropy": 0.09259, "router_noise": 0.0, "grad_norm": 0.7059, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 1.204, "probe_sec": 0.1012, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06431, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.11375, "n_shadow": 7, "mean_delta_shadow": -0.30098, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.36716, "trunc_abs_frac": 0.88203, "mean_delta_full": -0.12654, "read_hit_rate": {"surface": 0.09298}, "mean_gate": {"surface": 0.14788}, "n_reads": 607, "n_cand": 6528, "read_suppressed_frac": 0.117, "dropout_reads_logged": 607, "inventory": {"size": 327000, "admitted": 65536, "staged": 261464, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261464, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.0733642578125, "n_util_obs": 56, "util_count_spearman": 0.6380724538619276}, "probe_overhead_sec_total": 205.55, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.035045, "alpha": 0.03, "t": 2227.9} |
| {"event": "consolidate", "step": 3250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261934, "t": 2277.2} |
| {"event": "train", "step": 3250, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.34976, "bpb": 1.94729, "chunk_loss": 0.03036, "F": 0.19839, "G": 0.21469, "chunks": 203.16, "bytes_per_chunk": 5.04, "boundary_entropy": 0.10758, "router_noise": 0.0, "grad_norm": 0.6487, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 1.006, "probe_sec": 0.0425, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02436, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.37281, "n_shadow": 4, "mean_delta_shadow": -0.21697, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.72338, "trunc_abs_frac": 1.33719, "mean_delta_full": 0.32158, "read_hit_rate": {"surface": 0.08471}, "mean_gate": {"surface": 0.14585}, "n_reads": 548, "n_cand": 6469, "read_suppressed_frac": 0.1022, "dropout_reads_logged": 548, "inventory": {"size": 327079, "admitted": 65536, "staged": 261543, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261543, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.239776611328125, "n_util_obs": 54, "util_count_spearman": 0.6566418906041548}, "probe_overhead_sec_total": 208.356, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.030727, "alpha": 0.03, "t": 2278.1} |
| {"event": "consolidate", "step": 3300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261930, "t": 2330.6} |
| {"event": "train", "step": 3300, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47624, "bpb": 2.12976, "chunk_loss": 0.030344, "F": 0.19754, "G": 0.21389, "chunks": 202.28, "bytes_per_chunk": 5.062, "boundary_entropy": 0.10527, "router_noise": 0.0, "grad_norm": 0.7372, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 1.066, "probe_sec": 0.0364, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02365, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.21033, "n_shadow": 9, "mean_delta_shadow": -0.05443, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.51409, "trunc_abs_frac": 0.86065, "mean_delta_full": 0.07175, "read_hit_rate": {"surface": 0.07965}, "mean_gate": {"surface": 0.15381}, "n_reads": 513, "n_cand": 6441, "read_suppressed_frac": 0.1092, "dropout_reads_logged": 513, "inventory": {"size": 327139, "admitted": 65536, "staged": 261603, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261603, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.464263916015625, "n_util_obs": 50, "util_count_spearman": 0.6235294117647059}, "probe_overhead_sec_total": 211.005, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.037249, "alpha": 0.03, "t": 2331.4} |
| {"event": "consolidate", "step": 3350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261923, "t": 2375.0} |
| {"event": "train", "step": 3350, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.4256, "bpb": 2.05671, "chunk_loss": 0.030297, "F": 0.19522, "G": 0.21207, "chunks": 199.91, "bytes_per_chunk": 5.122, "boundary_entropy": 0.10564, "router_noise": 0.0, "grad_norm": 0.7462, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.88, "probe_sec": 0.0333, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02181, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.05841, "n_shadow": 5, "mean_delta_shadow": 0.21537, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.50067, "trunc_abs_frac": 0.47843, "mean_delta_full": 0.66935, "read_hit_rate": {"surface": 0.08201}, "mean_gate": {"surface": 0.14498}, "n_reads": 522, "n_cand": 6365, "read_suppressed_frac": 0.1054, "dropout_reads_logged": 522, "inventory": {"size": 327177, "admitted": 65536, "staged": 261641, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261641, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.026885986328125, "n_util_obs": 44, "util_count_spearman": 0.6482029598308668}, "probe_overhead_sec_total": 213.691, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.035385, "alpha": 0.03, "t": 2375.4} |
| {"event": "consolidate", "step": 3400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261921, "t": 2429.1} |
| {"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3364.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.1933, "cps": 26.28125, "mean_p": 0.200121, "boundary_entropy": 0.083348, "bpb": 1.960625}, "tr": {"chunks": 3249.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179473, "cps": 25.382812, "mean_p": 0.22408, "boundary_entropy": 0.218607, "bpb": 2.146352}, "fi": {"chunks": 3337.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177396, "cps": 26.070312, "mean_p": 0.206275, "boundary_entropy": 0.170321, "bpb": 2.308289}}, "bpb": {"en": 1.96062, "tr": 2.14635, "fi": 2.30829}, "gini_cpc": 0.019271, "gini_cps": 0.007705, "bpic": 5.701709, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9284687542016639, "fi": 0.9177247753499657}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.312667846679688, "n_util_obs": 43, "util_count_spearman": 0.6424041075203866}, "t": 2430.1} |
| {"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 1.97929, "tr": 2.15438, "fi": 2.31625}, "bytes_per_chunk": {"en": 4.901, "tr": 5.936, "fi": 5.676}, "bpb_mean": 2.14997, "frozen_inventory": true, "t": 2431.3} |
| {"event": "train", "step": 3400, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.38402, "bpb": 1.99672, "chunk_loss": 0.030343, "F": 0.19882, "G": 0.21245, "chunks": 203.59, "bytes_per_chunk": 5.03, "boundary_entropy": 0.09753, "router_noise": 0.0, "grad_norm": 0.7029, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 1.207, "probe_sec": 0.0481, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02403, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.05804, "n_shadow": 6, "mean_delta_shadow": -0.23286, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.24578, "trunc_abs_frac": 0.24578, "mean_delta_full": 0.38367, "read_hit_rate": {"surface": 0.09008}, "mean_gate": {"surface": 0.1493}, "n_reads": 584, "n_cand": 6483, "read_suppressed_frac": 0.1045, "dropout_reads_logged": 584, "inventory": {"size": 327085, "admitted": 65536, "staged": 261549, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261549, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.321578979492188, "n_util_obs": 43, "util_count_spearman": 0.6466324373301118}, "probe_overhead_sec_total": 217.101, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.037393, "alpha": 0.03, "t": 2435.8} |
| {"event": "consolidate", "step": 3450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261914, "t": 2485.2} |
| {"event": "train", "step": 3450, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.54267, "bpb": 2.22561, "chunk_loss": 0.030396, "F": 0.19781, "G": 0.21645, "chunks": 202.56, "bytes_per_chunk": 5.055, "boundary_entropy": 0.10907, "router_noise": 0.0, "grad_norm": 0.768, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 1.01, "probe_sec": 0.074, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02813, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.41956, "n_shadow": 3, "mean_delta_shadow": 0.01477, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.62675, "trunc_abs_frac": 0.61335, "mean_delta_full": -0.38411, "read_hit_rate": {"surface": 0.07225}, "mean_gate": {"surface": 0.14736}, "n_reads": 466, "n_cand": 6450, "read_suppressed_frac": 0.1094, "dropout_reads_logged": 466, "inventory": {"size": 327238, "admitted": 65536, "staged": 261702, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261702, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 28.009628295898438, "n_util_obs": 47, "util_count_spearman": 0.6931082331174838}, "probe_overhead_sec_total": 220.255, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.034828, "alpha": 0.03, "t": 2486.3} |
| {"event": "consolidate", "step": 3500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261914, "t": 2534.2} |
| {"event": "train", "step": 3500, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41739, "bpb": 2.04487, "chunk_loss": 0.030336, "F": 0.19708, "G": 0.21458, "chunks": 201.81, "bytes_per_chunk": 5.074, "boundary_entropy": 0.10811, "router_noise": 0.0, "grad_norm": 0.7507, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.973, "probe_sec": 0.0836, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06671, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.46764, "n_shadow": 5, "mean_delta_shadow": -0.12593, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.10867, "trunc_abs_frac": 0.69547, "mean_delta_full": 0.41096, "read_hit_rate": {"surface": 0.08606}, "mean_gate": {"surface": 0.15291}, "n_reads": 553, "n_cand": 6426, "read_suppressed_frac": 0.0976, "dropout_reads_logged": 553, "inventory": {"size": 327104, "admitted": 65536, "staged": 261568, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261568, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 28.767974853515625, "n_util_obs": 50, "util_count_spearman": 0.7176470588235294}, "probe_overhead_sec_total": 223.429, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.039705, "alpha": 0.03, "t": 2535.0} |
| {"event": "consolidate", "step": 3550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261913, "t": 2585.5} |
| {"event": "train", "step": 3550, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.40385, "bpb": 2.02532, "chunk_loss": 0.030279, "F": 0.19489, "G": 0.20893, "chunks": 199.56, "bytes_per_chunk": 5.131, "boundary_entropy": 0.09924, "router_noise": 0.0, "grad_norm": 0.715, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 1.028, "probe_sec": 0.037, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02087, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.04239, "n_shadow": 6, "mean_delta_shadow": -0.05611, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.28173, "trunc_abs_frac": 1.30605, "mean_delta_full": 0.44824, "read_hit_rate": {"surface": 0.09223}, "mean_gate": {"surface": 0.14257}, "n_reads": 586, "n_cand": 6354, "read_suppressed_frac": 0.1092, "dropout_reads_logged": 586, "inventory": {"size": 327177, "admitted": 65536, "staged": 261641, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261641, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 28.971145629882812, "n_util_obs": 48, "util_count_spearman": 0.7004993486756405}, "probe_overhead_sec_total": 226.773, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.029746, "alpha": 0.03, "t": 2586.4} |
| {"event": "consolidate", "step": 3600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261908, "t": 2632.0} |
| {"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3365.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193357, "cps": 26.289062, "mean_p": 0.199838, "boundary_entropy": 0.080711, "bpb": 1.945925}, "tr": {"chunks": 3244.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179197, "cps": 25.34375, "mean_p": 0.223252, "boundary_entropy": 0.213853, "bpb": 2.138277}, "fi": {"chunks": 3256.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17309, "cps": 25.4375, "mean_p": 0.200463, "boundary_entropy": 0.160242, "bpb": 2.288717}}, "bpb": {"en": 1.94593, "tr": 2.13828, "fi": 2.28872}, "gini_cpc": 0.024762, "gini_cps": 0.008177, "bpic": 5.750836, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9267644063165382, "fi": 0.8951824605801528}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 29.363616943359375, "n_util_obs": 48, "util_count_spearman": 0.708315240990013}, "t": 2632.6} |
| {"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.96268, "tr": 2.14458, "fi": 2.29834}, "bytes_per_chunk": {"en": 4.898, "tr": 5.937, "fi": 5.739}, "bpb_mean": 2.1352, "frozen_inventory": true, "t": 2633.6} |
| {"event": "train", "step": 3600, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.43055, "bpb": 2.06384, "chunk_loss": 0.030289, "F": 0.19397, "G": 0.21637, "chunks": 198.62, "bytes_per_chunk": 5.155, "boundary_entropy": 0.11835, "router_noise": 0.0, "grad_norm": 0.7225, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 1.045, "probe_sec": 0.0388, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02626, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.34888, "n_shadow": 8, "mean_delta_shadow": 0.0094, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 3.75587, "trunc_abs_frac": 1.01204, "mean_delta_full": 0.0455, "read_hit_rate": {"surface": 0.07638}, "mean_gate": {"surface": 0.1406}, "n_reads": 483, "n_cand": 6324, "read_suppressed_frac": 0.1077, "dropout_reads_logged": 483, "inventory": {"size": 327224, "admitted": 65536, "staged": 261688, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261688, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 29.370986938476562, "n_util_obs": 48, "util_count_spearman": 0.708315240990013}, "probe_overhead_sec_total": 230.104, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.035147, "alpha": 0.03, "t": 2638.7} |
| {"event": "consolidate", "step": 3650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261903, "t": 2685.6} |
| {"event": "train", "step": 3650, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.35702, "bpb": 1.95777, "chunk_loss": 0.030335, "F": 0.19931, "G": 0.21082, "chunks": 204.09, "bytes_per_chunk": 5.017, "boundary_entropy": 0.09004, "router_noise": 0.0, "grad_norm": 0.6631, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.959, "probe_sec": 0.061, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02719, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.24557, "n_shadow": 3, "mean_delta_shadow": -0.23218, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.97693, "trunc_abs_frac": 1.08865, "mean_delta_full": 0.32143, "read_hit_rate": {"surface": 0.09371}, "mean_gate": {"surface": 0.14649}, "n_reads": 609, "n_cand": 6499, "read_suppressed_frac": 0.1002, "dropout_reads_logged": 609, "inventory": {"size": 327012, "admitted": 65536, "staged": 261476, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261476, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 29.699996948242188, "n_util_obs": 48, "util_count_spearman": 0.7087494572297004}, "probe_overhead_sec_total": 233.19, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.035589, "alpha": 0.03, "t": 2686.6} |
| {"event": "consolidate", "step": 3700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261902, "t": 2743.5} |
| {"event": "train", "step": 3700, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46299, "bpb": 2.11065, "chunk_loss": 0.030251, "F": 0.18985, "G": 0.21183, "chunks": 194.41, "bytes_per_chunk": 5.267, "boundary_entropy": 0.11529, "router_noise": 0.0, "grad_norm": 0.7843, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 1.158, "probe_sec": 0.0564, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03118, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.28654, "n_shadow": 6, "mean_delta_shadow": -0.21714, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.62598, "trunc_abs_frac": 0.45477, "mean_delta_full": 0.078, "read_hit_rate": {"surface": 0.08806}, "mean_gate": {"surface": 0.138}, "n_reads": 545, "n_cand": 6189, "read_suppressed_frac": 0.1009, "dropout_reads_logged": 545, "inventory": {"size": 327287, "admitted": 65536, "staged": 261751, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261751, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 30.550918579101562, "n_util_obs": 51, "util_count_spearman": 0.6607239819004525}, "probe_overhead_sec_total": 236.689, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.03444, "alpha": 0.03, "t": 2744.5} |
| {"event": "consolidate", "step": 3750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261898, "t": 2803.0} |
| {"event": "train", "step": 3750, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44584, "bpb": 2.08591, "chunk_loss": 0.030408, "F": 0.19794, "G": 0.21706, "chunks": 202.69, "bytes_per_chunk": 5.052, "boundary_entropy": 0.10745, "router_noise": 0.0, "grad_norm": 0.7405, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 1.191, "probe_sec": 0.1507, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.10064, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.20366, "n_shadow": 4, "mean_delta_shadow": -0.47962, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.08692}, "mean_gate": {"surface": 0.14481}, "n_reads": 561, "n_cand": 6454, "read_suppressed_frac": 0.0963, "dropout_reads_logged": 561, "inventory": {"size": 327248, "admitted": 65536, "staged": 261712, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261712, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 31.333740234375, "n_util_obs": 53, "util_count_spearman": 0.685292694726657}, "probe_overhead_sec_total": 240.519, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.033383, "alpha": 0.03, "t": 2804.1} |
| {"event": "consolidate", "step": 3800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261896, "t": 2862.5} |
| {"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3356.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19284, "cps": 26.21875, "mean_p": 0.19905, "boundary_entropy": 0.078807, "bpb": 1.931219}, "tr": {"chunks": 3334.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184168, "cps": 26.046875, "mean_p": 0.22178, "boundary_entropy": 0.208786, "bpb": 2.124152}, "fi": {"chunks": 3346.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177875, "cps": 26.140625, "mean_p": 0.203023, "boundary_entropy": 0.156537, "bpb": 2.270979}}, "bpb": {"en": 1.93122, "tr": 2.12415, "fi": 2.27098}, "gini_cpc": 0.017981, "gini_cps": 0.001461, "bpic": 5.65285, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9550304353640013, "fi": 0.922393473146624}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 32.789581298828125, "n_util_obs": 58, "util_count_spearman": 0.6867944261589098}, "t": 2863.7} |
| {"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.94923, "tr": 2.13161, "fi": 2.28111}, "bytes_per_chunk": {"en": 4.899, "tr": 5.831, "fi": 5.667}, "bpb_mean": 2.12065, "frozen_inventory": true, "t": 2864.7} |
| {"event": "train", "step": 3800, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.34151, "bpb": 1.93539, "chunk_loss": 0.030352, "F": 0.19992, "G": 0.21057, "chunks": 204.72, "bytes_per_chunk": 5.002, "boundary_entropy": 0.08783, "router_noise": 0.0, "grad_norm": 0.7737, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 1.379, "probe_sec": 0.0844, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04764, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.55082, "n_shadow": 7, "mean_delta_shadow": 0.09169, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.79825, "trunc_abs_frac": 1.34397, "mean_delta_full": -0.20898, "read_hit_rate": {"surface": 0.11566}, "mean_gate": {"surface": 0.1407}, "n_reads": 754, "n_cand": 6519, "read_suppressed_frac": 0.0849, "dropout_reads_logged": 754, "inventory": {"size": 327032, "admitted": 65536, "staged": 261496, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261496, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 32.80108642578125, "n_util_obs": 58, "util_count_spearman": 0.6863637761850564}, "probe_overhead_sec_total": 244.39, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.032152, "alpha": 0.03, "t": 2873.0} |
| {"event": "consolidate", "step": 3850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261894, "t": 2937.9} |
| {"event": "train", "step": 3850, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.35777, "bpb": 1.95885, "chunk_loss": 0.030355, "F": 0.19983, "G": 0.21183, "chunks": 204.62, "bytes_per_chunk": 5.004, "boundary_entropy": 0.0915, "router_noise": 0.0, "grad_norm": 0.7052, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 1.322, "probe_sec": 0.0594, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02928, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.41877, "n_shadow": 3, "mean_delta_shadow": -0.28579, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.08901, "trunc_abs_frac": 2.58889, "mean_delta_full": 0.91107, "read_hit_rate": {"surface": 0.10789}, "mean_gate": {"surface": 0.14092}, "n_reads": 703, "n_cand": 6516, "read_suppressed_frac": 0.0996, "dropout_reads_logged": 703, "inventory": {"size": 327043, "admitted": 65536, "staged": 261507, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261507, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 34.13578796386719, "n_util_obs": 61, "util_count_spearman": 0.6754627181385511}, "probe_overhead_sec_total": 248.514, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.034536, "alpha": 0.03, "t": 2939.1} |
| {"event": "consolidate", "step": 3900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261893, "t": 2996.1} |
| {"event": "train", "step": 3900, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42747, "bpb": 2.0594, "chunk_loss": 0.030297, "F": 0.19565, "G": 0.21396, "chunks": 200.34, "bytes_per_chunk": 5.111, "boundary_entropy": 0.10445, "router_noise": 0.0, "grad_norm": 0.6985, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 1.159, "probe_sec": 0.0899, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04648, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.41808, "n_shadow": 8, "mean_delta_shadow": -0.07302, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.32847, "trunc_abs_frac": 0.72849, "mean_delta_full": 0.59601, "read_hit_rate": {"surface": 0.10613}, "mean_gate": {"surface": 0.13505}, "n_reads": 677, "n_cand": 6379, "read_suppressed_frac": 0.0827, "dropout_reads_logged": 677, "inventory": {"size": 327159, "admitted": 65536, "staged": 261623, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261623, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 36.5899658203125, "n_util_obs": 65, "util_count_spearman": 0.5928321678321679}, "probe_overhead_sec_total": 252.329, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.031961, "alpha": 0.03, "t": 2997.1} |
| {"event": "consolidate", "step": 3950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261893, "t": 3052.0} |
| {"event": "train", "step": 3950, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42253, "bpb": 2.05228, "chunk_loss": 0.030334, "F": 0.19861, "G": 0.20919, "chunks": 203.38, "bytes_per_chunk": 5.035, "boundary_entropy": 0.08736, "router_noise": 0.0, "grad_norm": 0.7135, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 1.131, "probe_sec": 0.0974, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05097, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.07751, "n_shadow": 8, "mean_delta_shadow": 0.16338, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.0, "trunc_abs_frac": 1.0, "mean_delta_full": -0.56384, "read_hit_rate": {"surface": 0.11844}, "mean_gate": {"surface": 0.13115}, "n_reads": 767, "n_cand": 6476, "read_suppressed_frac": 0.0926, "dropout_reads_logged": 767, "inventory": {"size": 327165, "admitted": 65536, "staged": 261629, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261629, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 37.916229248046875, "n_util_obs": 71, "util_count_spearman": 0.573608316566063}, "probe_overhead_sec_total": 255.969, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.033204, "alpha": 0.03, "t": 3053.6} |
| {"event": "consolidate", "step": 4000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261893, "t": 3109.2} |
| {"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3380.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194219, "cps": 26.40625, "mean_p": 0.199539, "boundary_entropy": 0.078025, "bpb": 1.921017}, "tr": {"chunks": 3217.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177705, "cps": 25.132812, "mean_p": 0.220586, "boundary_entropy": 0.208327, "bpb": 2.106995}, "fi": {"chunks": 3373.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17931, "cps": 26.351562, "mean_p": 0.204679, "boundary_entropy": 0.158426, "bpb": 2.265214}}, "bpb": {"en": 1.92102, "tr": 2.107, "fi": 2.26521}, "gini_cpc": 0.019972, "gini_cps": 0.010899, "bpic": 5.690271, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9149722642329053, "fi": 0.9232341865942091}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 38.60028076171875, "n_util_obs": 74, "util_count_spearman": 0.5605479452054795}, "t": 3109.7} |
| {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.93936, "tr": 2.1134, "fi": 2.27336}, "bytes_per_chunk": {"en": 4.885, "tr": 5.919, "fi": 5.628}, "bpb_mean": 2.10871, "frozen_inventory": true, "t": 3110.7} |
| {"event": "train", "step": 4000, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.36185, "bpb": 1.96474, "chunk_loss": 0.030274, "F": 0.19305, "G": 0.21179, "chunks": 197.69, "bytes_per_chunk": 5.18, "boundary_entropy": 0.11084, "router_noise": 0.0, "grad_norm": 0.707, "lr": 0.000104, "router_trainable": true, "sec_per_step": 1.29, "probe_sec": 0.0911, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04722, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.01762, "n_shadow": 7, "mean_delta_shadow": -0.10142, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.78809, "trunc_abs_frac": 0.92494, "mean_delta_full": 0.2608, "read_hit_rate": {"surface": 0.11455}, "mean_gate": {"surface": 0.13298}, "n_reads": 721, "n_cand": 6294, "read_suppressed_frac": 0.1096, "dropout_reads_logged": 721, "inventory": {"size": 327118, "admitted": 65536, "staged": 261582, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261582, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 38.61128234863281, "n_util_obs": 74, "util_count_spearman": 0.5605479452054795}, "probe_overhead_sec_total": 259.787, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.034445, "alpha": 0.03, "t": 3118.1} |
| {"event": "consolidate", "step": 4050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261892, "t": 3173.3} |
| {"event": "train", "step": 4050, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.49623, "bpb": 2.15861, "chunk_loss": 0.030315, "F": 0.19266, "G": 0.21173, "chunks": 197.28, "bytes_per_chunk": 5.191, "boundary_entropy": 0.10833, "router_noise": 0.0, "grad_norm": 0.7841, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 1.124, "probe_sec": 0.056, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02157, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.33198, "n_shadow": 8, "mean_delta_shadow": -0.2176, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.27876, "trunc_abs_frac": 0.43629, "mean_delta_full": -0.14111, "read_hit_rate": {"surface": 0.0968}, "mean_gate": {"surface": 0.13088}, "n_reads": 608, "n_cand": 6281, "read_suppressed_frac": 0.1036, "dropout_reads_logged": 608, "inventory": {"size": 327288, "admitted": 65536, "staged": 261752, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261752, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 39.31724548339844, "n_util_obs": 76, "util_count_spearman": 0.5602734107997266}, "probe_overhead_sec_total": 263.625, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.034748, "alpha": 0.03, "t": 3174.3} |
| {"event": "consolidate", "step": 4100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261885, "t": 3231.4} |
| {"event": "train", "step": 4100, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.43364, "bpb": 2.06831, "chunk_loss": 0.030325, "F": 0.19614, "G": 0.21637, "chunks": 200.84, "bytes_per_chunk": 5.098, "boundary_entropy": 0.11089, "router_noise": 0.0, "grad_norm": 0.715, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 1.181, "probe_sec": 0.121, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0699, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.02012, "n_shadow": 8, "mean_delta_shadow": 0.13799, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.86807, "trunc_abs_frac": 0.48434, "mean_delta_full": -0.13387, "read_hit_rate": {"surface": 0.11212}, "mean_gate": {"surface": 0.13105}, "n_reads": 717, "n_cand": 6395, "read_suppressed_frac": 0.0962, "dropout_reads_logged": 717, "inventory": {"size": 327224, "admitted": 65536, "staged": 261688, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261688, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 40.16184997558594, "n_util_obs": 81, "util_count_spearman": 0.5467479674796748}, "probe_overhead_sec_total": 267.547, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.034195, "alpha": 0.03, "t": 3233.3} |
| {"event": "consolidate", "step": 4150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261888, "t": 3291.1} |
| {"event": "train", "step": 4150, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.37266, "bpb": 1.98033, "chunk_loss": 0.030392, "F": 0.2002, "G": 0.21325, "chunks": 205.0, "bytes_per_chunk": 4.995, "boundary_entropy": 0.089, "router_noise": 0.0, "grad_norm": 0.6726, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 1.175, "probe_sec": 0.105, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05957, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.08292, "n_shadow": 6, "mean_delta_shadow": -0.54041, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.96362, "trunc_abs_frac": 0.80441, "mean_delta_full": -1.16952, "read_hit_rate": {"surface": 0.11964}, "mean_gate": {"surface": 0.13231}, "n_reads": 781, "n_cand": 6528, "read_suppressed_frac": 0.0999, "dropout_reads_logged": 781, "inventory": {"size": 327065, "admitted": 65536, "staged": 261529, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261529, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 40.72483825683594, "n_util_obs": 81, "util_count_spearman": 0.5294941282746161}, "probe_overhead_sec_total": 271.646, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.033396, "alpha": 0.03, "t": 3292.1} |
| {"event": "consolidate", "step": 4200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261884, "t": 3347.5} |
| {"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3347.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192323, "cps": 26.148438, "mean_p": 0.197647, "boundary_entropy": 0.074957, "bpb": 1.910838}, "tr": {"chunks": 3224.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178092, "cps": 25.1875, "mean_p": 0.221565, "boundary_entropy": 0.210099, "bpb": 2.092917}, "fi": {"chunks": 3276.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.174153, "cps": 25.59375, "mean_p": 0.197512, "boundary_entropy": 0.155589, "bpb": 2.243339}}, "bpb": {"en": 1.91084, "tr": 2.09292, "fi": 2.24334}, "gini_cpc": 0.022244, "gini_cps": 0.008327, "bpic": 5.761349, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9260040572865745, "fi": 0.9055249427588765}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 41.55134582519531, "n_util_obs": 84, "util_count_spearman": 0.5427558975397387}, "t": 3348.1} |
| {"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.92868, "tr": 2.10012, "fi": 2.25065}, "bytes_per_chunk": {"en": 4.914, "tr": 5.891, "fi": 5.724}, "bpb_mean": 2.09315, "frozen_inventory": true, "t": 3349.4} |
| {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 3355.1} |
| {"event": "train", "step": 4200, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.45092, "bpb": 2.09324, "chunk_loss": 0.030377, "F": 0.19989, "G": 0.21436, "chunks": 204.69, "bytes_per_chunk": 5.003, "boundary_entropy": 0.09604, "router_noise": 0.0, "grad_norm": 0.7396, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 1.288, "probe_sec": 0.0967, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05949, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.10953, "n_shadow": 4, "mean_delta_shadow": -0.09821, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.36799, "trunc_abs_frac": 126.4418, "mean_delta_full": -0.15887, "read_hit_rate": {"surface": 0.11}, "mean_gate": {"surface": 0.13227}, "n_reads": 717, "n_cand": 6518, "read_suppressed_frac": 0.1046, "dropout_reads_logged": 717, "inventory": {"size": 327185, "admitted": 65536, "staged": 261649, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261649, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 41.562286376953125, "n_util_obs": 84, "util_count_spearman": 0.5427963956667005}, "probe_overhead_sec_total": 275.159, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.035273, "alpha": 0.03, "t": 3356.5} |
| {"event": "consolidate", "step": 4250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261881, "t": 3411.1} |
| {"event": "train", "step": 4250, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.34572, "bpb": 1.94147, "chunk_loss": 0.030268, "F": 0.19507, "G": 0.20653, "chunks": 199.75, "bytes_per_chunk": 5.126, "boundary_entropy": 0.08827, "router_noise": 0.0, "grad_norm": 0.6951, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 1.104, "probe_sec": 0.0473, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.024, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.34016, "n_shadow": 6, "mean_delta_shadow": -0.06844, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.83046, "trunc_abs_frac": 0.76337, "mean_delta_full": 0.60486, "read_hit_rate": {"surface": 0.12642}, "mean_gate": {"surface": 0.13118}, "n_reads": 804, "n_cand": 6360, "read_suppressed_frac": 0.1045, "dropout_reads_logged": 804, "inventory": {"size": 327108, "admitted": 65536, "staged": 261572, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261572, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 42.1080322265625, "n_util_obs": 85, "util_count_spearman": 0.5511041625952706}, "probe_overhead_sec_total": 278.646, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.030625, "alpha": 0.03, "t": 3411.7} |
| {"event": "consolidate", "step": 4300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261877, "t": 3457.1} |
| {"event": "train", "step": 4300, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35147, "bpb": 1.94977, "chunk_loss": 0.030233, "F": 0.19028, "G": 0.21303, "chunks": 194.84, "bytes_per_chunk": 5.255, "boundary_entropy": 0.11644, "router_noise": 0.0, "grad_norm": 0.7345, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.929, "probe_sec": 0.0521, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03008, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.29594, "n_shadow": 4, "mean_delta_shadow": 0.0933, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -2.84995, "trunc_abs_frac": 1.64408, "mean_delta_full": 0.03117, "read_hit_rate": {"surface": 0.10495}, "mean_gate": {"surface": 0.12712}, "n_reads": 651, "n_cand": 6203, "read_suppressed_frac": 0.1075, "dropout_reads_logged": 651, "inventory": {"size": 327148, "admitted": 65536, "staged": 261612, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261612, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 42.32548522949219, "n_util_obs": 82, "util_count_spearman": 0.5297613217096027}, "probe_overhead_sec_total": 281.99, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.034203, "alpha": 0.03, "t": 3458.1} |
| {"event": "consolidate", "step": 4350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261874, "t": 3509.4} |
| {"event": "train", "step": 4350, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33182, "bpb": 1.92141, "chunk_loss": 0.030271, "F": 0.19394, "G": 0.21093, "chunks": 198.59, "bytes_per_chunk": 5.156, "boundary_entropy": 0.10081, "router_noise": 0.0, "grad_norm": 0.738, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 1.044, "probe_sec": 0.0329, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02054, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.08692, "n_shadow": 8, "mean_delta_shadow": -0.0872, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.24352, "trunc_abs_frac": 0.56144, "mean_delta_full": -0.21533, "read_hit_rate": {"surface": 0.11893}, "mean_gate": {"surface": 0.12958}, "n_reads": 752, "n_cand": 6323, "read_suppressed_frac": 0.1037, "dropout_reads_logged": 752, "inventory": {"size": 327111, "admitted": 65536, "staged": 261575, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261575, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 43.060211181640625, "n_util_obs": 83, "util_count_spearman": 0.5097812854204273}, "probe_overhead_sec_total": 285.722, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026224, "alpha": 0.03, "t": 3510.3} |
| {"event": "consolidate", "step": 4400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261871, "t": 3563.9} |
| {"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3374.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193875, "cps": 26.359375, "mean_p": 0.19871, "boundary_entropy": 0.071838, "bpb": 1.903194}, "tr": {"chunks": 3230.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178423, "cps": 25.234375, "mean_p": 0.221388, "boundary_entropy": 0.207444, "bpb": 2.073902}, "fi": {"chunks": 3311.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176014, "cps": 25.867188, "mean_p": 0.199042, "boundary_entropy": 0.146821, "bpb": 2.224174}}, "bpb": {"en": 1.90319, "tr": 2.0739, "fi": 2.22417}, "gini_cpc": 0.021716, "gini_cps": 0.009682, "bpic": 5.721836, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9203033710709131, "fi": 0.9078755896997674}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 43.656097412109375, "n_util_obs": 84, "util_count_spearman": 0.5348587627822213}, "t": 3564.9} |
| {"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.92086, "tr": 2.08128, "fi": 2.2341}, "bytes_per_chunk": {"en": 4.89, "tr": 5.85, "fi": 5.684}, "bpb_mean": 2.07875, "frozen_inventory": true, "t": 3566.0} |
| {"event": "train", "step": 4400, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36991, "bpb": 1.97636, "chunk_loss": 0.030401, "F": 0.20355, "G": 0.21178, "chunks": 208.44, "bytes_per_chunk": 4.913, "boundary_entropy": 0.07942, "router_noise": 0.0, "grad_norm": 0.7925, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 1.257, "probe_sec": 0.0497, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02262, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.15664, "n_shadow": 5, "mean_delta_shadow": -0.23348, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.8475, "trunc_abs_frac": 1.27673, "mean_delta_full": -0.24894, "read_hit_rate": {"surface": 0.12142}, "mean_gate": {"surface": 0.12708}, "n_reads": 806, "n_cand": 6638, "read_suppressed_frac": 0.0943, "dropout_reads_logged": 806, "inventory": {"size": 327075, "admitted": 65536, "staged": 261539, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261539, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 43.66839599609375, "n_util_obs": 84, "util_count_spearman": 0.5322466335931963}, "probe_overhead_sec_total": 289.322, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.029249, "alpha": 0.03, "t": 3573.2} |
| {"event": "consolidate", "step": 4450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261872, "t": 3621.0} |
| {"event": "train", "step": 4450, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39129, "bpb": 2.0072, "chunk_loss": 0.030237, "F": 0.19223, "G": 0.20974, "chunks": 196.84, "bytes_per_chunk": 5.202, "boundary_entropy": 0.09827, "router_noise": 0.0, "grad_norm": 0.696, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.979, "probe_sec": 0.1008, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04313, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.00427, "n_shadow": 9, "mean_delta_shadow": -0.21368, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.5851, "trunc_abs_frac": 1.52341, "mean_delta_full": -0.394, "read_hit_rate": {"surface": 0.1101}, "mean_gate": {"surface": 0.1289}, "n_reads": 690, "n_cand": 6267, "read_suppressed_frac": 0.1174, "dropout_reads_logged": 690, "inventory": {"size": 327162, "admitted": 65536, "staged": 261626, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261626, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 44.228240966796875, "n_util_obs": 86, "util_count_spearman": 0.5169419168033212}, "probe_overhead_sec_total": 292.914, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.031949, "alpha": 0.03, "t": 3622.2} |
| {"event": "consolidate", "step": 4500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261865, "t": 3676.6} |
| {"event": "train", "step": 4500, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.4083, "bpb": 2.03175, "chunk_loss": 0.030276, "F": 0.19403, "G": 0.20794, "chunks": 198.69, "bytes_per_chunk": 5.154, "boundary_entropy": 0.09306, "router_noise": 0.0, "grad_norm": 0.7092, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 1.1, "probe_sec": 0.0469, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02476, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.30487, "n_shadow": 7, "mean_delta_shadow": -0.20434, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 2.3456, "trunc_abs_frac": 0.89579, "mean_delta_full": 0.24316, "read_hit_rate": {"surface": 0.12061}, "mean_gate": {"surface": 0.12609}, "n_reads": 763, "n_cand": 6326, "read_suppressed_frac": 0.0996, "dropout_reads_logged": 763, "inventory": {"size": 327203, "admitted": 65536, "staged": 261667, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261667, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 44.80976867675781, "n_util_obs": 87, "util_count_spearman": 0.49577167019027485}, "probe_overhead_sec_total": 296.712, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.031488, "alpha": 0.03, "t": 3677.1} |
| {"event": "consolidate", "step": 4550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261860, "t": 3731.5} |
| {"event": "train", "step": 4550, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32904, "bpb": 1.9174, "chunk_loss": 0.030245, "F": 0.19025, "G": 0.2084, "chunks": 194.81, "bytes_per_chunk": 5.256, "boundary_entropy": 0.10418, "router_noise": 0.0, "grad_norm": 0.6751, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 1.109, "probe_sec": 0.1006, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05475, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.03508, "n_shadow": 6, "mean_delta_shadow": -0.26031, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.56583, "trunc_abs_frac": 0.89966, "mean_delta_full": -0.22397, "read_hit_rate": {"surface": 0.11964}, "mean_gate": {"surface": 0.13112}, "n_reads": 742, "n_cand": 6202, "read_suppressed_frac": 0.0916, "dropout_reads_logged": 742, "inventory": {"size": 327154, "admitted": 65536, "staged": 261618, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261618, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 45.48268127441406, "n_util_obs": 89, "util_count_spearman": 0.49933605720122576}, "probe_overhead_sec_total": 300.524, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.027905, "alpha": 0.03, "t": 3732.6} |
| {"event": "consolidate", "step": 4600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261859, "t": 3785.2} |
| {"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3360.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19307, "cps": 26.25, "mean_p": 0.197375, "boundary_entropy": 0.071507, "bpb": 1.890848}, "tr": {"chunks": 3234.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178644, "cps": 25.265625, "mean_p": 0.218507, "boundary_entropy": 0.201748, "bpb": 2.065868}, "fi": {"chunks": 3341.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177609, "cps": 26.101562, "mean_p": 0.200935, "boundary_entropy": 0.148576, "bpb": 2.214003}}, "bpb": {"en": 1.89085, "tr": 2.06587, "fi": 2.214}, "gini_cpc": 0.018764, "gini_cps": 0.008455, "bpic": 5.710317, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9252824117549577, "fi": 0.9199186741040577}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 46.03810119628906, "n_util_obs": 89, "util_count_spearman": 0.4926796050391556}, "t": 3785.7} |
| {"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.90843, "tr": 2.0736, "fi": 2.22264}, "bytes_per_chunk": {"en": 4.904, "tr": 5.857, "fi": 5.646}, "bpb_mean": 2.06823, "frozen_inventory": true, "t": 3786.8} |
| {"event": "train", "step": 4600, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39219, "bpb": 2.0085, "chunk_loss": 0.03026, "F": 0.19324, "G": 0.20597, "chunks": 197.88, "bytes_per_chunk": 5.175, "boundary_entropy": 0.08987, "router_noise": 0.0, "grad_norm": 0.7346, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 1.234, "probe_sec": 0.0854, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06996, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.07732, "n_shadow": 5, "mean_delta_shadow": -0.24699, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.41175, "trunc_abs_frac": 0.61585, "mean_delta_full": -0.28265, "read_hit_rate": {"surface": 0.11937}, "mean_gate": {"surface": 0.12764}, "n_reads": 752, "n_cand": 6300, "read_suppressed_frac": 0.0957, "dropout_reads_logged": 752, "inventory": {"size": 327178, "admitted": 65536, "staged": 261642, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261642, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 46.04957580566406, "n_util_obs": 89, "util_count_spearman": 0.4926796050391556}, "probe_overhead_sec_total": 304.124, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026699, "alpha": 0.03, "t": 3794.3} |
| {"event": "consolidate", "step": 4650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261857, "t": 3850.9} |
| {"event": "train", "step": 4650, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39355, "bpb": 2.01047, "chunk_loss": 0.03034, "F": 0.19919, "G": 0.213, "chunks": 203.97, "bytes_per_chunk": 5.02, "boundary_entropy": 0.09324, "router_noise": 0.0, "grad_norm": 0.7336, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 1.156, "probe_sec": 0.054, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02467, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.02753, "n_shadow": 8, "mean_delta_shadow": -0.1198, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.00454, "trunc_abs_frac": 1.11536, "mean_delta_full": -0.26218, "read_hit_rate": {"surface": 0.12902}, "mean_gate": {"surface": 0.13017}, "n_reads": 838, "n_cand": 6495, "read_suppressed_frac": 0.1014, "dropout_reads_logged": 838, "inventory": {"size": 327159, "admitted": 65536, "staged": 261623, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261623, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 46.823150634765625, "n_util_obs": 94, "util_count_spearman": 0.477383231586172}, "probe_overhead_sec_total": 307.903, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026193, "alpha": 0.03, "t": 3852.1} |
| {"event": "consolidate", "step": 4700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261854, "t": 3903.4} |
| {"event": "train", "step": 4700, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37261, "bpb": 1.98026, "chunk_loss": 0.030287, "F": 0.19397, "G": 0.21242, "chunks": 198.62, "bytes_per_chunk": 5.155, "boundary_entropy": 0.10681, "router_noise": 0.0, "grad_norm": 0.7209, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 1.058, "probe_sec": 0.058, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02571, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.10321, "n_shadow": 8, "mean_delta_shadow": -0.30725, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.24167, "trunc_abs_frac": 0.91283, "mean_delta_full": -0.51758, "read_hit_rate": {"surface": 0.11749}, "mean_gate": {"surface": 0.12611}, "n_reads": 743, "n_cand": 6324, "read_suppressed_frac": 0.1009, "dropout_reads_logged": 743, "inventory": {"size": 327173, "admitted": 65536, "staged": 261637, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261637, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 47.49140930175781, "n_util_obs": 95, "util_count_spearman": 0.473488241881299}, "probe_overhead_sec_total": 311.466, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.02917, "alpha": 0.03, "t": 3905.0} |
| {"event": "consolidate", "step": 4750, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 261852, "t": 3954.0} |
| {"event": "train", "step": 4750, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.46275, "bpb": 2.11031, "chunk_loss": 0.030287, "F": 0.19202, "G": 0.20911, "chunks": 196.62, "bytes_per_chunk": 5.208, "boundary_entropy": 0.10188, "router_noise": 0.0, "grad_norm": 0.8753, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 1.019, "probe_sec": 0.1296, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07487, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.06806, "n_shadow": 4, "mean_delta_shadow": -0.18808, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.48885, "trunc_abs_frac": 0.82391, "mean_delta_full": 0.33704, "read_hit_rate": {"surface": 0.10815}, "mean_gate": {"surface": 0.12823}, "n_reads": 677, "n_cand": 6260, "read_suppressed_frac": 0.0975, "dropout_reads_logged": 677, "inventory": {"size": 327265, "admitted": 65536, "staged": 261729, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261729, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 48.052978515625, "n_util_obs": 96, "util_count_spearman": 0.5126288659793814}, "probe_overhead_sec_total": 314.851, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.030064, "alpha": 0.03, "t": 3956.0} |
| {"event": "consolidate", "step": 4800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261846, "t": 4007.8} |
| {"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3352.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19261, "cps": 26.1875, "mean_p": 0.197304, "boundary_entropy": 0.071342, "bpb": 1.881654}, "tr": {"chunks": 3225.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178147, "cps": 25.195312, "mean_p": 0.215742, "boundary_entropy": 0.19667, "bpb": 2.056131}, "fi": {"chunks": 3314.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176174, "cps": 25.890625, "mean_p": 0.19921, "boundary_entropy": 0.146228, "bpb": 2.201317}}, "bpb": {"en": 1.88165, "tr": 2.05613, "fi": 2.20132}, "gini_cpc": 0.020035, "gini_cps": 0.00856, "bpic": 5.735719, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9249095799862811, "fi": 0.914662198741431}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 48.93235778808594, "n_util_obs": 101, "util_count_spearman": 0.46613861386138616}, "t": 4008.6} |
| {"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.90015, "tr": 2.0629, "fi": 2.20952}, "bytes_per_chunk": {"en": 4.908, "tr": 5.91, "fi": 5.681}, "bpb_mean": 2.05752, "frozen_inventory": true, "t": 4010.1} |
| {"event": "train", "step": 4800, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.34661, "bpb": 1.94274, "chunk_loss": 0.030258, "F": 0.19485, "G": 0.20591, "chunks": 199.53, "bytes_per_chunk": 5.132, "boundary_entropy": 0.08263, "router_noise": 0.0, "grad_norm": 0.7202, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 1.205, "probe_sec": 0.1126, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06554, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.11838, "n_shadow": 8, "mean_delta_shadow": -0.01439, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.23333, "trunc_abs_frac": 0.88878, "mean_delta_full": -0.02747, "read_hit_rate": {"surface": 0.1264}, "mean_gate": {"surface": 0.12957}, "n_reads": 803, "n_cand": 6353, "read_suppressed_frac": 0.1183, "dropout_reads_logged": 803, "inventory": {"size": 327034, "admitted": 65536, "staged": 261498, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261498, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 48.944610595703125, "n_util_obs": 101, "util_count_spearman": 0.46613861386138616}, "probe_overhead_sec_total": 318.428, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.028721, "alpha": 0.03, "t": 4016.3} |
| {"event": "consolidate", "step": 4850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261849, "t": 4069.3} |
| {"event": "train", "step": 4850, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33794, "bpb": 1.93024, "chunk_loss": 0.030383, "F": 0.20175, "G": 0.21442, "chunks": 206.59, "bytes_per_chunk": 4.957, "boundary_entropy": 0.08945, "router_noise": 0.0, "grad_norm": 0.7061, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 1.095, "probe_sec": 0.1094, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04434, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.59243, "n_shadow": 10, "mean_delta_shadow": -0.00854, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.10492, "trunc_abs_frac": 0.50642, "mean_delta_full": 0.35382, "read_hit_rate": {"surface": 0.12388}, "mean_gate": {"surface": 0.1262}, "n_reads": 815, "n_cand": 6579, "read_suppressed_frac": 0.0724, "dropout_reads_logged": 815, "inventory": {"size": 327107, "admitted": 65536, "staged": 261571, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261571, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 49.52107238769531, "n_util_obs": 103, "util_count_spearman": 0.46787549971444886}, "probe_overhead_sec_total": 322.296, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.033081, "alpha": 0.03, "t": 4071.0} |
| {"event": "consolidate", "step": 4900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261840, "t": 4127.3} |
| {"event": "train", "step": 4900, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39719, "bpb": 2.01571, "chunk_loss": 0.030276, "F": 0.19473, "G": 0.20963, "chunks": 199.41, "bytes_per_chunk": 5.135, "boundary_entropy": 0.09333, "router_noise": 0.0, "grad_norm": 0.8386, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 1.16, "probe_sec": 0.0932, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02422, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.23378, "n_shadow": 6, "mean_delta_shadow": -0.05836, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.23128, "trunc_abs_frac": 3.88942, "mean_delta_full": -0.54724, "read_hit_rate": {"surface": 0.11671}, "mean_gate": {"surface": 0.12893}, "n_reads": 741, "n_cand": 6349, "read_suppressed_frac": 0.1039, "dropout_reads_logged": 741, "inventory": {"size": 327137, "admitted": 65536, "staged": 261601, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261601, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 50.19422912597656, "n_util_obs": 104, "util_count_spearman": 0.47518403926170916}, "probe_overhead_sec_total": 326.119, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.030017, "alpha": 0.03, "t": 4129.0} |
| {"event": "consolidate", "step": 4950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261834, "t": 4190.9} |
| {"event": "train", "step": 4950, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33059, "bpb": 1.91963, "chunk_loss": 0.030459, "F": 0.1969, "G": 0.21541, "chunks": 201.62, "bytes_per_chunk": 5.079, "boundary_entropy": 0.10381, "router_noise": 0.0, "grad_norm": 0.7102, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 1.275, "probe_sec": 0.0599, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02645, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.07075, "n_shadow": 5, "mean_delta_shadow": -0.09433, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.12042, "trunc_abs_frac": 1.57409, "mean_delta_full": -0.13094, "read_hit_rate": {"surface": 0.12555}, "mean_gate": {"surface": 0.12776}, "n_reads": 806, "n_cand": 6420, "read_suppressed_frac": 0.0993, "dropout_reads_logged": 806, "inventory": {"size": 327151, "admitted": 65536, "staged": 261615, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261615, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 50.79444885253906, "n_util_obs": 104, "util_count_spearman": 0.42689640456630745}, "probe_overhead_sec_total": 329.865, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.028224, "alpha": 0.03, "t": 4192.8} |
| {"event": "consolidate", "step": 5000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261830, "t": 4249.1} |
| {"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3343.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192093, "cps": 26.117188, "mean_p": 0.196563, "boundary_entropy": 0.069399, "bpb": 1.874107}, "tr": {"chunks": 3221.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177926, "cps": 25.164062, "mean_p": 0.217336, "boundary_entropy": 0.196799, "bpb": 2.044763}, "fi": {"chunks": 3287.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.174738, "cps": 25.679688, "mean_p": 0.195929, "boundary_entropy": 0.143258, "bpb": 2.188166}}, "bpb": {"en": 1.87411, "tr": 2.04476, "fi": 2.18817}, "gini_cpc": 0.021239, "gini_cps": 0.008256, "bpic": 5.759009, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9262493516633614, "fi": 0.9096525980774515}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 51.60368347167969, "n_util_obs": 107, "util_count_spearman": 0.4356766394325908}, "t": 4250.0} |
| {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.89214, "tr": 2.05243, "fi": 2.19855}, "bytes_per_chunk": {"en": 4.916, "tr": 5.885, "fi": 5.719}, "bpb_mean": 2.04771, "frozen_inventory": true, "t": 4251.7} |
| {"event": "train", "step": 5000, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.44097, "bpb": 2.07888, "chunk_loss": 0.03029, "F": 0.19266, "G": 0.20968, "chunks": 197.28, "bytes_per_chunk": 5.191, "boundary_entropy": 0.09907, "router_noise": 0.0, "grad_norm": 0.797, "lr": 5e-05, "router_trainable": true, "sec_per_step": 1.308, "probe_sec": 0.059, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02896, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.13083, "n_shadow": 3, "mean_delta_shadow": -0.23114, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.01637, "trunc_abs_frac": 0.81936, "mean_delta_full": -0.41397, "read_hit_rate": {"surface": 0.10651}, "mean_gate": {"surface": 0.1265}, "n_reads": 669, "n_cand": 6281, "read_suppressed_frac": 0.1091, "dropout_reads_logged": 669, "inventory": {"size": 327310, "admitted": 65536, "staged": 261774, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261774, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 51.6138916015625, "n_util_obs": 107, "util_count_spearman": 0.4359215501871118}, "probe_overhead_sec_total": 333.756, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.03004, "alpha": 0.03, "t": 4258.2} |
| {"event": "consolidate", "step": 5050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261830, "t": 4290.0} |
| {"event": "train", "step": 5050, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.28622, "bpb": 1.85562, "chunk_loss": 0.030297, "F": 0.19778, "G": 0.20994, "chunks": 202.53, "bytes_per_chunk": 5.056, "boundary_entropy": 0.08782, "router_noise": 0.0, "grad_norm": 0.7251, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.654, "probe_sec": 0.0975, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0754, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.27332, "n_shadow": 7, "mean_delta_shadow": -0.63429, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.38677, "trunc_abs_frac": 4.06356, "mean_delta_full": -0.54431, "read_hit_rate": {"surface": 0.13366}, "mean_gate": {"surface": 0.12675}, "n_reads": 862, "n_cand": 6449, "read_suppressed_frac": 0.0719, "dropout_reads_logged": 862, "inventory": {"size": 327008, "admitted": 65536, "staged": 261472, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261472, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 52.96051025390625, "n_util_obs": 112, "util_count_spearman": 0.4253486236611713}, "probe_overhead_sec_total": 337.03, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026723, "alpha": 0.03, "t": 4290.9} |
| {"event": "consolidate", "step": 5100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261827, "t": 4347.6} |
| {"event": "train", "step": 5100, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33751, "bpb": 1.92961, "chunk_loss": 0.030212, "F": 0.19122, "G": 0.20416, "chunks": 195.81, "bytes_per_chunk": 5.229, "boundary_entropy": 0.09, "router_noise": 0.0, "grad_norm": 0.7513, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 1.156, "probe_sec": 0.0573, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02795, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.10685, "n_shadow": 5, "mean_delta_shadow": 0.00828, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.41088, "trunc_abs_frac": 2.26457, "mean_delta_full": 0.07294, "read_hit_rate": {"surface": 0.12737}, "mean_gate": {"surface": 0.12513}, "n_reads": 794, "n_cand": 6234, "read_suppressed_frac": 0.0882, "dropout_reads_logged": 794, "inventory": {"size": 327126, "admitted": 65536, "staged": 261590, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261590, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 53.509185791015625, "n_util_obs": 113, "util_count_spearman": 0.4357658526848094}, "probe_overhead_sec_total": 340.881, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.02875, "alpha": 0.03, "t": 4348.7} |
| {"event": "consolidate", "step": 5150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261820, "t": 4406.4} |
| {"event": "train", "step": 5150, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37668, "bpb": 1.98613, "chunk_loss": 0.030367, "F": 0.19696, "G": 0.21166, "chunks": 201.69, "bytes_per_chunk": 5.077, "boundary_entropy": 0.0955, "router_noise": 0.0, "grad_norm": 0.803, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 1.183, "probe_sec": 0.0529, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02587, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.04507, "n_shadow": 8, "mean_delta_shadow": 0.11185, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 1.1538, "trunc_abs_frac": 11.64957, "mean_delta_full": 0.0709, "read_hit_rate": {"surface": 0.13796}, "mean_gate": {"surface": 0.12385}, "n_reads": 886, "n_cand": 6422, "read_suppressed_frac": 0.1163, "dropout_reads_logged": 886, "inventory": {"size": 327193, "admitted": 65536, "staged": 261657, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261657, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 54.3521728515625, "n_util_obs": 116, "util_count_spearman": 0.44081036404874485}, "probe_overhead_sec_total": 344.735, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023275, "alpha": 0.03, "t": 4407.8} |
| {"event": "consolidate", "step": 5200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261814, "t": 4462.0} |
| {"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3363.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193243, "cps": 26.273438, "mean_p": 0.197734, "boundary_entropy": 0.069379, "bpb": 1.867432}, "tr": {"chunks": 3209.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177263, "cps": 25.070312, "mean_p": 0.217425, "boundary_entropy": 0.197202, "bpb": 2.038884}, "fi": {"chunks": 3325.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176758, "cps": 25.976562, "mean_p": 0.199106, "boundary_entropy": 0.142784, "bpb": 2.182574}}, "bpb": {"en": 1.86743, "tr": 2.03888, "fi": 2.18257}, "gini_cpc": 0.020081, "gini_cps": 0.010374, "bpic": 5.732242, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.917310613767596, "fi": 0.9146965037586194}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 55.32057189941406, "n_util_obs": 119, "util_count_spearman": 0.4246874104254887}, "t": 4462.7} |
| {"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.88578, "tr": 2.04525, "fi": 2.19289}, "bytes_per_chunk": {"en": 4.901, "tr": 5.939, "fi": 5.66}, "bpb_mean": 2.04131, "frozen_inventory": true, "t": 4463.8} |
| {"event": "train", "step": 5200, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32326, "bpb": 1.90906, "chunk_loss": 0.030341, "F": 0.19833, "G": 0.21059, "chunks": 203.09, "bytes_per_chunk": 5.042, "boundary_entropy": 0.08614, "router_noise": 0.0, "grad_norm": 0.7047, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 1.264, "probe_sec": 0.0579, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02754, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.26105, "n_shadow": 6, "mean_delta_shadow": -0.10821, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.03409, "trunc_abs_frac": 1.34939, "mean_delta_full": -0.30121, "read_hit_rate": {"surface": 0.13561}, "mean_gate": {"surface": 0.12389}, "n_reads": 877, "n_cand": 6467, "read_suppressed_frac": 0.1106, "dropout_reads_logged": 877, "inventory": {"size": 327089, "admitted": 65536, "staged": 261553, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261553, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 55.333953857421875, "n_util_obs": 119, "util_count_spearman": 0.4239681383857638}, "probe_overhead_sec_total": 348.435, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024072, "alpha": 0.03, "t": 4471.0} |
| {"event": "consolidate", "step": 5250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261810, "t": 4522.5} |
| {"event": "train", "step": 5250, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31247, "bpb": 1.89349, "chunk_loss": 0.030284, "F": 0.1944, "G": 0.20995, "chunks": 199.06, "bytes_per_chunk": 5.144, "boundary_entropy": 0.09548, "router_noise": 0.0, "grad_norm": 0.726, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 1.044, "probe_sec": 0.066, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02422, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.09783, "n_shadow": 6, "mean_delta_shadow": -0.04777, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.62074, "trunc_abs_frac": 0.34969, "mean_delta_full": -0.18787, "read_hit_rate": {"surface": 0.12149}, "mean_gate": {"surface": 0.12472}, "n_reads": 770, "n_cand": 6338, "read_suppressed_frac": 0.1156, "dropout_reads_logged": 770, "inventory": {"size": 327146, "admitted": 65536, "staged": 261610, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261610, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 56.66468811035156, "n_util_obs": 125, "util_count_spearman": 0.42234715821812596}, "probe_overhead_sec_total": 352.485, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025579, "alpha": 0.03, "t": 4523.2} |
| {"event": "consolidate", "step": 5300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261807, "t": 4568.2} |
| {"event": "train", "step": 5300, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31376, "bpb": 1.89535, "chunk_loss": 0.030282, "F": 0.19702, "G": 0.20727, "chunks": 201.75, "bytes_per_chunk": 5.076, "boundary_entropy": 0.08286, "router_noise": 0.0, "grad_norm": 0.7016, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.925, "probe_sec": 0.0631, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03619, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.28481, "n_shadow": 7, "mean_delta_shadow": -0.11385, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.46945, "trunc_abs_frac": 0.46945, "mean_delta_full": 0.55042, "read_hit_rate": {"surface": 0.13948}, "mean_gate": {"surface": 0.12321}, "n_reads": 896, "n_cand": 6424, "read_suppressed_frac": 0.0815, "dropout_reads_logged": 896, "inventory": {"size": 327073, "admitted": 65536, "staged": 261537, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261537, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 57.62751770019531, "n_util_obs": 128, "util_count_spearman": 0.4071369712506867}, "probe_overhead_sec_total": 356.016, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025008, "alpha": 0.03, "t": 4569.5} |
| {"event": "consolidate", "step": 5350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261808, "t": 4614.2} |
| {"event": "train", "step": 5350, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31364, "bpb": 1.89519, "chunk_loss": 0.030295, "F": 0.1962, "G": 0.20801, "chunks": 200.91, "bytes_per_chunk": 5.097, "boundary_entropy": 0.08633, "router_noise": 0.0, "grad_norm": 0.7275, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.913, "probe_sec": 0.0599, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04062, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.01684, "n_shadow": 3, "mean_delta_shadow": 0.13605, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.19134, "trunc_abs_frac": 1.20238, "mean_delta_full": 0.40546, "read_hit_rate": {"surface": 0.13928}, "mean_gate": {"surface": 0.12279}, "n_reads": 891, "n_cand": 6397, "read_suppressed_frac": 0.0943, "dropout_reads_logged": 891, "inventory": {"size": 327036, "admitted": 65536, "staged": 261500, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261500, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 58.95332336425781, "n_util_obs": 133, "util_count_spearman": 0.40494232088745596}, "probe_overhead_sec_total": 359.544, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026361, "alpha": 0.03, "t": 4615.2} |
| {"event": "consolidate", "step": 5400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261804, "t": 4646.1} |
| {"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3358.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192955, "cps": 26.234375, "mean_p": 0.197293, "boundary_entropy": 0.070142, "bpb": 1.863036}, "tr": {"chunks": 3242.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179086, "cps": 25.328125, "mean_p": 0.219583, "boundary_entropy": 0.19981, "bpb": 2.03123}, "fi": {"chunks": 3337.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177396, "cps": 26.070312, "mean_p": 0.199531, "boundary_entropy": 0.144903, "bpb": 2.173218}}, "bpb": {"en": 1.86304, "tr": 2.03123, "fi": 2.17322}, "gini_cpc": 0.018879, "gini_cps": 0.007782, "bpic": 5.709168, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9281237529789911, "fi": 0.9193645456454094}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 60.88995361328125, "n_util_obs": 141, "util_count_spearman": 0.3938267905304165}, "t": 4646.6} |
| {"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.88174, "tr": 2.03847, "fi": 2.18347}, "bytes_per_chunk": {"en": 4.906, "tr": 5.852, "fi": 5.655}, "bpb_mean": 2.03456, "frozen_inventory": true, "t": 4647.6} |
| {"event": "train", "step": 5400, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38342, "bpb": 1.99585, "chunk_loss": 0.030283, "F": 0.19305, "G": 0.21621, "chunks": 197.69, "bytes_per_chunk": 5.18, "boundary_entropy": 0.114, "router_noise": 0.0, "grad_norm": 0.8851, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.735, "probe_sec": 0.0422, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02753, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.02675, "n_shadow": 4, "mean_delta_shadow": -0.03891, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.53855, "trunc_abs_frac": 0.76506, "mean_delta_full": -0.22443, "read_hit_rate": {"surface": 0.12901}, "mean_gate": {"surface": 0.12095}, "n_reads": 812, "n_cand": 6294, "read_suppressed_frac": 0.1195, "dropout_reads_logged": 812, "inventory": {"size": 327213, "admitted": 65536, "staged": 261677, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261677, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 60.90234375, "n_util_obs": 141, "util_count_spearman": 0.3934885911211953}, "probe_overhead_sec_total": 362.567, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.027903, "alpha": 0.03, "t": 4651.9} |
| {"event": "consolidate", "step": 5450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261808, "t": 4688.8} |
| {"event": "train", "step": 5450, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31612, "bpb": 1.89877, "chunk_loss": 0.030283, "F": 0.19553, "G": 0.20862, "chunks": 200.22, "bytes_per_chunk": 5.114, "boundary_entropy": 0.08743, "router_noise": 0.0, "grad_norm": 0.7177, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.754, "probe_sec": 0.0636, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02918, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.15664, "n_shadow": 6, "mean_delta_shadow": -0.11734, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -6.50566, "trunc_abs_frac": 0.82298, "mean_delta_full": -0.02157, "read_hit_rate": {"surface": 0.14212}, "mean_gate": {"surface": 0.12332}, "n_reads": 906, "n_cand": 6375, "read_suppressed_frac": 0.0784, "dropout_reads_logged": 906, "inventory": {"size": 327077, "admitted": 65536, "staged": 261541, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261541, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 61.6431884765625, "n_util_obs": 142, "util_count_spearman": 0.39355100997294584}, "probe_overhead_sec_total": 365.658, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025538, "alpha": 0.03, "t": 4689.6} |
| {"event": "consolidate", "step": 5500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261805, "t": 4729.5} |
| {"event": "train", "step": 5500, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31493, "bpb": 1.89704, "chunk_loss": 0.030363, "F": 0.20068, "G": 0.21418, "chunks": 205.5, "bytes_per_chunk": 4.983, "boundary_entropy": 0.08776, "router_noise": 0.0, "grad_norm": 0.7252, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.802, "probe_sec": 0.0783, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06946, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.05509, "n_shadow": 5, "mean_delta_shadow": -0.11092, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 2.0, "trunc_abs_frac": 2.0, "mean_delta_full": 0.03467, "read_hit_rate": {"surface": 0.13661}, "mean_gate": {"surface": 0.12176}, "n_reads": 894, "n_cand": 6544, "read_suppressed_frac": 0.0895, "dropout_reads_logged": 894, "inventory": {"size": 327088, "admitted": 65536, "staged": 261552, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261552, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 62.48234558105469, "n_util_obs": 143, "util_count_spearman": 0.38786605341017066}, "probe_overhead_sec_total": 368.996, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.019983, "alpha": 0.03, "t": 4729.7} |
| {"event": "consolidate", "step": 5550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 861, "uncertain_staged": 261798, "t": 4764.0} |
| {"event": "train", "step": 5550, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.30938, "bpb": 1.88903, "chunk_loss": 0.030328, "F": 0.19559, "G": 0.21099, "chunks": 200.28, "bytes_per_chunk": 5.113, "boundary_entropy": 0.09239, "router_noise": 0.0, "grad_norm": 0.6998, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.704, "probe_sec": 0.0755, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04116, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.04665, "n_shadow": 5, "mean_delta_shadow": 0.07922, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.77942, "trunc_abs_frac": 0.84631, "mean_delta_full": 0.36758, "read_hit_rate": {"surface": 0.13361}, "mean_gate": {"surface": 0.11948}, "n_reads": 852, "n_cand": 6377, "read_suppressed_frac": 0.1021, "dropout_reads_logged": 852, "inventory": {"size": 327268, "admitted": 65536, "staged": 261732, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261732, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 63.524810791015625, "n_util_obs": 146, "util_count_spearman": 0.3854800304628228}, "probe_overhead_sec_total": 372.379, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024726, "alpha": 0.03, "t": 4764.9} |
| {"event": "consolidate", "step": 5600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 989, "uncertain_staged": 261794, "t": 4786.1} |
| {"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3362.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193185, "cps": 26.265625, "mean_p": 0.19736, "boundary_entropy": 0.067024, "bpb": 1.857713}, "tr": {"chunks": 3226.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178203, "cps": 25.203125, "mean_p": 0.217019, "boundary_entropy": 0.195198, "bpb": 2.027827}, "fi": {"chunks": 3317.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176333, "cps": 25.914062, "mean_p": 0.198105, "boundary_entropy": 0.14021, "bpb": 2.169416}}, "bpb": {"en": 1.85771, "tr": 2.02783, "fi": 2.16942}, "gini_cpc": 0.020512, "gini_cps": 0.009154, "bpic": 5.727612, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9224444510677762, "fi": 0.9127671447696427}, "inventory": {"size": 326691, "admitted": 65536, "staged": 261155, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261155, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 65.28196716308594, "n_util_obs": 155, "util_count_spearman": 0.3433663111082466}, "t": 4786.5} |
| {"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.87724, "tr": 2.03413, "fi": 2.17962}, "bytes_per_chunk": {"en": 4.901, "tr": 5.873, "fi": 5.689}, "bpb_mean": 2.03033, "frozen_inventory": true, "t": 4787.3} |
| {"event": "train", "step": 5600, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36649, "bpb": 1.97143, "chunk_loss": 0.030246, "F": 0.19189, "G": 0.20851, "chunks": 196.5, "bytes_per_chunk": 5.211, "boundary_entropy": 0.09643, "router_noise": 0.0, "grad_norm": 0.7752, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.557, "probe_sec": 0.1009, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05916, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.03991, "n_shadow": 5, "mean_delta_shadow": -0.23474, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 4.03816, "trunc_abs_frac": 3.73337, "mean_delta_full": -0.03558, "read_hit_rate": {"surface": 0.13635}, "mean_gate": {"surface": 0.12067}, "n_reads": 853, "n_cand": 6256, "read_suppressed_frac": 0.1114, "dropout_reads_logged": 853, "inventory": {"size": 327199, "admitted": 65536, "staged": 261663, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261663, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 65.29498291015625, "n_util_obs": 155, "util_count_spearman": 0.34305049788920755}, "probe_overhead_sec_total": 374.792, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025929, "alpha": 0.03, "t": 4792.8} |
| {"event": "consolidate", "step": 5650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 988, "uncertain_staged": 261795, "t": 4828.9} |
| {"event": "train", "step": 5650, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32006, "bpb": 1.90444, "chunk_loss": 0.03027, "F": 0.19235, "G": 0.20608, "chunks": 196.97, "bytes_per_chunk": 5.199, "boundary_entropy": 0.09126, "router_noise": 0.0, "grad_norm": 0.7453, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.74, "probe_sec": 0.1056, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06935, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.064, "n_shadow": 4, "mean_delta_shadow": -0.098, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.51518, "trunc_abs_frac": 1.22727, "mean_delta_full": 0.21346, "read_hit_rate": {"surface": 0.14065}, "mean_gate": {"surface": 0.12033}, "n_reads": 882, "n_cand": 6271, "read_suppressed_frac": 0.1066, "dropout_reads_logged": 882, "inventory": {"size": 327210, "admitted": 65536, "staged": 261674, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261674, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 67.18051147460938, "n_util_obs": 159, "util_count_spearman": 0.3249808372499392}, "probe_overhead_sec_total": 377.83, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.021988, "alpha": 0.03, "t": 4829.8} |
| {"event": "consolidate", "step": 5700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 738, "uncertain_staged": 261796, "t": 4862.2} |
| {"event": "train", "step": 5700, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.45366, "bpb": 2.09718, "chunk_loss": 0.030263, "F": 0.19183, "G": 0.20967, "chunks": 196.44, "bytes_per_chunk": 5.213, "boundary_entropy": 0.10222, "router_noise": 0.0, "grad_norm": 0.9471, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.672, "probe_sec": 0.0488, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02627, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.1086, "n_shadow": 5, "mean_delta_shadow": 0.07635, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.11644, "trunc_abs_frac": 0.79017, "mean_delta_full": -0.11909, "read_hit_rate": {"surface": 0.12008}, "mean_gate": {"surface": 0.11702}, "n_reads": 751, "n_cand": 6254, "read_suppressed_frac": 0.1052, "dropout_reads_logged": 751, "inventory": {"size": 327650, "admitted": 65536, "staged": 262114, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262114, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 68.23585510253906, "n_util_obs": 163, "util_count_spearman": 0.33609046724414077}, "probe_overhead_sec_total": 380.696, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.021451, "alpha": 0.03, "t": 4863.4} |
| {"event": "consolidate", "step": 5750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261797, "t": 4889.3} |
| {"event": "train", "step": 5750, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.27514, "bpb": 1.83964, "chunk_loss": 0.030366, "F": 0.19913, "G": 0.21313, "chunks": 203.91, "bytes_per_chunk": 5.022, "boundary_entropy": 0.09023, "router_noise": 0.0, "grad_norm": 0.7621, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.529, "probe_sec": 0.0886, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07028, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.0306, "n_shadow": 4, "mean_delta_shadow": 0.05225, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.13509, "trunc_abs_frac": 0.43152, "mean_delta_full": 0.18638, "read_hit_rate": {"surface": 0.15032}, "mean_gate": {"surface": 0.11616}, "n_reads": 976, "n_cand": 6493, "read_suppressed_frac": 0.1025, "dropout_reads_logged": 976, "inventory": {"size": 327020, "admitted": 65536, "staged": 261484, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261484, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 69.03521728515625, "n_util_obs": 163, "util_count_spearman": 0.337304160362667}, "probe_overhead_sec_total": 383.703, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.02104, "alpha": 0.03, "t": 4889.8} |
| {"event": "consolidate", "step": 5800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 715, "uncertain_staged": 261790, "t": 4908.4} |
| {"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3369.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193587, "cps": 26.320312, "mean_p": 0.197324, "boundary_entropy": 0.068279, "bpb": 1.853764}, "tr": {"chunks": 3207.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177153, "cps": 25.054688, "mean_p": 0.217308, "boundary_entropy": 0.197348, "bpb": 2.020672}, "fi": {"chunks": 3331.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177077, "cps": 26.023438, "mean_p": 0.198879, "boundary_entropy": 0.141509, "bpb": 2.162384}}, "bpb": {"en": 1.85376, "tr": 2.02067, "fi": 2.16238}, "gini_cpc": 0.020092, "gini_cps": 0.010901, "bpic": 5.726456, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9151062420150569, "fi": 0.9147151211661505}, "inventory": {"size": 326965, "admitted": 65536, "staged": 261429, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261429, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 69.83009338378906, "n_util_obs": 166, "util_count_spearman": 0.3155891940106643}, "t": 4909.0} |
| {"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.87241, "tr": 2.02727, "fi": 2.17261}, "bytes_per_chunk": {"en": 4.903, "tr": 5.88, "fi": 5.663}, "bpb_mean": 2.0241, "frozen_inventory": true, "t": 4909.9} |
| {"event": "train", "step": 5800, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.26213, "bpb": 1.82087, "chunk_loss": 0.030319, "F": 0.19717, "G": 0.21154, "chunks": 201.91, "bytes_per_chunk": 5.072, "boundary_entropy": 0.09207, "router_noise": 0.0, "grad_norm": 0.7393, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.535, "probe_sec": 0.0621, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03475, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.57525, "n_shadow": 8, "mean_delta_shadow": 0.09807, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.14855}, "mean_gate": {"surface": 0.11867}, "n_reads": 955, "n_cand": 6429, "read_suppressed_frac": 0.1037, "dropout_reads_logged": 955, "inventory": {"size": 327337, "admitted": 65536, "staged": 261801, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261801, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 69.84466552734375, "n_util_obs": 166, "util_count_spearman": 0.31560231125919025}, "probe_overhead_sec_total": 386.329, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.02489, "alpha": 0.03, "t": 4916.6} |
| {"event": "consolidate", "step": 5850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 877, "uncertain_staged": 261787, "t": 4942.5} |
| {"event": "train", "step": 5850, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38198, "bpb": 1.99377, "chunk_loss": 0.03033, "F": 0.19461, "G": 0.21399, "chunks": 199.28, "bytes_per_chunk": 5.138, "boundary_entropy": 0.10626, "router_noise": 0.0, "grad_norm": 0.7821, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.535, "probe_sec": 0.0863, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06946, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.072, "n_shadow": 7, "mean_delta_shadow": -0.18125, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.37258, "trunc_abs_frac": 0.93533, "mean_delta_full": -0.36165, "read_hit_rate": {"surface": 0.13948}, "mean_gate": {"surface": 0.11594}, "n_reads": 885, "n_cand": 6345, "read_suppressed_frac": 0.0904, "dropout_reads_logged": 885, "inventory": {"size": 327427, "admitted": 65536, "staged": 261891, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261891, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 70.73701477050781, "n_util_obs": 168, "util_count_spearman": 0.31173662818066317}, "probe_overhead_sec_total": 388.905, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022977, "alpha": 0.03, "t": 4943.3} |
| {"event": "consolidate", "step": 5900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 717, "uncertain_staged": 261785, "t": 4960.9} |
| {"event": "train", "step": 5900, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36351, "bpb": 1.96713, "chunk_loss": 0.030339, "F": 0.19788, "G": 0.21223, "chunks": 202.62, "bytes_per_chunk": 5.054, "boundary_entropy": 0.09362, "router_noise": 0.0, "grad_norm": 0.7474, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.362, "probe_sec": 0.0256, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01878, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.19512, "n_shadow": 6, "mean_delta_shadow": -0.49941, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.28714, "trunc_abs_frac": 0.67445, "mean_delta_full": 0.2248, "read_hit_rate": {"surface": 0.146}, "mean_gate": {"surface": 0.11837}, "n_reads": 942, "n_cand": 6452, "read_suppressed_frac": 0.0998, "dropout_reads_logged": 942, "inventory": {"size": 327414, "admitted": 65536, "staged": 261878, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261878, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 71.6280517578125, "n_util_obs": 172, "util_count_spearman": 0.3342059275086493}, "probe_overhead_sec_total": 391.265, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023083, "alpha": 0.03, "t": 4961.4} |
| {"event": "consolidate", "step": 5950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 914, "uncertain_staged": 261779, "t": 4992.4} |
| {"event": "train", "step": 5950, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31184, "bpb": 1.89258, "chunk_loss": 0.030243, "F": 0.19034, "G": 0.20611, "chunks": 194.91, "bytes_per_chunk": 5.254, "boundary_entropy": 0.09584, "router_noise": 0.0, "grad_norm": 0.7463, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.623, "probe_sec": 0.0337, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02109, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.0004, "n_shadow": 3, "mean_delta_shadow": -0.338, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 10, "trunc_ratio_sum": 0.28472, "trunc_abs_frac": 1.31295, "mean_delta_full": -0.32675, "read_hit_rate": {"surface": 0.14376}, "mean_gate": {"surface": 0.11563}, "n_reads": 892, "n_cand": 6205, "read_suppressed_frac": 0.0953, "dropout_reads_logged": 892, "inventory": {"size": 327263, "admitted": 65536, "staged": 261727, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261727, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 73.25633239746094, "n_util_obs": 177, "util_count_spearman": 0.3217677152403871}, "probe_overhead_sec_total": 394.105, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022154, "alpha": 0.03, "t": 4992.5} |
| {"event": "consolidate", "step": 6000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 136, "uncertain_staged": 261782, "t": 5020.8} |
| {"event": "train", "step": 5999, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35666, "bpb": 1.95724, "chunk_loss": 0.03032, "F": 0.19693, "G": 0.20942, "chunks": 201.66, "bytes_per_chunk": 5.078, "boundary_entropy": 0.08677, "router_noise": 0.0, "grad_norm": 0.7427, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.565, "probe_sec": 0.0702, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05203, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.16133, "n_shadow": 5, "mean_delta_shadow": -0.15871, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.84453, "trunc_abs_frac": 0.84453, "mean_delta_full": -0.79968, "read_hit_rate": {"surface": 0.14764}, "mean_gate": {"surface": 0.11731}, "n_reads": 948, "n_cand": 6421, "read_suppressed_frac": 0.1076, "dropout_reads_logged": 948, "inventory": {"size": 327544, "admitted": 65536, "staged": 262008, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262008, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 74.45860290527344, "n_util_obs": 179, "util_count_spearman": 0.30405080241876425}, "probe_overhead_sec_total": 396.65, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022809, "alpha": 0.03, "t": 5020.8} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3347.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192323, "cps": 26.148438, "mean_p": 0.196445, "boundary_entropy": 0.067322, "bpb": 1.852313}, "tr": {"chunks": 3199.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176711, "cps": 24.992188, "mean_p": 0.215687, "boundary_entropy": 0.19266, "bpb": 2.016337}, "fi": {"chunks": 3283.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.174526, "cps": 25.648438, "mean_p": 0.196468, "boundary_entropy": 0.138808, "bpb": 2.158804}}, "bpb": {"en": 1.85231, "tr": 2.01634, "fi": 2.1588}, "gini_cpc": 0.021828, "gini_cps": 0.010038, "bpic": 5.771899, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9188235047331735, "fi": 0.907459825115199}, "inventory": {"size": 327544, "admitted": 65536, "staged": 262008, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262008, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 74.45860290527344, "n_util_obs": 179, "util_count_spearman": 0.30405080241876425}, "t": 5021.3} |
| {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.87054, "tr": 2.0226, "fi": 2.1696}, "bytes_per_chunk": {"en": 4.917, "tr": 5.926, "fi": 5.707}, "bpb_mean": 2.02091, "frozen_inventory": true, "t": 5022.4} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s1/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s1/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s1/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s1/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s1/probe/reps/th0.70"], "t": 5026.8} |
| {"event": "final", "run": "ali_ali-surface-only_s1", "arch": "ali", "objective": "ratio", "condition": "A", "step": 6000, "params": 22785088, "seed": 1, "per_lang": {"en": {"cpc": 0.19232316267310234, "cps": 26.1484375, "bpb": 1.875639544137256, "bytes_per_chunk": 4.908787895811097, "reads_per_chunk": 0.15780012358856244}, "tr": {"cpc": 0.17671104236866816, "cps": 24.9921875, "bpb": 2.014741552158955, "bytes_per_chunk": 5.944375786573544, "reads_per_chunk": 0.11853875894284516}, "fi": {"cpc": 0.1745255435649354, "cps": 25.6484375, "bpb": 2.1875872872062025, "bytes_per_chunk": 5.681860545765871, "reads_per_chunk": 0.10773348939029412}}, "bpb_mean": 2.025989461167471, "gini_cpc": 0.021828473200707466, "gini_cps": 0.010038321972394737, "bpic": 5.771899481127276, "mean_boundary_entropy": 0.08677133917808533, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "probe_overhead_sec_total": 396.6500906944275, "probe_overhead_seq_frac": 0.4999166666666667, "wall_sec": 5028.66376209259, "inventory": {"size": 327544, "admitted": 65536, "staged": 262008, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262008, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 74.45860290527344, "n_util_obs": 179, "util_count_spearman": 0.30405080241876425}, "ali_params": 268096, "ali_config": {"enabled": true, "budget_B": 65536, "staging_capacity": 262144, "policy": "utility", "candidate_types": ["surface"], "gate": "vector", "retrieval": true, "memory_scope": "persistent", "read_dropout": 0.1, "write_rule": "mean", "eta_min": 0.0, "consolidate_every": 50, "probe_seqs": 2, "probe_reps": 8, "horizon": 32, "predictor": "mlp", "lambda_util": 0.1, "max_span": 64, "byte_classes_path": "/local/scratch/sas245/beetle-hnet/scratch/data/byte_classes.json", "gate_hidden": 0, "type_dim": 16, "gate_bias_init": -2.0}, "util_vs_frequency_spearman": 0.30405080241876425, "paired_vs_dropout": {"n": 290, "pearson": 0.08867178982987495, "spearman": 0.09496959863805834, "mean_abs_discrepancy": 1.5305712223052979, "mean_paired": 0.04795457422733307, "mean_dropout": -0.2840459942817688}, "predictor_calibration": {"n": 3861, "all": {"n": 3861, "pearson": 0.11957240591027775, "spearman": 0.13307890437234768, "mae": 0.13828004896640778, "reliability": [{"pred": -0.043622732162475586, "obs": -0.09359260648488998, "n": 386}, {"pred": -0.03244829922914505, "obs": -0.07524032145738602, "n": 386}, {"pred": -0.030230719596147537, "obs": -0.0725184753537178, "n": 386}, {"pred": -0.02858923375606537, "obs": -0.04640570282936096, "n": 386}, {"pred": -0.02678457647562027, "obs": -0.0839860662817955, "n": 386}, {"pred": -0.024645311757922173, "obs": -0.05091825872659683, "n": 387}, {"pred": -0.021980097517371178, "obs": -0.030947843566536903, "n": 385}, {"pred": -0.01830235868692398, "obs": -0.047378309071063995, "n": 386}, {"pred": -0.008728406392037868, "obs": -0.019148128107190132, "n": 386}, {"pred": 0.015720533207058907, "obs": 0.0024626790545880795, "n": 387}]}, "admitted": {"n": 673, "pearson": -0.15085461266206335, "spearman": -0.1291867491801365, "mae": 0.10682331025600433, "reliability": [{"pred": -0.0453169159591198, "obs": 0.10118170827627182, "n": 68}, {"pred": -0.026866426691412926, "obs": 0.035534102469682693, "n": 67}, {"pred": -0.016908537596464157, "obs": 0.0515558086335659, "n": 67}, {"pred": -0.002194872358813882, "obs": 0.002599082887172699, "n": 67}, {"pred": 0.002584868110716343, "obs": -0.007116600871086121, "n": 67}, {"pred": 0.005670230835676193, "obs": -0.05251315236091614, "n": 68}, {"pred": 0.009499579668045044, "obs": 0.0032736591529101133, "n": 67}, {"pred": 0.013429771177470684, "obs": 0.018256882205605507, "n": 67}, {"pred": 0.0196695514023304, "obs": 0.02072785049676895, "n": 67}, {"pred": 0.03902362287044525, "obs": 0.005367133300751448, "n": 68}]}, "staged": {"n": 3188, "pearson": 0.1313404807838373, "spearman": 0.1149103473702627, "mae": 0.14492067694664001, "reliability": [{"pred": -0.043184101581573486, "obs": -0.13311763107776642, "n": 319}, {"pred": -0.0327070951461792, "obs": -0.08183448761701584, "n": 319}, {"pred": -0.030713099986314774, "obs": -0.06850270181894302, "n": 319}, {"pred": -0.029209403321146965, "obs": -0.06776915490627289, "n": 318}, {"pred": -0.027830740436911583, "obs": -0.06021697074174881, "n": 319}, {"pred": -0.026130879297852516, "obs": -0.08210696280002594, "n": 319}, {"pred": -0.024189531803131104, "obs": -0.049394555389881134, "n": 318}, {"pred": -0.021769022569060326, "obs": -0.04106095805764198, "n": 319}, {"pred": -0.01857171393930912, "obs": -0.05606129392981529, "n": 319}, {"pred": -0.011259413324296474, "obs": -0.024495001882314682, "n": 319}]}}, "t": 5028.7} |
|
|
