| {"event": "start", "run": "parity_B_ste_s0", "arch": "hnet", "objective": "parity", "condition": "B", "params": 22516992, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "parity_a", "p3_objective": "none", "freeze_in_p3": true, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "device": "cuda", "t": 1.9} |
| {"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": 1.9} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6384.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.366833, "cps": 49.875, "mean_p": 0.496725, "boundary_entropy": 0.686995, "bpb": 8.125775}, "de": {"chunks": 8628.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.420202, "cps": 67.40625, "mean_p": 0.498889, "boundary_entropy": 0.687821, "bpb": 8.121816}, "vi": {"chunks": 13115.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.711418, "cps": 102.460938, "mean_p": 0.505745, "boundary_entropy": 0.688385, "bpb": 8.087584}, "tr": {"chunks": 5290.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.292217, "cps": 41.328125, "mean_p": 0.489567, "boundary_entropy": 0.687288, "bpb": 8.14748}, "fi": {"chunks": 8895.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.472862, "cps": 69.492188, "mean_p": 0.500688, "boundary_entropy": 0.687512, "bpb": 8.208588}, "ru": {"chunks": 10744.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.566966, "cps": 83.9375, "mean_p": 0.493655, "boundary_entropy": 0.689893, "bpb": 8.082947}, "el": {"chunks": 15039.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.732788, "cps": 117.492188, "mean_p": 0.497482, "boundary_entropy": 0.690234, "bpb": 8.196862}, "hi": {"chunks": 2563.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.152333, "cps": 20.023438, "mean_p": 0.484638, "boundary_entropy": 0.690256, "bpb": 8.272031}, "ta": {"chunks": 38089.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.03739, "cps": 297.570312, "mean_p": 0.512976, "boundary_entropy": 0.690378, "bpb": 8.008816}, "th": {"chunks": 34006.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.114144, "cps": 265.671875, "mean_p": 0.511916, "boundary_entropy": 0.690552, "bpb": 8.270247}, "zh": {"chunks": 11821.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.001524, "cps": 92.351562, "mean_p": 0.52004, "boundary_entropy": 0.685674, "bpb": 8.131836}, "my": {"chunks": 48201.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.543857, "cps": 376.570312, "mean_p": 0.521563, "boundary_entropy": 0.690206, "bpb": 8.106955}}, "bpb": {"en": 8.12577, "de": 8.12182, "vi": 8.08758, "tr": 8.14748, "fi": 8.20859, "ru": 8.08295, "el": 8.19686, "hi": 8.27203, "ta": 8.00882, "th": 8.27025, "zh": 8.13184, "my": 8.10696}, "gini_cpc": 0.424496, "gini_cps": 0.425905, "bpic": 1.892645, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1454838515955792, "vi": 1.939350894526797, "tr": 0.7965927738475616, "fi": 1.2890367847879758, "ru": 1.5455676790921895, "el": 1.997603802482379, "hi": 0.41526447061189375, "ta": 5.553993192060118, "th": 5.7632270176529286, "zh": 5.456221801328583, "my": 6.934639901884223}, "t": 4.8} |
| {"event": "train", "step": 0, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 5.63738, "bpb": 8.13302, "chunk_loss": 0.05247, "F": 0.47223, "G": 0.4995, "chunks": 483.56, "bytes_per_chunk": 2.118, "boundary_entropy": 0.69158, "router_noise": 0.0, "grad_norm": 7.4901, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.008, "alpha": 0.03, "t": 5.2} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6343.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.364477, "cps": 49.554688, "mean_p": 0.49658, "boundary_entropy": 0.68699, "bpb": 8.105599}, "de": {"chunks": 8608.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.419228, "cps": 67.25, "mean_p": 0.498815, "boundary_entropy": 0.68782, "bpb": 8.104416}, "vi": {"chunks": 13053.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.708055, "cps": 101.976562, "mean_p": 0.505613, "boundary_entropy": 0.688386, "bpb": 8.073837}, "tr": {"chunks": 5261.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.290615, "cps": 41.101562, "mean_p": 0.489476, "boundary_entropy": 0.687283, "bpb": 8.134032}, "fi": {"chunks": 8851.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.470523, "cps": 69.148438, "mean_p": 0.500541, "boundary_entropy": 0.68751, "bpb": 8.192063}, "ru": {"chunks": 10637.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.561319, "cps": 83.101562, "mean_p": 0.493511, "boundary_entropy": 0.689889, "bpb": 8.047803}, "el": {"chunks": 14969.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.729377, "cps": 116.945312, "mean_p": 0.497407, "boundary_entropy": 0.690233, "bpb": 8.183434}, "hi": {"chunks": 2504.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.148826, "cps": 19.5625, "mean_p": 0.484458, "boundary_entropy": 0.690244, "bpb": 8.262059}, "ta": {"chunks": 37842.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.024178, "cps": 295.640625, "mean_p": 0.512671, "boundary_entropy": 0.690392, "bpb": 7.987582}, "th": {"chunks": 33877.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.106124, "cps": 264.664062, "mean_p": 0.511786, "boundary_entropy": 0.690558, "bpb": 8.25511}, "zh": {"chunks": 11817.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.000847, "cps": 92.320312, "mean_p": 0.520017, "boundary_entropy": 0.685676, "bpb": 8.129336}, "my": {"chunks": 48279.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.547973, "cps": 377.179688, "mean_p": 0.521729, "boundary_entropy": 0.690191, "bpb": 8.087031}}, "bpb": {"en": 8.1056, "de": 8.10442, "vi": 8.07384, "tr": 8.13403, "fi": 8.19206, "ru": 8.0478, "el": 8.18343, "hi": 8.26206, "ta": 7.98758, "th": 8.25511, "zh": 8.12934, "my": 8.08703}, "gini_cpc": 0.42557, "gini_cps": 0.426928, "bpic": 1.89952, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1502156171177025, "vi": 1.9426591360426666, "tr": 0.7973466229454836, "fi": 1.2909513205503071, "ru": 1.5400660732937688, "el": 2.001157873891602, "hi": 0.4083275288667481, "ta": 5.553643789505498, "th": 5.778475651898752, "zh": 5.489631613123976, "my": 6.9907584889847305}, "t": 6.6} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6283.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.36103, "cps": 49.085938, "mean_p": 0.496334, "boundary_entropy": 0.686981, "bpb": 8.060631}, "de": {"chunks": 8562.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.416987, "cps": 66.890625, "mean_p": 0.498717, "boundary_entropy": 0.687817, "bpb": 8.065172}, "vi": {"chunks": 12904.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.699973, "cps": 100.8125, "mean_p": 0.505287, "boundary_entropy": 0.688389, "bpb": 8.042936}, "tr": {"chunks": 5222.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.28846, "cps": 40.796875, "mean_p": 0.489332, "boundary_entropy": 0.687276, "bpb": 8.102083}, "fi": {"chunks": 8787.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.46712, "cps": 68.648438, "mean_p": 0.500294, "boundary_entropy": 0.687505, "bpb": 8.154054}, "ru": {"chunks": 10492.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.553668, "cps": 81.96875, "mean_p": 0.493302, "boundary_entropy": 0.689883, "bpb": 7.989209}, "el": {"chunks": 14969.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.729377, "cps": 116.945312, "mean_p": 0.497401, "boundary_entropy": 0.690233, "bpb": 8.168167}, "hi": {"chunks": 2423.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.144012, "cps": 18.929688, "mean_p": 0.484205, "boundary_entropy": 0.690227, "bpb": 8.235433}, "ta": {"chunks": 37437.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.002514, "cps": 292.476562, "mean_p": 0.512172, "boundary_entropy": 0.690416, "bpb": 7.950504}, "th": {"chunks": 33651.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.092073, "cps": 262.898438, "mean_p": 0.511603, "boundary_entropy": 0.690565, "bpb": 8.233769}, "zh": {"chunks": 11803.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.998476, "cps": 92.210938, "mean_p": 0.519959, "boundary_entropy": 0.685679, "bpb": 8.123245}, "my": {"chunks": 48412.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.554993, "cps": 378.21875, "mean_p": 0.521973, "boundary_entropy": 0.690169, "bpb": 8.054594}}, "bpb": {"en": 8.06063, "de": 8.06517, "vi": 8.04294, "tr": 8.10208, "fi": 8.15405, "ru": 7.98921, "el": 8.16817, "hi": 8.23543, "ta": 7.9505, "th": 8.23377, "zh": 8.12324, "my": 8.05459}, "gini_cpc": 0.427098, "gini_cps": 0.428429, "bpic": 1.909881, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.154994395384025, "vi": 1.9388234904083979, "tr": 0.7989937397238708, "fi": 1.2938555809092809, "ru": 1.5335789123139585, "el": 2.02026808755283, "hi": 0.39889206919116393, "ta": 5.546673859427944, "th": 5.794740201948199, "zh": 5.535489419117988, "my": 7.076959480415855}, "t": 8.3} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6013.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.345515, "cps": 46.976562, "mean_p": 0.495308, "boundary_entropy": 0.686943, "bpb": 7.908855}, "de": {"chunks": 8400.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.409098, "cps": 65.625, "mean_p": 0.498184, "boundary_entropy": 0.687805, "bpb": 7.930131}, "vi": {"chunks": 12514.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.678817, "cps": 97.765625, "mean_p": 0.504185, "boundary_entropy": 0.688395, "bpb": 7.941786}, "tr": {"chunks": 5040.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.278407, "cps": 39.375, "mean_p": 0.488692, "boundary_entropy": 0.687243, "bpb": 7.994076}, "fi": {"chunks": 8462.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.449843, "cps": 66.109375, "mean_p": 0.499195, "boundary_entropy": 0.687484, "bpb": 8.023531}, "ru": {"chunks": 9653.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.509393, "cps": 75.414062, "mean_p": 0.492207, "boundary_entropy": 0.689848, "bpb": 7.791348}, "el": {"chunks": 14754.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.718901, "cps": 115.265625, "mean_p": 0.497197, "boundary_entropy": 0.690228, "bpb": 8.098101}, "hi": {"chunks": 2128.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.126478, "cps": 16.625, "mean_p": 0.483036, "boundary_entropy": 0.690144, "bpb": 8.137073}, "ta": {"chunks": 35787.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.914255, "cps": 279.585938, "mean_p": 0.510287, "boundary_entropy": 0.690497, "bpb": 7.818933}, "th": {"chunks": 32662.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.030588, "cps": 255.171875, "mean_p": 0.5108, "boundary_entropy": 0.690594, "bpb": 8.143551}, "zh": {"chunks": 11758.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.990857, "cps": 91.859375, "mean_p": 0.519677, "boundary_entropy": 0.685698, "bpb": 8.097824}, "my": {"chunks": 48597.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.564756, "cps": 379.664062, "mean_p": 0.522461, "boundary_entropy": 0.690122, "bpb": 7.976977}}, "bpb": {"en": 7.90885, "de": 7.93013, "vi": 7.94179, "tr": 7.99408, "fi": 8.02353, "ru": 7.79135, "el": 8.0981, "hi": 8.13707, "ta": 7.81893, "th": 8.14355, "zh": 8.09782, "my": 7.97698}, "gini_cpc": 0.434183, "gini_cps": 0.434382, "bpic": 1.960387, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1840220634638683, "vi": 1.9646533135353466, "tr": 0.8057733534476145, "fi": 1.3019492444660972, "ru": 1.4743004843946295, "el": 2.0806635080698928, "hi": 0.3660576329295442, "ta": 5.540293089675943, "th": 5.8769855862499965, "zh": 5.761995696309649, "my": 7.422992135363878}, "t": 10.1} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5171.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.297133, "cps": 40.398438, "mean_p": 0.492005, "boundary_entropy": 0.68679, "bpb": 7.479582}, "de": {"chunks": 7862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.382896, "cps": 61.421875, "mean_p": 0.496423, "boundary_entropy": 0.687763, "bpb": 7.540031}, "vi": {"chunks": 11072.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.600597, "cps": 86.5, "mean_p": 0.500661, "boundary_entropy": 0.688379, "bpb": 7.633172}, "tr": {"chunks": 4419.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.244103, "cps": 34.523438, "mean_p": 0.486566, "boundary_entropy": 0.687118, "bpb": 7.665698}, "fi": {"chunks": 7409.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.393865, "cps": 57.882812, "mean_p": 0.495672, "boundary_entropy": 0.687386, "bpb": 7.637518}, "ru": {"chunks": 6532.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.344697, "cps": 51.03125, "mean_p": 0.487689, "boundary_entropy": 0.689644, "bpb": 7.249578}, "el": {"chunks": 13140.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.640257, "cps": 102.65625, "mean_p": 0.495739, "boundary_entropy": 0.690195, "bpb": 7.831401}, "hi": {"chunks": 1172.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.069658, "cps": 9.15625, "mean_p": 0.477945, "boundary_entropy": 0.68972, "bpb": 7.799842}, "ta": {"chunks": 28552.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.527253, "cps": 223.0625, "mean_p": 0.503461, "boundary_entropy": 0.690678, "bpb": 7.39242}, "th": {"chunks": 28025.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.742306, "cps": 218.945312, "mean_p": 0.506755, "boundary_entropy": 0.69071, "bpb": 7.702984}, "zh": {"chunks": 11560.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.957332, "cps": 90.3125, "mean_p": 0.518678, "boundary_entropy": 0.68576, "bpb": 8.01338}, "my": {"chunks": 49639.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.619749, "cps": 387.804688, "mean_p": 0.525227, "boundary_entropy": 0.689847, "bpb": 7.651174}}, "bpb": {"en": 7.47958, "de": 7.54003, "vi": 7.63317, "tr": 7.6657, "fi": 7.63752, "ru": 7.24958, "el": 7.8314, "hi": 7.79984, "ta": 7.39242, "th": 7.70298, "zh": 8.01338, "my": 7.65117}, "gini_cpc": 0.465794, "gini_cps": 0.457843, "bpic": 2.19865, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2886358661568635, "vi": 2.0213081057452693, "tr": 0.8215292533123355, "fi": 1.3255536007442907, "ru": 1.1600762727388232, "el": 2.154785788113469, "hi": 0.23443482238785687, "ta": 5.139970753635663, "th": 5.8637323462963, "zh": 6.587399065734475, "my": 8.816764286481497}, "t": 12.0} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2561.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.147159, "cps": 20.007812, "mean_p": 0.480125, "boundary_entropy": 0.685967, "bpb": 6.886635}, "de": {"chunks": 4867.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.237033, "cps": 38.023438, "mean_p": 0.487289, "boundary_entropy": 0.687439, "bpb": 6.972577}, "vi": {"chunks": 7622.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.413453, "cps": 59.546875, "mean_p": 0.491765, "boundary_entropy": 0.688149, "bpb": 7.19776}, "tr": {"chunks": 2857.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.157819, "cps": 22.320312, "mean_p": 0.480193, "boundary_entropy": 0.686637, "bpb": 7.184915}, "fi": {"chunks": 4150.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.220616, "cps": 32.421875, "mean_p": 0.484379, "boundary_entropy": 0.686812, "bpb": 7.078723}, "ru": {"chunks": 708.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.037361, "cps": 5.53125, "mean_p": 0.46809, "boundary_entropy": 0.687785, "bpb": 6.529837}, "el": {"chunks": 4385.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.213663, "cps": 34.257812, "mean_p": 0.483304, "boundary_entropy": 0.689588, "bpb": 7.204378}, "hi": {"chunks": 367.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.021813, "cps": 2.867188, "mean_p": 0.465299, "boundary_entropy": 0.688219, "bpb": 7.148144}, "ta": {"chunks": 14379.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.769136, "cps": 112.335938, "mean_p": 0.489884, "boundary_entropy": 0.690512, "bpb": 6.707952}, "th": {"chunks": 15795.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.981971, "cps": 123.398438, "mean_p": 0.493995, "boundary_entropy": 0.69066, "bpb": 6.554702}, "zh": {"chunks": 10988.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.860481, "cps": 85.84375, "mean_p": 0.515845, "boundary_entropy": 0.685925, "bpb": 7.796966}, "my": {"chunks": 52123.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.750844, "cps": 407.210938, "mean_p": 0.533894, "boundary_entropy": 0.688774, "bpb": 6.828333}}, "bpb": {"en": 6.88663, "de": 6.97258, "vi": 7.19776, "tr": 7.18492, "fi": 7.07872, "ru": 6.52984, "el": 7.20438, "hi": 7.14814, "ta": 6.70795, "th": 6.5547, "zh": 7.79697, "my": 6.82833}, "gini_cpc": 0.596818, "gini_cps": 0.579032, "bpic": 3.176942, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.6107327196058694, "vi": 2.8095731522693312, "tr": 1.0724430292079146, "fi": 1.4991695583999416, "ru": 0.2538859016771771, "el": 1.451922028109794, "hi": 0.14822639048523603, "ta": 5.226581849515314, "th": 6.672876801383512, "zh": 12.642697589582886, "my": 18.69306730840327}, "t": 14.1} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 1732.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.099523, "cps": 13.53125, "mean_p": 0.484147, "boundary_entropy": 0.686808, "bpb": 6.387136}, "de": {"chunks": 4742.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.230945, "cps": 37.046875, "mean_p": 0.491202, "boundary_entropy": 0.688101, "bpb": 6.470757}, "vi": {"chunks": 5583.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.302848, "cps": 43.617188, "mean_p": 0.48879, "boundary_entropy": 0.688309, "bpb": 6.932916}, "tr": {"chunks": 4703.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.259791, "cps": 36.742188, "mean_p": 0.491653, "boundary_entropy": 0.687805, "bpb": 6.739405}, "fi": {"chunks": 3467.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184307, "cps": 27.085938, "mean_p": 0.488017, "boundary_entropy": 0.68759, "bpb": 6.544552}, "ru": {"chunks": 3728.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.196728, "cps": 29.125, "mean_p": 0.485623, "boundary_entropy": 0.689697, "bpb": 5.838962}, "el": {"chunks": 222.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.010817, "cps": 1.734375, "mean_p": 0.446241, "boundary_entropy": 0.684195, "bpb": 6.772894}, "hi": {"chunks": 441.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.026211, "cps": 3.445312, "mean_p": 0.476745, "boundary_entropy": 0.689696, "bpb": 6.160623}, "ta": {"chunks": 3159.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.168976, "cps": 24.679688, "mean_p": 0.477764, "boundary_entropy": 0.689824, "bpb": 5.974537}, "th": {"chunks": 3939.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.244887, "cps": 30.773438, "mean_p": 0.487447, "boundary_entropy": 0.69058, "bpb": 5.602339}, "zh": {"chunks": 8908.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.508297, "cps": 69.59375, "mean_p": 0.506249, "boundary_entropy": 0.686321, "bpb": 7.386188}, "my": {"chunks": 53214.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.808423, "cps": 415.734375, "mean_p": 0.541255, "boundary_entropy": 0.687626, "bpb": 5.834365}}, "bpb": {"en": 6.38714, "de": 6.47076, "vi": 6.93292, "tr": 6.73941, "fi": 6.54455, "ru": 5.83896, "el": 6.77289, "hi": 6.16062, "ta": 5.97454, "th": 5.60234, "zh": 7.38619, "my": 5.83437}, "gini_cpc": 0.637312, "gini_cps": 0.619779, "bpic": 4.089825, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 2.3205203160259456, "vi": 3.042991354055288, "tr": 2.6103615263390814, "fi": 1.8519028102936677, "ru": 1.9767098295624197, "el": 0.10868969293618526, "hi": 0.26336602507129153, "ta": 1.697854182893377, "th": 2.460600727514984, "zh": 15.155246279471537, "my": 28.218814309345962}, "t": 17.3} |
| {"event": "train", "step": 50, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 4.2327, "bpb": 6.10649, "chunk_loss": 0.039679, "F": 0.29611, "G": 0.4924, "chunks": 303.22, "bytes_per_chunk": 3.377, "boundary_entropy": 0.69175, "router_noise": 0.0, "grad_norm": 1.5923, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.287, "alpha": 0.03, "t": 19.6} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2360.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.135609, "cps": 18.4375, "mean_p": 0.489219, "boundary_entropy": 0.687276, "bpb": 5.811872}, "de": {"chunks": 3858.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.187893, "cps": 30.140625, "mean_p": 0.490697, "boundary_entropy": 0.688184, "bpb": 5.928605}, "vi": {"chunks": 3593.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.194901, "cps": 28.070312, "mean_p": 0.483365, "boundary_entropy": 0.688041, "bpb": 6.283079}, "tr": {"chunks": 3804.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210131, "cps": 29.71875, "mean_p": 0.490946, "boundary_entropy": 0.687908, "bpb": 6.223389}, "fi": {"chunks": 2940.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.156292, "cps": 22.96875, "mean_p": 0.490069, "boundary_entropy": 0.687904, "bpb": 5.977518}, "ru": {"chunks": 7479.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.39467, "cps": 58.429688, "mean_p": 0.494305, "boundary_entropy": 0.690259, "bpb": 5.383825}, "el": {"chunks": 659.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.03211, "cps": 5.148438, "mean_p": 0.479871, "boundary_entropy": 0.689625, "bpb": 5.74863}, "hi": {"chunks": 8960.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.532541, "cps": 70.0, "mean_p": 0.491356, "boundary_entropy": 0.690634, "bpb": 5.169952}, "ta": {"chunks": 3147.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.168334, "cps": 24.585938, "mean_p": 0.479132, "boundary_entropy": 0.689863, "bpb": 5.13703}, "th": {"chunks": 9989.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.621013, "cps": 78.039062, "mean_p": 0.489956, "boundary_entropy": 0.690503, "bpb": 4.881402}, "zh": {"chunks": 3503.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.593126, "cps": 27.367188, "mean_p": 0.488421, "boundary_entropy": 0.686314, "bpb": 6.995373}, "my": {"chunks": 38069.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.00913, "cps": 297.414062, "mean_p": 0.509662, "boundary_entropy": 0.690889, "bpb": 4.935646}}, "bpb": {"en": 5.81187, "de": 5.92861, "vi": 6.28308, "tr": 6.22339, "fi": 5.97752, "ru": 5.38382, "el": 5.74863, "hi": 5.16995, "ta": 5.13703, "th": 4.8814, "zh": 6.99537, "my": 4.93565}, "gini_cpc": 0.503435, "gini_cps": 0.514223, "bpic": 4.34333, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.3855491408208531, "vi": 1.4372297306615547, "tr": 1.549537439716425, "fi": 1.1525175947358604, "ru": 2.910358145878986, "el": 0.23678636288182667, "hi": 3.927037550054146, "ta": 1.2413189650092247, "th": 4.579447295353603, "zh": 4.373798908894718, "my": 14.815632949946863}, "t": 21.8} |
| {"event": "train", "step": 100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 3.25925, "bpb": 4.7021, "chunk_loss": 0.038609, "F": 0.29138, "G": 0.47384, "chunks": 298.38, "bytes_per_chunk": 3.432, "boundary_entropy": 0.68612, "router_noise": 0.0, "grad_norm": 1.8565, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 24.9} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5530.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.317761, "cps": 43.203125, "mean_p": 0.445619, "boundary_entropy": 0.667706, "bpb": 4.444776}, "de": {"chunks": 5877.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.286222, "cps": 45.914062, "mean_p": 0.44244, "boundary_entropy": 0.668233, "bpb": 4.663788}, "vi": {"chunks": 6848.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.371467, "cps": 53.5, "mean_p": 0.4727, "boundary_entropy": 0.683109, "bpb": 4.820838}, "tr": {"chunks": 8738.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.482682, "cps": 68.265625, "mean_p": 0.467173, "boundary_entropy": 0.674117, "bpb": 4.91265}, "fi": {"chunks": 7580.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.402956, "cps": 59.21875, "mean_p": 0.466351, "boundary_entropy": 0.675401, "bpb": 4.677465}, "ru": {"chunks": 10287.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.54285, "cps": 80.367188, "mean_p": 0.493587, "boundary_entropy": 0.690016, "bpb": 3.60653}, "el": {"chunks": 19414.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.945963, "cps": 151.671875, "mean_p": 0.486598, "boundary_entropy": 0.687374, "bpb": 4.326317}, "hi": {"chunks": 18313.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.08844, "cps": 143.070312, "mean_p": 0.410623, "boundary_entropy": 0.623741, "bpb": 3.375399}, "ta": {"chunks": 7120.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.38085, "cps": 55.625, "mean_p": 0.405144, "boundary_entropy": 0.639104, "bpb": 3.133301}, "th": {"chunks": 6740.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.419024, "cps": 52.65625, "mean_p": 0.434565, "boundary_entropy": 0.66568, "bpb": 3.039006}, "zh": {"chunks": 4116.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.696918, "cps": 32.15625, "mean_p": 0.471027, "boundary_entropy": 0.680577, "bpb": 6.127289}, "my": {"chunks": 24575.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.296971, "cps": 191.992188, "mean_p": 0.496224, "boundary_entropy": 0.691021, "bpb": 3.557298}}, "bpb": {"en": 4.44478, "de": 4.66379, "vi": 4.82084, "tr": 4.91265, "fi": 4.67747, "ru": 3.60653, "el": 4.32632, "hi": 3.3754, "ta": 3.1333, "th": 3.03901, "zh": 6.12729, "my": 3.5573}, "gini_cpc": 0.278152, "gini_cps": 0.306942, "bpic": 3.066862, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9007458553244991, "vi": 1.1690136940749254, "tr": 1.5190094575354869, "fi": 1.268108200705543, "ru": 1.7083565392891735, "el": 2.9769611635831508, "hi": 3.4253378330812083, "ta": 1.1985427053776068, "th": 1.318675144083674, "zh": 2.1932135095397514, "my": 4.0815877641329985}, "t": 28.0} |
| {"event": "train", "step": 150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.81738, "bpb": 4.06462, "chunk_loss": 0.037979, "F": 0.30212, "G": 0.43839, "chunks": 309.38, "bytes_per_chunk": 3.31, "boundary_entropy": 0.66171, "router_noise": 0.0, "grad_norm": 1.8042, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 30.0} |
| {"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4479.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.257369, "cps": 34.992188, "mean_p": 0.417313, "boundary_entropy": 0.647322, "bpb": 3.814088}, "de": {"chunks": 5729.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.279014, "cps": 44.757812, "mean_p": 0.411453, "boundary_entropy": 0.64362, "bpb": 4.063855}, "vi": {"chunks": 8354.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.45316, "cps": 65.265625, "mean_p": 0.455047, "boundary_entropy": 0.66698, "bpb": 3.577099}, "tr": {"chunks": 7522.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.415511, "cps": 58.765625, "mean_p": 0.446169, "boundary_entropy": 0.655753, "bpb": 4.109204}, "fi": {"chunks": 6592.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.350433, "cps": 51.5, "mean_p": 0.442011, "boundary_entropy": 0.656661, "bpb": 4.107896}, "ru": {"chunks": 5435.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.286807, "cps": 42.460938, "mean_p": 0.461048, "boundary_entropy": 0.684459, "bpb": 2.80486}, "el": {"chunks": 16769.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.817083, "cps": 131.007812, "mean_p": 0.482645, "boundary_entropy": 0.686545, "bpb": 3.085979}, "hi": {"chunks": 4453.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.264666, "cps": 34.789062, "mean_p": 0.376185, "boundary_entropy": 0.597354, "bpb": 2.374269}, "ta": {"chunks": 2283.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.122118, "cps": 17.835938, "mean_p": 0.359244, "boundary_entropy": 0.609031, "bpb": 2.212962}, "th": {"chunks": 13240.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.823127, "cps": 103.4375, "mean_p": 0.420789, "boundary_entropy": 0.650191, "bpb": 2.266522}, "zh": {"chunks": 5481.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.928039, "cps": 42.820312, "mean_p": 0.395703, "boundary_entropy": 0.601541, "bpb": 4.819294}, "my": {"chunks": 25952.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.369643, "cps": 202.75, "mean_p": 0.489611, "boundary_entropy": 0.689992, "bpb": 2.38434}}, "bpb": {"en": 3.81409, "de": 4.06385, "vi": 3.5771, "tr": 4.1092, "fi": 4.1079, "ru": 2.80486, "el": 3.08598, "hi": 2.37427, "ta": 2.21296, "th": 2.26652, "zh": 4.81929, "my": 2.38434}, "gini_cpc": 0.352372, "gini_cps": 0.356035, "bpic": 3.610731, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0841003205628277, "vi": 1.7607365790377698, "tr": 1.6144545950298024, "fi": 1.361596332652693, "ru": 1.1143802123188777, "el": 3.174748870290399, "hi": 1.0283493549391962, "ta": 0.47448610588380075, "th": 3.198232097940841, "zh": 3.605864618473522, "my": 5.3217015412576965}, "t": 35.0} |
| {"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.8189, "de": 4.06401, "vi": 3.58066, "tr": 4.11731, "fi": 4.11129, "ru": 2.78973, "el": 3.07972, "hi": 2.36938, "ta": 2.22937, "th": 2.30062, "zh": 4.82792, "my": 2.40172}, "bytes_per_chunk": {"en": 3.701, "de": 3.58, "vi": 2.863, "tr": 2.517, "fi": 2.825, "ru": 6.013, "el": 2.21, "hi": 13.092, "ta": 21.088, "th": 3.589, "zh": 2.61, "my": 2.105}, "bpb_mean": 3.30755, "frozen_inventory": false, "t": 37.1} |
| {"event": "train", "step": 200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.43812, "bpb": 3.51746, "chunk_loss": 0.036487, "F": 0.27832, "G": 0.42691, "chunks": 285.0, "bytes_per_chunk": 3.593, "boundary_entropy": 0.6533, "router_noise": 0.0, "grad_norm": 1.6249, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 37.3} |
| {"event": "train", "step": 250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.51534, "bpb": 3.62887, "chunk_loss": 0.034526, "F": 0.25162, "G": 0.42335, "chunks": 257.66, "bytes_per_chunk": 3.974, "boundary_entropy": 0.6516, "router_noise": 0.0, "grad_norm": 1.6205, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 40.9} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4716.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.270988, "cps": 36.84375, "mean_p": 0.434672, "boundary_entropy": 0.65308, "bpb": 3.579124}, "de": {"chunks": 5501.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.26791, "cps": 42.976562, "mean_p": 0.425502, "boundary_entropy": 0.652002, "bpb": 3.875477}, "vi": {"chunks": 7735.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.419582, "cps": 60.429688, "mean_p": 0.4615, "boundary_entropy": 0.669673, "bpb": 3.170073}, "tr": {"chunks": 7281.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.402199, "cps": 56.882812, "mean_p": 0.444318, "boundary_entropy": 0.658825, "bpb": 3.802974}, "fi": {"chunks": 6606.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.351178, "cps": 51.609375, "mean_p": 0.445651, "boundary_entropy": 0.658599, "bpb": 3.873628}, "ru": {"chunks": 5269.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.278047, "cps": 41.164062, "mean_p": 0.479837, "boundary_entropy": 0.688192, "bpb": 2.650679}, "el": {"chunks": 5103.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.248648, "cps": 39.867188, "mean_p": 0.468478, "boundary_entropy": 0.686487, "bpb": 2.74322}, "hi": {"chunks": 10387.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.617355, "cps": 81.148438, "mean_p": 0.377057, "boundary_entropy": 0.596336, "bpb": 2.097735}, "ta": {"chunks": 3056.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.163466, "cps": 23.875, "mean_p": 0.364039, "boundary_entropy": 0.597965, "bpb": 1.905938}, "th": {"chunks": 638.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.039664, "cps": 4.984375, "mean_p": 0.385898, "boundary_entropy": 0.629353, "bpb": 2.053419}, "zh": {"chunks": 3774.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.639011, "cps": 29.484375, "mean_p": 0.393571, "boundary_entropy": 0.608301, "bpb": 4.453416}, "my": {"chunks": 5623.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.29676, "cps": 43.929688, "mean_p": 0.48318, "boundary_entropy": 0.690051, "bpb": 2.165924}}, "bpb": {"en": 3.57912, "de": 3.87548, "vi": 3.17007, "tr": 3.80297, "fi": 3.87363, "ru": 2.65068, "el": 2.74322, "hi": 2.09773, "ta": 1.90594, "th": 2.05342, "zh": 4.45342, "my": 2.16592}, "gini_cpc": 0.267119, "gini_cps": 0.234871, "bpic": 5.842394, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9886431498776438, "vi": 1.5483441581072621, "tr": 1.48419445050017, "fi": 1.2959164708321953, "ru": 1.0260518506582879, "el": 0.9175612132059617, "hi": 2.278166086302266, "ta": 0.603223433351912, "th": 0.1463692802470337, "zh": 2.358081314641857, "my": 1.0951031576442505}, "t": 42.4} |
| {"event": "train", "step": 300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.29146, "bpb": 3.30588, "chunk_loss": 0.038404, "F": 0.30933, "G": 0.43146, "chunks": 316.75, "bytes_per_chunk": 3.233, "boundary_entropy": 0.65187, "router_noise": 0.0, "grad_norm": 1.813, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 45.9} |
| {"event": "train", "step": 350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.02209, "bpb": 2.91725, "chunk_loss": 0.035541, "F": 0.25952, "G": 0.42289, "chunks": 265.75, "bytes_per_chunk": 3.853, "boundary_entropy": 0.64788, "router_noise": 0.0, "grad_norm": 0.783, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 49.3} |
| {"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4999.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.287249, "cps": 39.054688, "mean_p": 0.444517, "boundary_entropy": 0.661369, "bpb": 3.211374}, "de": {"chunks": 5533.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.269469, "cps": 43.226562, "mean_p": 0.439478, "boundary_entropy": 0.663179, "bpb": 3.501841}, "vi": {"chunks": 7930.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.43016, "cps": 61.953125, "mean_p": 0.451185, "boundary_entropy": 0.66859, "bpb": 2.777583}, "tr": {"chunks": 7381.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.407722, "cps": 57.664062, "mean_p": 0.455534, "boundary_entropy": 0.662883, "bpb": 3.410275}, "fi": {"chunks": 5010.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.266334, "cps": 39.140625, "mean_p": 0.431925, "boundary_entropy": 0.657674, "bpb": 3.559692}, "ru": {"chunks": 6179.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.326069, "cps": 48.273438, "mean_p": 0.414667, "boundary_entropy": 0.643745, "bpb": 2.354941}, "el": {"chunks": 7248.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.353165, "cps": 56.625, "mean_p": 0.463574, "boundary_entropy": 0.678233, "bpb": 2.546753}, "hi": {"chunks": 11702.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.695513, "cps": 91.421875, "mean_p": 0.374935, "boundary_entropy": 0.609866, "bpb": 1.858502}, "ta": {"chunks": 6754.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.361273, "cps": 52.765625, "mean_p": 0.35345, "boundary_entropy": 0.613205, "bpb": 1.608868}, "th": {"chunks": 2273.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.141312, "cps": 17.757812, "mean_p": 0.384595, "boundary_entropy": 0.63282, "bpb": 1.927083}, "zh": {"chunks": 3176.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.537758, "cps": 24.8125, "mean_p": 0.407667, "boundary_entropy": 0.637798, "bpb": 4.132382}, "my": {"chunks": 30290.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.598586, "cps": 236.640625, "mean_p": 0.475736, "boundary_entropy": 0.680466, "bpb": 1.921484}}, "bpb": {"en": 3.21137, "de": 3.50184, "vi": 2.77758, "tr": 3.41028, "fi": 3.55969, "ru": 2.35494, "el": 2.54675, "hi": 1.8585, "ta": 1.60887, "th": 1.92708, "zh": 4.13238, "my": 1.92148}, "gini_cpc": 0.333918, "gini_cps": 0.35797, "bpic": 3.897243, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9381002387590944, "vi": 1.4975144744164455, "tr": 1.4194027337755017, "fi": 0.9271859156265891, "ru": 1.135141402950775, "el": 1.2294710952130494, "hi": 2.4212855171331444, "ta": 1.2576985790311312, "th": 0.4919481751494843, "zh": 1.872095651775123, "my": 5.565150078780797}, "t": 54.0} |
| {"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 3.22092, "de": 3.49882, "vi": 2.78313, "tr": 3.41757, "fi": 3.55034, "ru": 2.33822, "el": 2.54053, "hi": 1.85514, "ta": 1.62428, "th": 1.95495, "zh": 4.13895, "my": 1.93749}, "bytes_per_chunk": {"en": 3.292, "de": 3.638, "vi": 3.29, "tr": 2.587, "fi": 3.9, "ru": 6.255, "el": 5.136, "hi": 3.164, "ta": 7.629, "th": 14.75, "zh": 4.301, "my": 1.789}, "bpb_mean": 2.73836, "frozen_inventory": false, "t": 55.3} |
| {"event": "train", "step": 400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.02185, "bpb": 2.91691, "chunk_loss": 0.034792, "F": 0.2475, "G": 0.42567, "chunks": 253.44, "bytes_per_chunk": 4.04, "boundary_entropy": 0.65418, "router_noise": 0.0, "grad_norm": 0.883, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.124, "alpha": 0.03, "t": 55.6} |
| {"event": "train", "step": 450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.11455, "bpb": 3.05066, "chunk_loss": 0.033198, "F": 0.21478, "G": 0.41642, "chunks": 219.94, "bytes_per_chunk": 4.656, "boundary_entropy": 0.65257, "router_noise": 0.0, "grad_norm": 0.7094, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.067, "alpha": 0.03, "t": 58.9} |
| {"event": "train", "step": 500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.95315, "bpb": 2.8178, "chunk_loss": 0.036208, "F": 0.27435, "G": 0.43045, "chunks": 280.94, "bytes_per_chunk": 3.645, "boundary_entropy": 0.65838, "router_noise": 0.0, "grad_norm": 1.0143, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.064, "alpha": 0.03, "t": 62.1} |
| {"event": "train", "step": 550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.90859, "bpb": 2.75352, "chunk_loss": 0.035791, "F": 0.2627, "G": 0.43176, "chunks": 269.0, "bytes_per_chunk": 3.807, "boundary_entropy": 0.66018, "router_noise": 0.0, "grad_norm": 0.8762, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.068, "alpha": 0.03, "t": 65.5} |
| {"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5342.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.306959, "cps": 41.734375, "mean_p": 0.448559, "boundary_entropy": 0.663732, "bpb": 2.935401}, "de": {"chunks": 5515.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.268592, "cps": 43.085938, "mean_p": 0.442437, "boundary_entropy": 0.666249, "bpb": 3.245343}, "vi": {"chunks": 5723.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.310442, "cps": 44.710938, "mean_p": 0.449227, "boundary_entropy": 0.669796, "bpb": 2.573474}, "tr": {"chunks": 6706.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.370436, "cps": 52.390625, "mean_p": 0.454221, "boundary_entropy": 0.663418, "bpb": 3.14082}, "fi": {"chunks": 4663.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247887, "cps": 36.429688, "mean_p": 0.439457, "boundary_entropy": 0.663441, "bpb": 3.370109}, "ru": {"chunks": 7066.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.372876, "cps": 55.203125, "mean_p": 0.431, "boundary_entropy": 0.659563, "bpb": 2.149329}, "el": {"chunks": 9744.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.474784, "cps": 76.125, "mean_p": 0.442468, "boundary_entropy": 0.661458, "bpb": 2.242056}, "hi": {"chunks": 6562.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.390015, "cps": 51.265625, "mean_p": 0.405137, "boundary_entropy": 0.655118, "bpb": 1.643095}, "ta": {"chunks": 4122.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.220487, "cps": 32.203125, "mean_p": 0.376331, "boundary_entropy": 0.636342, "bpb": 1.414553}, "th": {"chunks": 11719.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.728567, "cps": 91.554688, "mean_p": 0.395637, "boundary_entropy": 0.647184, "bpb": 1.73937}, "zh": {"chunks": 5034.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.852354, "cps": 39.328125, "mean_p": 0.432255, "boundary_entropy": 0.652789, "bpb": 3.940008}, "my": {"chunks": 21198.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.118746, "cps": 165.609375, "mean_p": 0.456536, "boundary_entropy": 0.676561, "bpb": 1.705429}}, "bpb": {"en": 2.9354, "de": 3.24534, "vi": 2.57347, "tr": 3.14082, "fi": 3.37011, "ru": 2.14933, "el": 2.24206, "hi": 1.6431, "ta": 1.41455, "th": 1.73937, "zh": 3.94001, "my": 1.70543}, "gini_cpc": 0.288777, "gini_cps": 0.265311, "bpic": 4.109268, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8750106644082706, "vi": 1.0113485135068856, "tr": 1.2067942554184672, "fi": 0.8075580831937533, "ru": 1.214743699799172, "el": 1.5467376841385856, "hi": 1.2705781707993455, "ta": 0.7182948529714369, "th": 2.373502675032153, "zh": 2.7767706168637494, "my": 3.644615755413919}, "t": 70.1} |
| {"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.94407, "de": 3.24266, "vi": 2.57859, "tr": 3.15126, "fi": 3.36131, "ru": 2.13021, "el": 2.23466, "hi": 1.63956, "ta": 1.42698, "th": 1.76968, "zh": 3.95151, "my": 1.71991}, "bytes_per_chunk": {"en": 3.202, "de": 3.679, "vi": 4.441, "tr": 2.808, "fi": 4.231, "ru": 5.458, "el": 3.916, "hi": 5.26, "ta": 12.003, "th": 3.785, "zh": 2.653, "my": 2.541}, "bpb_mean": 2.51253, "frozen_inventory": false, "t": 71.4} |
| {"event": "train", "step": 600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.93578, "bpb": 2.79274, "chunk_loss": 0.036523, "F": 0.27774, "G": 0.43436, "chunks": 284.41, "bytes_per_chunk": 3.6, "boundary_entropy": 0.66325, "router_noise": 0.0, "grad_norm": 0.8678, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.123, "alpha": 0.03, "t": 71.6} |
| {"event": "train", "step": 650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.71934, "bpb": 2.48048, "chunk_loss": 0.03353, "F": 0.22409, "G": 0.42925, "chunks": 229.47, "bytes_per_chunk": 4.462, "boundary_entropy": 0.66313, "router_noise": 0.0, "grad_norm": 0.8415, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.066, "alpha": 0.03, "t": 74.9} |
| {"event": "train", "step": 700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.75718, "bpb": 2.53508, "chunk_loss": 0.033331, "F": 0.21765, "G": 0.43048, "chunks": 222.88, "bytes_per_chunk": 4.595, "boundary_entropy": 0.66462, "router_noise": 0.0, "grad_norm": 0.7186, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.061, "alpha": 0.03, "t": 77.9} |
| {"event": "train", "step": 750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.68812, "bpb": 2.43544, "chunk_loss": 0.03676, "F": 0.27753, "G": 0.43999, "chunks": 284.19, "bytes_per_chunk": 3.603, "boundary_entropy": 0.66715, "router_noise": 0.0, "grad_norm": 0.7038, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.063, "alpha": 0.03, "t": 81.1} |
| {"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4912.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.28225, "cps": 38.375, "mean_p": 0.443501, "boundary_entropy": 0.664394, "bpb": 2.734705}, "de": {"chunks": 5567.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.271125, "cps": 43.492188, "mean_p": 0.44528, "boundary_entropy": 0.668448, "bpb": 3.061101}, "vi": {"chunks": 6530.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.354218, "cps": 51.015625, "mean_p": 0.455871, "boundary_entropy": 0.672692, "bpb": 2.433895}, "tr": {"chunks": 5736.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.316854, "cps": 44.8125, "mean_p": 0.451551, "boundary_entropy": 0.664523, "bpb": 3.006814}, "fi": {"chunks": 5736.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.304928, "cps": 44.8125, "mean_p": 0.44873, "boundary_entropy": 0.666217, "bpb": 3.22702}, "ru": {"chunks": 8079.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.426332, "cps": 63.117188, "mean_p": 0.430969, "boundary_entropy": 0.663802, "bpb": 2.000283}, "el": {"chunks": 5440.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.265068, "cps": 42.5, "mean_p": 0.435507, "boundary_entropy": 0.661488, "bpb": 2.154549}, "hi": {"chunks": 10057.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.597741, "cps": 78.570312, "mean_p": 0.428131, "boundary_entropy": 0.666681, "bpb": 1.562586}, "ta": {"chunks": 8468.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.452955, "cps": 66.15625, "mean_p": 0.401792, "boundary_entropy": 0.653966, "bpb": 1.314943}, "th": {"chunks": 9247.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.574883, "cps": 72.242188, "mean_p": 0.43373, "boundary_entropy": 0.673503, "bpb": 1.624486}, "zh": {"chunks": 3459.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.585676, "cps": 27.023438, "mean_p": 0.428026, "boundary_entropy": 0.655186, "bpb": 3.838311}, "my": {"chunks": 17982.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.949018, "cps": 140.484375, "mean_p": 0.455389, "boundary_entropy": 0.679239, "bpb": 1.502008}}, "bpb": {"en": 2.7347, "de": 3.0611, "vi": 2.4339, "tr": 3.00681, "fi": 3.22702, "ru": 2.00028, "el": 2.15455, "hi": 1.56259, "ta": 1.31494, "th": 1.62449, "zh": 3.83831, "my": 1.50201}, "gini_cpc": 0.223271, "gini_cps": 0.233003, "bpic": 4.207525, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9605822917775484, "vi": 1.254977100632648, "tr": 1.1225982304298907, "fi": 1.0803463806003037, "ru": 1.5104771363866851, "el": 0.9391258968844202, "hi": 2.1177716941498756, "ta": 1.6048008359809838, "th": 2.036786718866283, "zh": 2.075022840185711, "my": 3.3623303388529826}, "t": 85.4} |
| {"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.74614, "de": 3.06157, "vi": 2.44134, "tr": 3.01885, "fi": 3.22336, "ru": 1.98284, "el": 2.14206, "hi": 1.55922, "ta": 1.32631, "th": 1.6518, "zh": 3.85075, "my": 1.51365}, "bytes_per_chunk": {"en": 3.322, "de": 3.703, "vi": 3.805, "tr": 3.441, "fi": 3.293, "ru": 4.252, "el": 6.646, "hi": 4.058, "ta": 5.821, "th": 4.73, "zh": 3.759, "my": 3.027}, "bpb_mean": 2.37649, "frozen_inventory": false, "t": 86.7} |
| {"event": "train", "step": 800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.82564, "bpb": 2.63385, "chunk_loss": 0.035836, "F": 0.26526, "G": 0.43888, "chunks": 271.62, "bytes_per_chunk": 3.77, "boundary_entropy": 0.66809, "router_noise": 0.0, "grad_norm": 0.8474, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.116, "alpha": 0.03, "t": 86.9} |
| {"event": "train", "step": 850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.85993, "bpb": 2.68332, "chunk_loss": 0.037305, "F": 0.28809, "G": 0.44249, "chunks": 295.0, "bytes_per_chunk": 3.471, "boundary_entropy": 0.6679, "router_noise": 0.0, "grad_norm": 0.8298, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.061, "alpha": 0.03, "t": 89.9} |
| {"event": "train", "step": 900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.67998, "bpb": 2.4237, "chunk_loss": 0.034251, "F": 0.23672, "G": 0.43518, "chunks": 242.41, "bytes_per_chunk": 4.224, "boundary_entropy": 0.66695, "router_noise": 0.0, "grad_norm": 0.9456, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.062, "alpha": 0.03, "t": 93.0} |
| {"event": "train", "step": 950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.81599, "bpb": 2.61993, "chunk_loss": 0.03406, "F": 0.23508, "G": 0.43655, "chunks": 240.72, "bytes_per_chunk": 4.254, "boundary_entropy": 0.66731, "router_noise": 0.0, "grad_norm": 0.7922, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.063, "alpha": 0.03, "t": 96.2} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4898.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.281446, "cps": 38.265625, "mean_p": 0.447127, "boundary_entropy": 0.66629, "bpb": 2.59666}, "de": {"chunks": 5502.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.267959, "cps": 42.984375, "mean_p": 0.449134, "boundary_entropy": 0.670512, "bpb": 2.926958}, "vi": {"chunks": 5545.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.300787, "cps": 43.320312, "mean_p": 0.446436, "boundary_entropy": 0.670386, "bpb": 2.317883}, "tr": {"chunks": 6476.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.357731, "cps": 50.59375, "mean_p": 0.454706, "boundary_entropy": 0.664367, "bpb": 2.900437}, "fi": {"chunks": 5326.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.283132, "cps": 41.609375, "mean_p": 0.450869, "boundary_entropy": 0.668667, "bpb": 3.152478}, "ru": {"chunks": 7923.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.4181, "cps": 61.898438, "mean_p": 0.429978, "boundary_entropy": 0.665329, "bpb": 1.93473}, "el": {"chunks": 8837.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.43059, "cps": 69.039062, "mean_p": 0.457557, "boundary_entropy": 0.670866, "bpb": 2.069153}, "hi": {"chunks": 3446.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.204814, "cps": 26.921875, "mean_p": 0.409333, "boundary_entropy": 0.659643, "bpb": 1.534182}, "ta": {"chunks": 5149.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.275421, "cps": 40.226562, "mean_p": 0.394597, "boundary_entropy": 0.65104, "bpb": 1.293053}, "th": {"chunks": 7314.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.454709, "cps": 57.140625, "mean_p": 0.426183, "boundary_entropy": 0.671179, "bpb": 1.578289}, "zh": {"chunks": 2765.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.468168, "cps": 21.601562, "mean_p": 0.425336, "boundary_entropy": 0.657916, "bpb": 3.682777}, "my": {"chunks": 20436.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.078531, "cps": 159.65625, "mean_p": 0.459181, "boundary_entropy": 0.679084, "bpb": 1.458518}}, "bpb": {"en": 2.59666, "de": 2.92696, "vi": 2.31788, "tr": 2.90044, "fi": 3.15248, "ru": 1.93473, "el": 2.06915, "hi": 1.53418, "ta": 1.29305, "th": 1.57829, "zh": 3.68278, "my": 1.45852}, "gini_cpc": 0.239823, "gini_cps": 0.277449, "bpic": 4.589748, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9520801668606365, "vi": 1.0687195351484358, "tr": 1.2710470530829134, "fi": 1.0059922107988777, "ru": 1.4855448942059168, "el": 1.5299222095967502, "hi": 0.7277220589778889, "ta": 0.9785944801088502, "th": 1.6156200351008887, "zh": 1.6634395857684368, "my": 3.8321090331424776}, "t": 100.4} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.60288, "de": 2.92927, "vi": 2.32367, "tr": 2.91301, "fi": 3.15302, "ru": 1.91718, "el": 2.05741, "hi": 1.52946, "ta": 1.30339, "th": 1.60204, "zh": 3.69385, "my": 1.46846}, "bytes_per_chunk": {"en": 3.451, "de": 3.75, "vi": 4.608, "tr": 3.017, "fi": 3.875, "ru": 4.171, "el": 4.181, "hi": 11.53, "ta": 9.877, "th": 5.791, "zh": 4.775, "my": 2.634}, "bpb_mean": 2.29114, "frozen_inventory": false, "t": 101.7} |
| {"event": "train", "step": 1000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.75532, "bpb": 2.5324, "chunk_loss": 0.034231, "F": 0.23672, "G": 0.43658, "chunks": 242.41, "bytes_per_chunk": 4.224, "boundary_entropy": 0.66781, "router_noise": 0.0, "grad_norm": 0.8238, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.117, "alpha": 0.03, "t": 102.1} |
| {"event": "train", "step": 1050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.61805, "bpb": 2.33435, "chunk_loss": 0.03315, "F": 0.2193, "G": 0.43693, "chunks": 224.56, "bytes_per_chunk": 4.56, "boundary_entropy": 0.66905, "router_noise": 0.0, "grad_norm": 0.8444, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.058, "alpha": 0.03, "t": 105.0} |
| {"event": "train", "step": 1100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.81472, "bpb": 2.61809, "chunk_loss": 0.034559, "F": 0.24347, "G": 0.43919, "chunks": 249.31, "bytes_per_chunk": 4.107, "boundary_entropy": 0.66939, "router_noise": 0.0, "grad_norm": 0.7697, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.061, "alpha": 0.03, "t": 108.0} |
| {"event": "train", "step": 1150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.6674, "bpb": 2.40555, "chunk_loss": 0.03533, "F": 0.25516, "G": 0.43978, "chunks": 261.28, "bytes_per_chunk": 3.919, "boundary_entropy": 0.66906, "router_noise": 0.0, "grad_norm": 0.7494, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 111.0} |
| {"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4504.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.258806, "cps": 35.1875, "mean_p": 0.442802, "boundary_entropy": 0.664973, "bpb": 2.493377}, "de": {"chunks": 5218.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.254128, "cps": 40.765625, "mean_p": 0.446477, "boundary_entropy": 0.669969, "bpb": 2.817789}, "vi": {"chunks": 6642.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.360293, "cps": 51.890625, "mean_p": 0.458607, "boundary_entropy": 0.674067, "bpb": 2.283653}, "tr": {"chunks": 6092.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.336519, "cps": 47.59375, "mean_p": 0.454199, "boundary_entropy": 0.666617, "bpb": 2.794133}, "fi": {"chunks": 5046.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.268247, "cps": 39.421875, "mean_p": 0.448928, "boundary_entropy": 0.669061, "bpb": 3.063499}, "ru": {"chunks": 5282.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.278734, "cps": 41.265625, "mean_p": 0.42378, "boundary_entropy": 0.664616, "bpb": 1.872729}, "el": {"chunks": 6356.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.309701, "cps": 49.65625, "mean_p": 0.444712, "boundary_entropy": 0.668476, "bpb": 1.998763}, "hi": {"chunks": 3267.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.194175, "cps": 25.523438, "mean_p": 0.405743, "boundary_entropy": 0.657368, "bpb": 1.519966}, "ta": {"chunks": 7167.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.383365, "cps": 55.992188, "mean_p": 0.397357, "boundary_entropy": 0.651812, "bpb": 1.269915}, "th": {"chunks": 7645.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.475288, "cps": 59.726562, "mean_p": 0.427087, "boundary_entropy": 0.670701, "bpb": 1.503785}, "zh": {"chunks": 3206.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.542838, "cps": 25.046875, "mean_p": 0.432117, "boundary_entropy": 0.661607, "bpb": 3.622764}, "my": {"chunks": 17518.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.92453, "cps": 136.859375, "mean_p": 0.451563, "boundary_entropy": 0.679219, "bpb": 1.411601}}, "bpb": {"en": 2.49338, "de": 2.81779, "vi": 2.28365, "tr": 2.79413, "fi": 3.0635, "ru": 1.87273, "el": 1.99876, "hi": 1.51997, "ta": 1.26991, "th": 1.50378, "zh": 3.62276, "my": 1.4116}, "gini_cpc": 0.236099, "gini_cps": 0.243425, "bpic": 4.923867, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9819229392809512, "vi": 1.3921353697716665, "tr": 1.3002746078317, "fi": 1.036480417102234, "ru": 1.0769980597721402, "el": 1.1966545094448033, "hi": 0.7502738328683897, "ta": 1.4812817538907497, "th": 1.8364629154287473, "zh": 2.0974702698703607, "my": 3.57229145129739}, "t": 114.9} |
| {"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.50432, "de": 2.82235, "vi": 2.29207, "tr": 2.8091, "fi": 3.06558, "ru": 1.85492, "el": 1.98817, "hi": 1.51832, "ta": 1.27799, "th": 1.52649, "zh": 3.63369, "my": 1.4207}, "bytes_per_chunk": {"en": 3.78, "de": 4.0, "vi": 3.826, "tr": 3.326, "fi": 4.133, "ru": 6.161, "el": 5.428, "hi": 13.318, "ta": 6.946, "th": 5.668, "zh": 4.334, "my": 3.053}, "bpb_mean": 2.22614, "frozen_inventory": false, "t": 116.2} |
| {"event": "train", "step": 1200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.72703, "bpb": 2.49158, "chunk_loss": 0.034196, "F": 0.23666, "G": 0.43766, "chunks": 242.34, "bytes_per_chunk": 4.225, "boundary_entropy": 0.6682, "router_noise": 0.0, "grad_norm": 0.8037, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 116.4} |
| {"event": "train", "step": 1250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.84229, "bpb": 2.65787, "chunk_loss": 0.033975, "F": 0.23264, "G": 0.43906, "chunks": 238.22, "bytes_per_chunk": 4.299, "boundary_entropy": 0.6693, "router_noise": 0.0, "grad_norm": 0.8192, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.06, "alpha": 0.03, "t": 119.4} |
| {"event": "train", "step": 1300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.52659, "bpb": 2.2024, "chunk_loss": 0.034434, "F": 0.24011, "G": 0.43647, "chunks": 245.88, "bytes_per_chunk": 4.165, "boundary_entropy": 0.66875, "router_noise": 0.0, "grad_norm": 0.7942, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 122.4} |
| {"event": "train", "step": 1350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.52559, "bpb": 2.20097, "chunk_loss": 0.033907, "F": 0.23175, "G": 0.43484, "chunks": 237.31, "bytes_per_chunk": 4.315, "boundary_entropy": 0.66843, "router_noise": 0.0, "grad_norm": 0.7966, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.06, "alpha": 0.03, "t": 125.4} |
| {"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4547.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.261277, "cps": 35.523438, "mean_p": 0.446103, "boundary_entropy": 0.666049, "bpb": 2.436432}, "de": {"chunks": 5182.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.252374, "cps": 40.484375, "mean_p": 0.446924, "boundary_entropy": 0.669713, "bpb": 2.754651}, "vi": {"chunks": 5104.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276865, "cps": 39.875, "mean_p": 0.451266, "boundary_entropy": 0.673161, "bpb": 2.213969}, "tr": {"chunks": 5591.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.308844, "cps": 43.679688, "mean_p": 0.452948, "boundary_entropy": 0.664937, "bpb": 2.727622}, "fi": {"chunks": 5420.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.288129, "cps": 42.34375, "mean_p": 0.452553, "boundary_entropy": 0.66911, "bpb": 3.001984}, "ru": {"chunks": 7295.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.38496, "cps": 56.992188, "mean_p": 0.435864, "boundary_entropy": 0.669164, "bpb": 1.804946}, "el": {"chunks": 8140.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.396628, "cps": 63.59375, "mean_p": 0.450104, "boundary_entropy": 0.669493, "bpb": 1.903281}, "hi": {"chunks": 6062.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.360297, "cps": 47.359375, "mean_p": 0.411036, "boundary_entropy": 0.661469, "bpb": 1.474969}, "ta": {"chunks": 8159.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.436427, "cps": 63.742188, "mean_p": 0.401775, "boundary_entropy": 0.65416, "bpb": 1.191094}, "th": {"chunks": 4859.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.302083, "cps": 37.960938, "mean_p": 0.420074, "boundary_entropy": 0.669633, "bpb": 1.465386}, "zh": {"chunks": 3173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.53725, "cps": 24.789062, "mean_p": 0.430311, "boundary_entropy": 0.661683, "bpb": 3.585908}, "my": {"chunks": 20757.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.095472, "cps": 162.164062, "mean_p": 0.46368, "boundary_entropy": 0.682058, "bpb": 1.369665}}, "bpb": {"en": 2.43643, "de": 2.75465, "vi": 2.21397, "tr": 2.72762, "fi": 3.00198, "ru": 1.80495, "el": 1.90328, "hi": 1.47497, "ta": 1.19109, "th": 1.46539, "zh": 3.58591, "my": 1.36966}, "gini_cpc": 0.233259, "gini_cps": 0.259687, "bpic": 4.553156, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9659266923127207, "vi": 1.0596603631575678, "tr": 1.182056149851215, "fi": 1.1027741291473738, "ru": 1.4733816201699867, "el": 1.518038288849939, "hi": 1.3789865335830396, "ta": 1.67036210844237, "th": 1.156178794764645, "zh": 2.056249432592174, "my": 4.192763589575239}, "t": 129.6} |
| {"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.4446, "de": 2.7613, "vi": 2.21854, "tr": 2.73977, "fi": 2.99952, "ru": 1.78487, "el": 1.89323, "hi": 1.47132, "ta": 1.19984, "th": 1.48469, "zh": 3.59238, "my": 1.37814}, "bytes_per_chunk": {"en": 3.863, "de": 4.046, "vi": 4.874, "tr": 3.533, "fi": 3.597, "ru": 4.58, "el": 4.551, "hi": 7.797, "ta": 6.112, "th": 8.409, "zh": 4.6, "my": 2.599}, "bpb_mean": 2.16402, "frozen_inventory": false, "t": 130.8} |
| {"event": "train", "step": 1400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.58463, "bpb": 2.28613, "chunk_loss": 0.034167, "F": 0.23508, "G": 0.43912, "chunks": 240.72, "bytes_per_chunk": 4.254, "boundary_entropy": 0.66987, "router_noise": 0.0, "grad_norm": 0.7793, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 131.1} |
| {"event": "train", "step": 1450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.67271, "bpb": 2.41321, "chunk_loss": 0.034688, "F": 0.24573, "G": 0.43595, "chunks": 251.62, "bytes_per_chunk": 4.07, "boundary_entropy": 0.66769, "router_noise": 0.0, "grad_norm": 0.7581, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.063, "alpha": 0.03, "t": 134.3} |
| {"event": "train", "step": 1500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.74867, "bpb": 2.5228, "chunk_loss": 0.034744, "F": 0.24518, "G": 0.44103, "chunks": 251.06, "bytes_per_chunk": 4.079, "boundary_entropy": 0.6708, "router_noise": 0.0, "grad_norm": 0.7523, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.058, "alpha": 0.03, "t": 137.2} |
| {"event": "train", "step": 1550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.61279, "bpb": 2.32677, "chunk_loss": 0.036113, "F": 0.2674, "G": 0.44502, "chunks": 273.81, "bytes_per_chunk": 3.74, "boundary_entropy": 0.67048, "router_noise": 0.0, "grad_norm": 0.8451, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.062, "alpha": 0.03, "t": 140.3} |
| {"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4331.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.248865, "cps": 33.835938, "mean_p": 0.445124, "boundary_entropy": 0.665808, "bpb": 2.374294}, "de": {"chunks": 5362.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.261141, "cps": 41.890625, "mean_p": 0.450607, "boundary_entropy": 0.671435, "bpb": 2.673036}, "vi": {"chunks": 6416.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.348034, "cps": 50.125, "mean_p": 0.455665, "boundary_entropy": 0.6742, "bpb": 2.178122}, "tr": {"chunks": 5884.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.325029, "cps": 45.96875, "mean_p": 0.45446, "boundary_entropy": 0.66711, "bpb": 2.651155}, "fi": {"chunks": 6341.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.33709, "cps": 49.539062, "mean_p": 0.463326, "boundary_entropy": 0.67229, "bpb": 2.899522}, "ru": {"chunks": 7539.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.397836, "cps": 58.898438, "mean_p": 0.43901, "boundary_entropy": 0.670594, "bpb": 1.744723}, "el": {"chunks": 8447.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.411587, "cps": 65.992188, "mean_p": 0.447857, "boundary_entropy": 0.667537, "bpb": 1.866056}, "hi": {"chunks": 6689.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.397563, "cps": 52.257812, "mean_p": 0.414273, "boundary_entropy": 0.662363, "bpb": 1.443552}, "ta": {"chunks": 10330.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.552554, "cps": 80.703125, "mean_p": 0.407702, "boundary_entropy": 0.654899, "bpb": 1.180318}, "th": {"chunks": 10427.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.648244, "cps": 81.460938, "mean_p": 0.440708, "boundary_entropy": 0.674602, "bpb": 1.380284}, "zh": {"chunks": 3516.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.595327, "cps": 27.46875, "mean_p": 0.43172, "boundary_entropy": 0.662344, "bpb": 3.56054}, "my": {"chunks": 7306.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.385582, "cps": 57.078125, "mean_p": 0.461352, "boundary_entropy": 0.683981, "bpb": 1.329533}}, "bpb": {"en": 2.37429, "de": 2.67304, "vi": 2.17812, "tr": 2.65115, "fi": 2.89952, "ru": 1.74472, "el": 1.86606, "hi": 1.44355, "ta": 1.18032, "th": 1.38028, "zh": 3.56054, "my": 1.32953}, "gini_cpc": 0.161738, "gini_cps": 0.164521, "bpic": 4.646934, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0493257706818886, "vi": 1.3984828658756416, "tr": 1.3060447239661328, "fi": 1.3545087229858006, "ru": 1.5986024177418854, "el": 1.6538555899681424, "hi": 1.597504387149144, "ta": 2.220295549929605, "th": 2.604799169615757, "zh": 2.3921662577486758, "my": 1.5493596030444026}, "t": 144.1} |
| {"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.38299, "de": 2.68149, "vi": 2.18581, "tr": 2.66255, "fi": 2.90333, "ru": 1.72364, "el": 1.85648, "hi": 1.44171, "ta": 1.18841, "th": 1.39732, "zh": 3.56522, "my": 1.33773}, "bytes_per_chunk": {"en": 3.802, "de": 3.781, "vi": 3.8, "tr": 3.301, "fi": 3.045, "ru": 4.531, "el": 4.421, "hi": 6.29, "ta": 4.887, "th": 4.194, "zh": 4.43, "my": 7.418}, "bpb_mean": 2.11056, "frozen_inventory": false, "t": 145.4} |
| {"event": "train", "step": 1600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.60821, "bpb": 2.32016, "chunk_loss": 0.034738, "F": 0.24509, "G": 0.44297, "chunks": 250.97, "bytes_per_chunk": 4.08, "boundary_entropy": 0.67059, "router_noise": 0.0, "grad_norm": 0.8486, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 145.6} |
| {"event": "train", "step": 1650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.58309, "bpb": 2.28391, "chunk_loss": 0.034384, "F": 0.2388, "G": 0.439, "chunks": 244.53, "bytes_per_chunk": 4.188, "boundary_entropy": 0.66916, "router_noise": 0.0, "grad_norm": 0.8772, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 148.6} |
| {"event": "train", "step": 1700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.55575, "bpb": 2.24447, "chunk_loss": 0.035081, "F": 0.25113, "G": 0.44311, "chunks": 257.16, "bytes_per_chunk": 3.982, "boundary_entropy": 0.67187, "router_noise": 0.0, "grad_norm": 0.843, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 151.5} |
| {"event": "train", "step": 1750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.56839, "bpb": 2.26271, "chunk_loss": 0.034633, "F": 0.24454, "G": 0.44265, "chunks": 250.41, "bytes_per_chunk": 4.089, "boundary_entropy": 0.67116, "router_noise": 0.0, "grad_norm": 0.7987, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 154.4} |
| {"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4515.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.259438, "cps": 35.273438, "mean_p": 0.445637, "boundary_entropy": 0.6659, "bpb": 2.335038}, "de": {"chunks": 5311.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.258657, "cps": 41.492188, "mean_p": 0.449817, "boundary_entropy": 0.671977, "bpb": 2.626674}, "vi": {"chunks": 8281.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.4492, "cps": 64.695312, "mean_p": 0.462677, "boundary_entropy": 0.674614, "bpb": 2.129862}, "tr": {"chunks": 6412.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.354195, "cps": 50.09375, "mean_p": 0.456991, "boundary_entropy": 0.667155, "bpb": 2.616529}, "fi": {"chunks": 5689.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.302429, "cps": 44.445312, "mean_p": 0.454797, "boundary_entropy": 0.669789, "bpb": 2.850127}, "ru": {"chunks": 7243.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.382216, "cps": 56.585938, "mean_p": 0.436627, "boundary_entropy": 0.669384, "bpb": 1.707771}, "el": {"chunks": 8533.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.415777, "cps": 66.664062, "mean_p": 0.453013, "boundary_entropy": 0.671183, "bpb": 1.81481}, "hi": {"chunks": 5374.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.319406, "cps": 41.984375, "mean_p": 0.415949, "boundary_entropy": 0.662797, "bpb": 1.406838}, "ta": {"chunks": 8065.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.431399, "cps": 63.007812, "mean_p": 0.406006, "boundary_entropy": 0.655909, "bpb": 1.140041}, "th": {"chunks": 7897.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.490954, "cps": 61.695312, "mean_p": 0.429294, "boundary_entropy": 0.670865, "bpb": 1.35215}, "zh": {"chunks": 3628.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.614291, "cps": 28.34375, "mean_p": 0.433588, "boundary_entropy": 0.662948, "bpb": 3.506362}, "my": {"chunks": 5920.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.312434, "cps": 46.25, "mean_p": 0.461313, "boundary_entropy": 0.68513, "bpb": 1.284834}}, "bpb": {"en": 2.33504, "de": 2.62667, "vi": 2.12986, "tr": 2.61653, "fi": 2.85013, "ru": 1.70777, "el": 1.81481, "hi": 1.40684, "ta": 1.14004, "th": 1.35215, "zh": 3.50636, "my": 1.28483}, "gini_cpc": 0.14405, "gini_cps": 0.135401, "bpic": 4.992728, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9969887546713961, "vi": 1.7314342662150486, "tr": 1.3652410174852172, "fi": 1.1657097149808613, "ru": 1.473247240947063, "el": 1.6026078512026778, "hi": 1.231144288725546, "ta": 1.6628201084199143, "th": 1.8923760297505878, "zh": 2.3677737488651105, "my": 1.2042722976048454}, "t": 158.5} |
| {"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.34489, "de": 2.63382, "vi": 2.13561, "tr": 2.62761, "fi": 2.85331, "ru": 1.68605, "el": 1.80565, "hi": 1.4043, "ta": 1.14845, "th": 1.37386, "zh": 3.51116, "my": 1.29435}, "bytes_per_chunk": {"en": 3.862, "de": 3.799, "vi": 2.825, "tr": 3.051, "fi": 3.454, "ru": 4.649, "el": 4.374, "hi": 8.633, "ta": 6.489, "th": 5.616, "zh": 4.669, "my": 9.335}, "bpb_mean": 2.06826, "frozen_inventory": false, "t": 159.8} |
| {"event": "train", "step": 1800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.53992, "bpb": 2.22164, "chunk_loss": 0.034868, "F": 0.24606, "G": 0.44424, "chunks": 251.97, "bytes_per_chunk": 4.064, "boundary_entropy": 0.67151, "router_noise": 0.0, "grad_norm": 0.764, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 160.0} |
| {"event": "train", "step": 1850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.58418, "bpb": 2.28549, "chunk_loss": 0.03582, "F": 0.26147, "G": 0.44739, "chunks": 267.75, "bytes_per_chunk": 3.824, "boundary_entropy": 0.67147, "router_noise": 0.0, "grad_norm": 0.7667, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.063, "alpha": 0.03, "t": 163.2} |
| {"event": "train", "step": 1900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.64255, "bpb": 2.36969, "chunk_loss": 0.033979, "F": 0.23294, "G": 0.43755, "chunks": 238.53, "bytes_per_chunk": 4.293, "boundary_entropy": 0.66831, "router_noise": 0.0, "grad_norm": 0.7514, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.061, "alpha": 0.03, "t": 166.3} |
| {"event": "train", "step": 1950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.48598, "bpb": 2.14382, "chunk_loss": 0.034041, "F": 0.23398, "G": 0.44259, "chunks": 239.59, "bytes_per_chunk": 4.274, "boundary_entropy": 0.67222, "router_noise": 0.0, "grad_norm": 0.7624, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.057, "alpha": 0.03, "t": 169.1} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4779.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.274608, "cps": 37.335938, "mean_p": 0.447284, "boundary_entropy": 0.665967, "bpb": 2.294401}, "de": {"chunks": 5392.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.262602, "cps": 42.125, "mean_p": 0.448899, "boundary_entropy": 0.671093, "bpb": 2.572732}, "vi": {"chunks": 6260.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.339571, "cps": 48.90625, "mean_p": 0.45576, "boundary_entropy": 0.67392, "bpb": 2.103589}, "tr": {"chunks": 6401.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.353588, "cps": 50.007812, "mean_p": 0.454385, "boundary_entropy": 0.665755, "bpb": 2.559772}, "fi": {"chunks": 6259.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.332731, "cps": 48.898438, "mean_p": 0.459726, "boundary_entropy": 0.670624, "bpb": 2.77655}, "ru": {"chunks": 6944.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.366438, "cps": 54.25, "mean_p": 0.436526, "boundary_entropy": 0.670151, "bpb": 1.672036}, "el": {"chunks": 8958.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.436486, "cps": 69.984375, "mean_p": 0.452452, "boundary_entropy": 0.66966, "bpb": 1.745312}, "hi": {"chunks": 5893.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.350253, "cps": 46.039062, "mean_p": 0.416739, "boundary_entropy": 0.663372, "bpb": 1.387996}, "ta": {"chunks": 6872.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.367585, "cps": 53.6875, "mean_p": 0.399218, "boundary_entropy": 0.652433, "bpb": 1.125159}, "th": {"chunks": 8169.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.507864, "cps": 63.820312, "mean_p": 0.425083, "boundary_entropy": 0.66907, "bpb": 1.317206}, "zh": {"chunks": 4042.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.684389, "cps": 31.578125, "mean_p": 0.438083, "boundary_entropy": 0.664629, "bpb": 3.463344}, "my": {"chunks": 8079.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.426377, "cps": 63.117188, "mean_p": 0.467643, "boundary_entropy": 0.685854, "bpb": 1.259994}}, "bpb": {"en": 2.2944, "de": 2.57273, "vi": 2.10359, "tr": 2.55977, "fi": 2.77655, "ru": 1.67204, "el": 1.74531, "hi": 1.388, "ta": 1.12516, "th": 1.31721, "zh": 3.46334, "my": 1.25999}, "gini_cpc": 0.141133, "gini_cps": 0.118122, "bpic": 4.917243, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9562788840000307, "vi": 1.2365687896480926, "tr": 1.2876100727791198, "fi": 1.2116582893175023, "ru": 1.3344047755102717, "el": 1.5894881797676266, "hi": 1.2754647421995748, "ta": 1.3385813535985114, "th": 1.8494173198292736, "zh": 2.492240540172262, "my": 1.5526777220001853}, "t": 173.0} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.30153, "de": 2.58162, "vi": 2.11043, "tr": 2.56661, "fi": 2.78423, "ru": 1.64821, "el": 1.73729, "hi": 1.38684, "ta": 1.13151, "th": 1.33462, "zh": 3.46775, "my": 1.26926}, "bytes_per_chunk": {"en": 3.515, "de": 3.793, "vi": 3.934, "tr": 2.975, "fi": 3.152, "ru": 4.977, "el": 4.227, "hi": 7.329, "ta": 7.363, "th": 5.482, "zh": 4.16, "my": 6.735}, "bpb_mean": 2.02666, "frozen_inventory": false, "t": 174.3} |
| {"event": "train", "step": 2000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.60555, "bpb": 2.31633, "chunk_loss": 0.035368, "F": 0.25519, "G": 0.44269, "chunks": 261.31, "bytes_per_chunk": 3.919, "boundary_entropy": 0.67078, "router_noise": 0.0, "grad_norm": 0.7857, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 174.5} |
| {"event": "train", "step": 2050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.62051, "bpb": 2.3379, "chunk_loss": 0.03406, "F": 0.2341, "G": 0.44059, "chunks": 239.72, "bytes_per_chunk": 4.272, "boundary_entropy": 0.67051, "router_noise": 0.0, "grad_norm": 0.788, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.06, "alpha": 0.03, "t": 177.5} |
| {"event": "train", "step": 2100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.55688, "bpb": 2.2461, "chunk_loss": 0.035256, "F": 0.25339, "G": 0.44193, "chunks": 259.47, "bytes_per_chunk": 3.947, "boundary_entropy": 0.67006, "router_noise": 0.0, "grad_norm": 0.7852, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.056, "alpha": 0.03, "t": 180.3} |
| {"event": "train", "step": 2150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.50949, "bpb": 2.17773, "chunk_loss": 0.034655, "F": 0.24438, "G": 0.44172, "chunks": 250.25, "bytes_per_chunk": 4.092, "boundary_entropy": 0.67024, "router_noise": 0.0, "grad_norm": 0.7903, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.057, "alpha": 0.03, "t": 183.2} |
| {"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4520.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.259725, "cps": 35.3125, "mean_p": 0.444155, "boundary_entropy": 0.664619, "bpb": 2.258084}, "de": {"chunks": 5357.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.260897, "cps": 41.851562, "mean_p": 0.449716, "boundary_entropy": 0.671383, "bpb": 2.528154}, "vi": {"chunks": 7271.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.394413, "cps": 56.804688, "mean_p": 0.460189, "boundary_entropy": 0.674648, "bpb": 2.091452}, "tr": {"chunks": 5584.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.308457, "cps": 43.625, "mean_p": 0.453914, "boundary_entropy": 0.667118, "bpb": 2.52765}, "fi": {"chunks": 5249.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.279039, "cps": 41.007812, "mean_p": 0.455025, "boundary_entropy": 0.670557, "bpb": 2.738697}, "ru": {"chunks": 7475.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.394459, "cps": 58.398438, "mean_p": 0.441055, "boundary_entropy": 0.671574, "bpb": 1.637845}, "el": {"chunks": 8344.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.406568, "cps": 65.1875, "mean_p": 0.449353, "boundary_entropy": 0.669303, "bpb": 1.718832}, "hi": {"chunks": 6540.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.388707, "cps": 51.09375, "mean_p": 0.425252, "boundary_entropy": 0.668718, "bpb": 1.351044}, "ta": {"chunks": 9564.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.511581, "cps": 74.71875, "mean_p": 0.41003, "boundary_entropy": 0.656511, "bpb": 1.097082}, "th": {"chunks": 8987.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.558719, "cps": 70.210938, "mean_p": 0.429854, "boundary_entropy": 0.670464, "bpb": 1.277714}, "zh": {"chunks": 3903.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.660853, "cps": 30.492188, "mean_p": 0.436856, "boundary_entropy": 0.664851, "bpb": 3.476179}, "my": {"chunks": 7466.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.394026, "cps": 58.328125, "mean_p": 0.464225, "boundary_entropy": 0.685639, "bpb": 1.228267}}, "bpb": {"en": 2.25808, "de": 2.52815, "vi": 2.09145, "tr": 2.52765, "fi": 2.7387, "ru": 1.63785, "el": 1.71883, "hi": 1.35104, "ta": 1.09708, "th": 1.27771, "zh": 3.47618, "my": 1.22827}, "gini_cpc": 0.159642, "gini_cps": 0.146509, "bpic": 4.781722, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0045115266639628, "vi": 1.5185765461523508, "tr": 1.187628315651002, "fi": 1.0743613461903059, "ru": 1.5187548158871742, "el": 1.5653776746917663, "hi": 1.496609028389591, "ta": 1.969698632212011, "th": 2.1511929297069496, "zh": 2.5444316788986154, "my": 1.5170863294232375}, "t": 187.0} |
| {"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.2672, "de": 2.53711, "vi": 2.09521, "tr": 2.53491, "fi": 2.74703, "ru": 1.61481, "el": 1.70948, "hi": 1.34884, "ta": 1.10237, "th": 1.29516, "zh": 3.48134, "my": 1.23723}, "bytes_per_chunk": {"en": 3.769, "de": 3.755, "vi": 3.288, "tr": 3.427, "fi": 3.782, "ru": 4.57, "el": 4.541, "hi": 6.377, "ta": 5.563, "th": 5.047, "zh": 4.178, "my": 7.369}, "bpb_mean": 1.99756, "frozen_inventory": false, "t": 188.2} |
| {"event": "train", "step": 2200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.4555, "bpb": 2.09984, "chunk_loss": 0.034332, "F": 0.23883, "G": 0.44181, "chunks": 244.56, "bytes_per_chunk": 4.187, "boundary_entropy": 0.67097, "router_noise": 0.0, "grad_norm": 0.7775, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 188.5} |
| {"event": "train", "step": 2250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.54777, "bpb": 2.23296, "chunk_loss": 0.033908, "F": 0.23233, "G": 0.43749, "chunks": 237.91, "bytes_per_chunk": 4.304, "boundary_entropy": 0.66964, "router_noise": 0.0, "grad_norm": 0.7787, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.055, "alpha": 0.03, "t": 191.2} |
| {"event": "train", "step": 2300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.48636, "bpb": 2.14437, "chunk_loss": 0.034604, "F": 0.24329, "G": 0.44157, "chunks": 249.12, "bytes_per_chunk": 4.11, "boundary_entropy": 0.67028, "router_noise": 0.0, "grad_norm": 0.7923, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 194.2} |
| {"event": "train", "step": 2350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.49504, "bpb": 2.15689, "chunk_loss": 0.033073, "F": 0.21857, "G": 0.44019, "chunks": 223.81, "bytes_per_chunk": 4.575, "boundary_entropy": 0.67085, "router_noise": 0.0, "grad_norm": 0.7156, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.06, "alpha": 0.03, "t": 197.2} |
| {"event": "probe", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4133.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.237488, "cps": 32.289062, "mean_p": 0.441443, "boundary_entropy": 0.663995, "bpb": 2.240275}, "de": {"chunks": 5263.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.256319, "cps": 41.117188, "mean_p": 0.448064, "boundary_entropy": 0.670664, "bpb": 2.487997}, "vi": {"chunks": 5945.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.322484, "cps": 46.445312, "mean_p": 0.455715, "boundary_entropy": 0.673784, "bpb": 2.071968}, "tr": {"chunks": 5880.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.324808, "cps": 45.9375, "mean_p": 0.454126, "boundary_entropy": 0.666152, "bpb": 2.481239}, "fi": {"chunks": 5236.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.278348, "cps": 40.90625, "mean_p": 0.456918, "boundary_entropy": 0.67069, "bpb": 2.692526}, "ru": {"chunks": 6765.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.356992, "cps": 52.851562, "mean_p": 0.439405, "boundary_entropy": 0.67168, "bpb": 1.613526}, "el": {"chunks": 8587.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.418409, "cps": 67.085938, "mean_p": 0.452381, "boundary_entropy": 0.672214, "bpb": 1.701787}, "hi": {"chunks": 6925.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.41159, "cps": 54.101562, "mean_p": 0.430085, "boundary_entropy": 0.670834, "bpb": 1.333084}, "ta": {"chunks": 9940.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.531693, "cps": 77.65625, "mean_p": 0.405373, "boundary_entropy": 0.653938, "bpb": 1.074691}, "th": {"chunks": 7428.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.461797, "cps": 58.03125, "mean_p": 0.431697, "boundary_entropy": 0.673114, "bpb": 1.259513}, "zh": {"chunks": 3148.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.533017, "cps": 24.59375, "mean_p": 0.435173, "boundary_entropy": 0.665939, "bpb": 3.411623}, "my": {"chunks": 18927.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.998892, "cps": 147.867188, "mean_p": 0.48191, "boundary_entropy": 0.686968, "bpb": 1.201259}}, "bpb": {"en": 2.24027, "de": 2.488, "vi": 2.07197, "tr": 2.48124, "fi": 2.69253, "ru": 1.61353, "el": 1.70179, "hi": 1.33308, "ta": 1.07469, "th": 1.25951, "zh": 3.41162, "my": 1.20126}, "gini_cpc": 0.221998, "gini_cps": 0.249563, "bpic": 4.352393, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.079293788831535, "vi": 1.3578988856486036, "tr": 1.3676831284766329, "fi": 1.1720508909908172, "ru": 1.5032018498385158, "el": 1.7618110626804206, "hi": 1.7330991918415508, "ta": 2.23882232951475, "th": 1.9445071515112862, "zh": 2.2443986353181042, "my": 4.206076050791855}, "t": 201.1} |
| {"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.24744, "de": 2.49514, "vi": 2.0774, "tr": 2.48998, "fi": 2.70127, "ru": 1.59134, "el": 1.69204, "hi": 1.33088, "ta": 1.08068, "th": 1.27833, "zh": 3.41793, "my": 1.21099}, "bytes_per_chunk": {"en": 4.152, "de": 3.929, "vi": 3.841, "tr": 3.269, "fi": 3.807, "ru": 5.096, "el": 4.571, "hi": 6.162, "ta": 5.073, "th": 6.099, "zh": 5.203, "my": 2.892}, "bpb_mean": 1.96779, "frozen_inventory": false, "t": 202.4} |
| {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 202.6} |
| {"event": "train", "step": 2400, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.37832, "bpb": 1.9885, "chunk_loss": 0.045553, "F": 0.23929, "G": 0.44307, "chunks": 245.03, "bytes_per_chunk": 4.179, "boundary_entropy": 0.67093, "router_noise": 0.0, "grad_norm": 0.7313, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.111, "n_target_mean": 10.027, "alpha": 0.03, "t": 202.7} |
| {"event": "train", "step": 2450, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.46746, "bpb": 2.11709, "chunk_loss": 0.040325, "F": 0.21313, "G": 0.43691, "chunks": 218.25, "bytes_per_chunk": 4.692, "boundary_entropy": 0.67026, "router_noise": 0.0, "grad_norm": 0.7399, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.057, "n_target_mean": 9.8405, "alpha": 0.03, "t": 205.5} |
| {"event": "train", "step": 2500, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.51043, "bpb": 2.17909, "chunk_loss": 0.036795, "F": 0.20468, "G": 0.43215, "chunks": 209.59, "bytes_per_chunk": 4.886, "boundary_entropy": 0.66751, "router_noise": 0.0, "grad_norm": 0.7662, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.054, "n_target_mean": 9.0154, "alpha": 0.03, "t": 208.2} |
| {"event": "train", "step": 2550, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.57321, "bpb": 2.26966, "chunk_loss": 0.038062, "F": 0.22202, "G": 0.43679, "chunks": 227.34, "bytes_per_chunk": 4.504, "boundary_entropy": 0.66828, "router_noise": 0.0, "grad_norm": 0.7437, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.056, "n_target_mean": 8.354, "alpha": 0.03, "t": 211.0} |
| {"event": "probe", "step": 2600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4147.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.238292, "cps": 32.398438, "mean_p": 0.44127, "boundary_entropy": 0.663959, "bpb": 2.207724}, "de": {"chunks": 5101.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.248429, "cps": 39.851562, "mean_p": 0.445312, "boundary_entropy": 0.669387, "bpb": 2.462904}, "vi": {"chunks": 4990.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.270681, "cps": 38.984375, "mean_p": 0.449242, "boundary_entropy": 0.672792, "bpb": 2.056541}, "tr": {"chunks": 5724.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.316191, "cps": 44.71875, "mean_p": 0.453176, "boundary_entropy": 0.665491, "bpb": 2.452728}, "fi": {"chunks": 4965.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.263941, "cps": 38.789062, "mean_p": 0.452175, "boundary_entropy": 0.669066, "bpb": 2.668266}, "ru": {"chunks": 5821.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.307177, "cps": 45.476562, "mean_p": 0.423638, "boundary_entropy": 0.664842, "bpb": 1.587511}, "el": {"chunks": 6387.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.311212, "cps": 49.898438, "mean_p": 0.429678, "boundary_entropy": 0.663139, "bpb": 1.68656}, "hi": {"chunks": 4856.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.288618, "cps": 37.9375, "mean_p": 0.41299, "boundary_entropy": 0.663256, "bpb": 1.321471}, "ta": {"chunks": 5279.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.282375, "cps": 41.242188, "mean_p": 0.384574, "boundary_entropy": 0.643068, "bpb": 1.077086}, "th": {"chunks": 5614.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.349021, "cps": 43.859375, "mean_p": 0.411378, "boundary_entropy": 0.663215, "bpb": 1.242517}, "zh": {"chunks": 3718.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.629529, "cps": 29.046875, "mean_p": 0.439561, "boundary_entropy": 0.666594, "bpb": 3.378518}, "my": {"chunks": 4378.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.231053, "cps": 34.203125, "mean_p": 0.445335, "boundary_entropy": 0.681345, "bpb": 1.195499}}, "bpb": {"en": 2.20772, "de": 2.4629, "vi": 2.05654, "tr": 2.45273, "fi": 2.66827, "ru": 1.58751, "el": 1.68656, "hi": 1.32147, "ta": 1.07709, "th": 1.24252, "zh": 3.37852, "my": 1.1955}, "gini_cpc": 0.138413, "gini_cps": 0.080204, "bpic": 6.293555, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0425406584642265, "vi": 1.1359193260374485, "tr": 1.3269029465544808, "fi": 1.1076369988433545, "ru": 1.2890758427270117, "el": 1.3060089581015641, "hi": 1.2111939420931581, "ta": 1.1849943436685235, "th": 1.4646755364837107, "zh": 2.6418370564125317, "my": 0.9696220117154363}, "t": 214.7} |
| {"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.21517, "de": 2.47023, "vi": 2.06059, "tr": 2.45887, "fi": 2.67583, "ru": 1.56445, "el": 1.67748, "hi": 1.31798, "ta": 1.08195, "th": 1.25961, "zh": 3.38123, "my": 1.20515}, "bytes_per_chunk": {"en": 4.1, "de": 4.028, "vi": 4.822, "tr": 3.364, "fi": 4.059, "ru": 5.834, "el": 6.071, "hi": 8.561, "ta": 9.298, "th": 7.953, "zh": 4.472, "my": 11.838}, "bpb_mean": 1.94738, "frozen_inventory": false, "t": 215.9} |
| {"event": "train", "step": 2600, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.60171, "bpb": 2.31078, "chunk_loss": 0.037267, "F": 0.22577, "G": 0.43712, "chunks": 231.19, "bytes_per_chunk": 4.429, "boundary_entropy": 0.66835, "router_noise": 0.0, "grad_norm": 0.8063, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.102, "n_target_mean": 8.0814, "alpha": 0.03, "t": 216.1} |
| {"event": "train", "step": 2650, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.50311, "bpb": 2.16854, "chunk_loss": 0.038499, "F": 0.21155, "G": 0.43489, "chunks": 216.62, "bytes_per_chunk": 4.727, "boundary_entropy": 0.66888, "router_noise": 0.0, "grad_norm": 0.7176, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.055, "n_target_mean": 9.2749, "alpha": 0.03, "t": 218.9} |
| {"event": "train", "step": 2700, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.5232, "bpb": 2.19752, "chunk_loss": 0.036562, "F": 0.21268, "G": 0.4366, "chunks": 217.78, "bytes_per_chunk": 4.702, "boundary_entropy": 0.66867, "router_noise": 0.0, "grad_norm": 0.7536, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.056, "n_target_mean": 8.6096, "alpha": 0.03, "t": 221.7} |
| {"event": "train", "step": 2750, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.40447, "bpb": 2.02622, "chunk_loss": 0.036305, "F": 0.20654, "G": 0.42926, "chunks": 211.5, "bytes_per_chunk": 4.842, "boundary_entropy": 0.66508, "router_noise": 0.0, "grad_norm": 0.7572, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.056, "n_target_mean": 8.5736, "alpha": 0.03, "t": 224.5} |
| {"event": "probe", "step": 2800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4144.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.23812, "cps": 32.375, "mean_p": 0.439681, "boundary_entropy": 0.662997, "bpb": 2.187296}, "de": {"chunks": 5062.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.24653, "cps": 39.546875, "mean_p": 0.445203, "boundary_entropy": 0.6696, "bpb": 2.430322}, "vi": {"chunks": 4868.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.264063, "cps": 38.03125, "mean_p": 0.44804, "boundary_entropy": 0.672206, "bpb": 2.046124}, "tr": {"chunks": 5416.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.299177, "cps": 42.3125, "mean_p": 0.450424, "boundary_entropy": 0.664997, "bpb": 2.428972}, "fi": {"chunks": 4945.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.262878, "cps": 38.632812, "mean_p": 0.453569, "boundary_entropy": 0.669698, "bpb": 2.641242}, "ru": {"chunks": 4581.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.241741, "cps": 35.789062, "mean_p": 0.414159, "boundary_entropy": 0.661173, "bpb": 1.565208}, "el": {"chunks": 6095.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.296984, "cps": 47.617188, "mean_p": 0.428394, "boundary_entropy": 0.66351, "bpb": 1.640899}, "hi": {"chunks": 2585.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.15364, "cps": 20.195312, "mean_p": 0.399841, "boundary_entropy": 0.657881, "bpb": 1.308899}, "ta": {"chunks": 5232.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.279861, "cps": 40.875, "mean_p": 0.378036, "boundary_entropy": 0.63927, "bpb": 1.079437}, "th": {"chunks": 4942.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.307243, "cps": 38.609375, "mean_p": 0.39353, "boundary_entropy": 0.653873, "bpb": 1.24324}, "zh": {"chunks": 3254.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.550965, "cps": 25.421875, "mean_p": 0.433254, "boundary_entropy": 0.665048, "bpb": 3.362561}, "my": {"chunks": 3856.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.203504, "cps": 30.125, "mean_p": 0.430431, "boundary_entropy": 0.677209, "bpb": 1.193986}}, "bpb": {"en": 2.1873, "de": 2.43032, "vi": 2.04612, "tr": 2.42897, "fi": 2.64124, "ru": 1.56521, "el": 1.6409, "hi": 1.3089, "ta": 1.07944, "th": 1.24324, "zh": 3.36256, "my": 1.19399}, "gini_cpc": 0.152206, "gini_cps": 0.111337, "bpic": 6.980375, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0353188162872573, "vi": 1.1089495807559273, "tr": 1.2564131630308506, "fi": 1.1039738490930613, "ru": 1.0152089806542313, "el": 1.2472032625793041, "hi": 0.6452230116977516, "ta": 1.1752943253879333, "th": 1.2902861907602348, "zh": 2.3138141860839117, "my": 0.8546297813238916}, "t": 228.3} |
| {"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.19565, "de": 2.43801, "vi": 2.04849, "tr": 2.43536, "fi": 2.64757, "ru": 1.5418, "el": 1.63137, "hi": 1.30522, "ta": 1.08355, "th": 1.26036, "zh": 3.36261, "my": 1.20301}, "bytes_per_chunk": {"en": 4.134, "de": 4.056, "vi": 4.993, "tr": 3.554, "fi": 4.083, "ru": 7.273, "el": 6.174, "hi": 15.787, "ta": 9.336, "th": 9.401, "zh": 5.147, "my": 13.643}, "bpb_mean": 1.92942, "frozen_inventory": false, "t": 229.5} |
| {"event": "train", "step": 2800, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.4758, "bpb": 2.12913, "chunk_loss": 0.036526, "F": 0.19791, "G": 0.43182, "chunks": 202.66, "bytes_per_chunk": 5.053, "boundary_entropy": 0.66773, "router_noise": 0.0, "grad_norm": 0.8067, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.105, "n_target_mean": 9.2272, "alpha": 0.03, "t": 229.8} |
| {"event": "train", "step": 2850, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.43701, "bpb": 2.07316, "chunk_loss": 0.037052, "F": 0.20248, "G": 0.4307, "chunks": 207.34, "bytes_per_chunk": 4.939, "boundary_entropy": 0.66622, "router_noise": 0.0, "grad_norm": 0.7319, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.054, "n_target_mean": 9.3574, "alpha": 0.03, "t": 232.4} |
| {"event": "train", "step": 2900, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.45887, "bpb": 2.1047, "chunk_loss": 0.036545, "F": 0.2037, "G": 0.43155, "chunks": 208.59, "bytes_per_chunk": 4.909, "boundary_entropy": 0.66691, "router_noise": 0.0, "grad_norm": 0.7644, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.051, "n_target_mean": 8.9121, "alpha": 0.03, "t": 235.0} |
| {"event": "train", "step": 2950, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.36253, "bpb": 1.96572, "chunk_loss": 0.035684, "F": 0.1864, "G": 0.42455, "chunks": 190.88, "bytes_per_chunk": 5.365, "boundary_entropy": 0.66394, "router_noise": 0.0, "grad_norm": 0.7321, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.056, "n_target_mean": 9.8268, "alpha": 0.03, "t": 237.8} |
| {"event": "probe", "step": 3000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3983.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.228869, "cps": 31.117188, "mean_p": 0.43832, "boundary_entropy": 0.662736, "bpb": 2.165337}, "de": {"chunks": 5257.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.256027, "cps": 41.070312, "mean_p": 0.447398, "boundary_entropy": 0.669891, "bpb": 2.406137}, "vi": {"chunks": 4577.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.248278, "cps": 35.757812, "mean_p": 0.442676, "boundary_entropy": 0.67086, "bpb": 2.02397}, "tr": {"chunks": 5319.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.293819, "cps": 41.554688, "mean_p": 0.449591, "boundary_entropy": 0.665329, "bpb": 2.401791}, "fi": {"chunks": 5096.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.270905, "cps": 39.8125, "mean_p": 0.453239, "boundary_entropy": 0.669528, "bpb": 2.591303}, "ru": {"chunks": 5515.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.291029, "cps": 43.085938, "mean_p": 0.416799, "boundary_entropy": 0.660691, "bpb": 1.546885}, "el": {"chunks": 5883.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.286654, "cps": 45.960938, "mean_p": 0.424226, "boundary_entropy": 0.661715, "bpb": 1.610059}, "hi": {"chunks": 3088.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.183536, "cps": 24.125, "mean_p": 0.402356, "boundary_entropy": 0.659097, "bpb": 1.284379}, "ta": {"chunks": 2883.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.154212, "cps": 22.523438, "mean_p": 0.367969, "boundary_entropy": 0.634885, "bpb": 1.058399}, "th": {"chunks": 3879.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.241156, "cps": 30.304688, "mean_p": 0.386518, "boundary_entropy": 0.650475, "bpb": 1.224722}, "zh": {"chunks": 3258.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.551642, "cps": 25.453125, "mean_p": 0.433951, "boundary_entropy": 0.664953, "bpb": 3.340513}, "my": {"chunks": 3405.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.179702, "cps": 26.601562, "mean_p": 0.42585, "boundary_entropy": 0.675494, "bpb": 1.169744}}, "bpb": {"en": 2.16534, "de": 2.40614, "vi": 2.02397, "tr": 2.40179, "fi": 2.5913, "ru": 1.54688, "el": 1.61006, "hi": 1.28438, "ta": 1.0584, "th": 1.22472, "zh": 3.34051, "my": 1.16974}, "gini_cpc": 0.170995, "gini_cps": 0.132288, "bpic": 7.360163, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1186632825704637, "vi": 1.0848047652482367, "tr": 1.2837878250420445, "fi": 1.1836719325967429, "ru": 1.2715988200511805, "el": 1.2524829340660386, "hi": 0.8019292095417411, "ta": 0.6738030719888222, "th": 1.0536891986778436, "zh": 2.410301944455946, "my": 0.7851769720488772}, "t": 241.6} |
| {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.17288, "de": 2.41385, "vi": 2.02512, "tr": 2.40879, "fi": 2.59895, "ru": 1.52359, "el": 1.60142, "hi": 1.28142, "ta": 1.06389, "th": 1.24375, "zh": 3.3409, "my": 1.17992}, "bytes_per_chunk": {"en": 4.304, "de": 3.983, "vi": 5.191, "tr": 3.626, "fi": 3.927, "ru": 6.137, "el": 6.48, "hi": 13.458, "ta": 16.362, "th": 11.61, "zh": 5.155, "my": 15.42}, "bpb_mean": 1.90454, "frozen_inventory": false, "t": 242.8} |
| {"event": "train", "step": 3000, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.43761, "bpb": 2.07404, "chunk_loss": 0.034787, "F": 0.18268, "G": 0.42233, "chunks": 187.06, "bytes_per_chunk": 5.474, "boundary_entropy": 0.66319, "router_noise": 0.0, "grad_norm": 0.707, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.108, "n_target_mean": 9.5059, "alpha": 0.03, "t": 243.2} |
| {"event": "train", "step": 3050, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.39541, "bpb": 2.01315, "chunk_loss": 0.037536, "F": 0.19998, "G": 0.42921, "chunks": 204.78, "bytes_per_chunk": 5.0, "boundary_entropy": 0.66626, "router_noise": 0.0, "grad_norm": 0.7134, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.054, "n_target_mean": 9.6662, "alpha": 0.03, "t": 245.9} |
| {"event": "train", "step": 3100, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.41267, "bpb": 2.03805, "chunk_loss": 0.036264, "F": 0.20853, "G": 0.42894, "chunks": 213.53, "bytes_per_chunk": 4.796, "boundary_entropy": 0.66468, "router_noise": 0.0, "grad_norm": 0.7697, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.056, "n_target_mean": 8.4014, "alpha": 0.03, "t": 248.7} |
| {"event": "train", "step": 3150, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.3605, "bpb": 1.96279, "chunk_loss": 0.036198, "F": 0.18457, "G": 0.4231, "chunks": 189.0, "bytes_per_chunk": 5.418, "boundary_entropy": 0.6633, "router_noise": 0.0, "grad_norm": 0.733, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.051, "n_target_mean": 9.7269, "alpha": 0.03, "t": 251.3} |
| {"event": "probe", "step": 3200, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4145.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.238177, "cps": 32.382812, "mean_p": 0.438448, "boundary_entropy": 0.661908, "bpb": 2.153774}, "de": {"chunks": 5217.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.254079, "cps": 40.757812, "mean_p": 0.444593, "boundary_entropy": 0.668472, "bpb": 2.372272}, "vi": {"chunks": 4514.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.24486, "cps": 35.265625, "mean_p": 0.444438, "boundary_entropy": 0.671447, "bpb": 2.010508}, "tr": {"chunks": 5504.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.304038, "cps": 43.0, "mean_p": 0.449204, "boundary_entropy": 0.663701, "bpb": 2.37844}, "fi": {"chunks": 4846.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257615, "cps": 37.859375, "mean_p": 0.451708, "boundary_entropy": 0.669118, "bpb": 2.562088}, "ru": {"chunks": 5117.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.270026, "cps": 39.976562, "mean_p": 0.408408, "boundary_entropy": 0.656149, "bpb": 1.517013}, "el": {"chunks": 6185.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.301369, "cps": 48.320312, "mean_p": 0.423152, "boundary_entropy": 0.660471, "bpb": 1.583917}, "hi": {"chunks": 3742.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.222407, "cps": 29.234375, "mean_p": 0.400959, "boundary_entropy": 0.657646, "bpb": 1.285139}, "ta": {"chunks": 4429.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.236908, "cps": 34.601562, "mean_p": 0.376885, "boundary_entropy": 0.639613, "bpb": 1.040333}, "th": {"chunks": 4828.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.300155, "cps": 37.71875, "mean_p": 0.38459, "boundary_entropy": 0.646346, "bpb": 1.216359}, "zh": {"chunks": 3438.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.58212, "cps": 26.859375, "mean_p": 0.435182, "boundary_entropy": 0.665207, "bpb": 3.301278}, "my": {"chunks": 4131.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.218018, "cps": 32.273438, "mean_p": 0.431733, "boundary_entropy": 0.677227, "bpb": 1.163079}}, "bpb": {"en": 2.15377, "de": 2.37227, "vi": 2.01051, "tr": 2.37844, "fi": 2.56209, "ru": 1.51701, "el": 1.58392, "hi": 1.28514, "ta": 1.04033, "th": 1.21636, "zh": 3.30128, "my": 1.16308}, "gini_cpc": 0.135576, "gini_cps": 0.089284, "bpic": 6.841504, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.066763173959222, "vi": 1.0280589022232125, "tr": 1.2765195166886816, "fi": 1.0816110405334147, "ru": 1.1337199474205182, "el": 1.2653143811641616, "hi": 0.9337880150631725, "ta": 0.9946717786730098, "th": 1.2602182989718749, "zh": 2.4440608536553983, "my": 0.9153588909320233}, "t": 254.8} |
| {"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.16015, "de": 2.38014, "vi": 2.01459, "tr": 2.38452, "fi": 2.57156, "ru": 1.49435, "el": 1.57657, "hi": 1.28057, "ta": 1.04612, "th": 1.23628, "zh": 3.30273, "my": 1.17176}, "bytes_per_chunk": {"en": 4.237, "de": 3.977, "vi": 5.233, "tr": 3.592, "fi": 4.091, "ru": 6.59, "el": 6.263, "hi": 11.354, "ta": 11.024, "th": 9.292, "zh": 4.942, "my": 12.7}, "bpb_mean": 1.88495, "frozen_inventory": false, "t": 256.1} |
| {"event": "train", "step": 3200, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.38563, "bpb": 1.99904, "chunk_loss": 0.03592, "F": 0.18668, "G": 0.42386, "chunks": 191.16, "bytes_per_chunk": 5.357, "boundary_entropy": 0.6634, "router_noise": 0.0, "grad_norm": 0.6953, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.104, "n_target_mean": 9.469, "alpha": 0.03, "t": 256.5} |
| {"event": "train", "step": 3250, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.45235, "bpb": 2.0953, "chunk_loss": 0.03637, "F": 0.20279, "G": 0.42861, "chunks": 207.66, "bytes_per_chunk": 4.931, "boundary_entropy": 0.66532, "router_noise": 0.0, "grad_norm": 0.7806, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.06, "n_target_mean": 8.943, "alpha": 0.03, "t": 259.5} |
| {"event": "train", "step": 3300, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.41742, "bpb": 2.0449, "chunk_loss": 0.035631, "F": 0.20349, "G": 0.42785, "chunks": 208.38, "bytes_per_chunk": 4.914, "boundary_entropy": 0.66437, "router_noise": 0.0, "grad_norm": 0.8437, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.058, "n_target_mean": 8.4233, "alpha": 0.03, "t": 262.4} |
| {"event": "train", "step": 3350, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.4768, "bpb": 2.13057, "chunk_loss": 0.036574, "F": 0.20587, "G": 0.42788, "chunks": 210.81, "bytes_per_chunk": 4.857, "boundary_entropy": 0.66439, "router_noise": 0.0, "grad_norm": 0.7488, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.05, "n_target_mean": 8.9185, "alpha": 0.03, "t": 264.9} |
| {"event": "probe", "step": 3400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3992.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.229386, "cps": 31.1875, "mean_p": 0.437005, "boundary_entropy": 0.662004, "bpb": 2.134438}, "de": {"chunks": 5194.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.252959, "cps": 40.578125, "mean_p": 0.444403, "boundary_entropy": 0.668824, "bpb": 2.346507}, "vi": {"chunks": 4585.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.248712, "cps": 35.820312, "mean_p": 0.441798, "boundary_entropy": 0.671114, "bpb": 1.990032}, "tr": {"chunks": 5477.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.302547, "cps": 42.789062, "mean_p": 0.449606, "boundary_entropy": 0.665022, "bpb": 2.348949}, "fi": {"chunks": 4876.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25921, "cps": 38.09375, "mean_p": 0.450556, "boundary_entropy": 0.668802, "bpb": 2.558534}, "ru": {"chunks": 5143.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.271398, "cps": 40.179688, "mean_p": 0.406984, "boundary_entropy": 0.655147, "bpb": 1.500527}, "el": {"chunks": 6168.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.300541, "cps": 48.1875, "mean_p": 0.418093, "boundary_entropy": 0.658009, "bpb": 1.578154}, "hi": {"chunks": 3346.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.198871, "cps": 26.140625, "mean_p": 0.399906, "boundary_entropy": 0.657593, "bpb": 1.264399}, "ta": {"chunks": 4799.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.2567, "cps": 37.492188, "mean_p": 0.373361, "boundary_entropy": 0.636323, "bpb": 1.033617}, "th": {"chunks": 4317.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.268387, "cps": 33.726562, "mean_p": 0.377277, "boundary_entropy": 0.642947, "bpb": 1.182914}, "zh": {"chunks": 3641.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.616492, "cps": 28.445312, "mean_p": 0.436268, "boundary_entropy": 0.665274, "bpb": 3.280833}, "my": {"chunks": 3357.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.177169, "cps": 26.226562, "mean_p": 0.42516, "boundary_entropy": 0.675726, "bpb": 1.147575}}, "bpb": {"en": 2.13444, "de": 2.34651, "vi": 1.99003, "tr": 2.34895, "fi": 2.55853, "ru": 1.50053, "el": 1.57815, "hi": 1.2644, "ta": 1.03362, "th": 1.18291, "zh": 3.28083, "my": 1.14758}, "gini_cpc": 0.15832, "gini_cps": 0.104197, "bpic": 6.991183, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1027653856390518, "vi": 1.0842508858237223, "tr": 1.318942240111887, "fi": 1.1300181032949956, "ru": 1.1831529682055404, "el": 1.3101985289098883, "hi": 0.8669707617314868, "ta": 1.119074160624003, "th": 1.1700234631526027, "zh": 2.687576431223436, "my": 0.7723631636293486}, "t": 268.5} |
| {"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.13918, "de": 2.35635, "vi": 1.99283, "tr": 2.35589, "fi": 2.56439, "ru": 1.47799, "el": 1.56975, "hi": 1.26165, "ta": 1.03968, "th": 1.20315, "zh": 3.28278, "my": 1.15618}, "bytes_per_chunk": {"en": 4.343, "de": 3.981, "vi": 5.197, "tr": 3.573, "fi": 4.059, "ru": 6.548, "el": 6.324, "hi": 12.285, "ta": 10.256, "th": 10.512, "zh": 4.496, "my": 15.751}, "bpb_mean": 1.86665, "frozen_inventory": false, "t": 269.7} |
| {"event": "train", "step": 3400, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.40309, "bpb": 2.02423, "chunk_loss": 0.035825, "F": 0.1882, "G": 0.42161, "chunks": 192.72, "bytes_per_chunk": 5.313, "boundary_entropy": 0.66303, "router_noise": 0.0, "grad_norm": 0.7301, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.102, "n_target_mean": 9.6778, "alpha": 0.03, "t": 270.0} |
| {"event": "train", "step": 3450, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.26332, "bpb": 1.82258, "chunk_loss": 0.037322, "F": 0.19577, "G": 0.42518, "chunks": 200.47, "bytes_per_chunk": 5.108, "boundary_entropy": 0.66308, "router_noise": 0.0, "grad_norm": 0.6923, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.054, "n_target_mean": 9.8297, "alpha": 0.03, "t": 272.7} |
| {"event": "train", "step": 3500, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.45301, "bpb": 2.09625, "chunk_loss": 0.034932, "F": 0.1962, "G": 0.42361, "chunks": 200.91, "bytes_per_chunk": 5.097, "boundary_entropy": 0.66237, "router_noise": 0.0, "grad_norm": 0.7875, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.053, "n_target_mean": 8.7543, "alpha": 0.03, "t": 275.3} |
| {"event": "train", "step": 3550, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.38403, "bpb": 1.99673, "chunk_loss": 0.036283, "F": 0.19144, "G": 0.4266, "chunks": 196.03, "bytes_per_chunk": 5.224, "boundary_entropy": 0.66455, "router_noise": 0.0, "grad_norm": 0.7528, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.049, "n_target_mean": 9.6795, "alpha": 0.03, "t": 277.8} |
| {"event": "probe", "step": 3600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3852.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.221341, "cps": 30.09375, "mean_p": 0.434589, "boundary_entropy": 0.661205, "bpb": 2.118584}, "de": {"chunks": 5026.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.244777, "cps": 39.265625, "mean_p": 0.44301, "boundary_entropy": 0.668057, "bpb": 2.339885}, "vi": {"chunks": 4437.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.240683, "cps": 34.664062, "mean_p": 0.440347, "boundary_entropy": 0.670317, "bpb": 1.99832}, "tr": {"chunks": 5399.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.298238, "cps": 42.179688, "mean_p": 0.448536, "boundary_entropy": 0.663506, "bpb": 2.328775}, "fi": {"chunks": 5260.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.279624, "cps": 41.09375, "mean_p": 0.452438, "boundary_entropy": 0.667827, "bpb": 2.509237}, "ru": {"chunks": 4946.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.261003, "cps": 38.640625, "mean_p": 0.404006, "boundary_entropy": 0.653186, "bpb": 1.491283}, "el": {"chunks": 5435.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.264825, "cps": 42.460938, "mean_p": 0.414964, "boundary_entropy": 0.65791, "bpb": 1.556194}, "hi": {"chunks": 4189.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.248975, "cps": 32.726562, "mean_p": 0.402122, "boundary_entropy": 0.657873, "bpb": 1.248332}, "ta": {"chunks": 4605.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.246323, "cps": 35.976562, "mean_p": 0.368543, "boundary_entropy": 0.63198, "bpb": 1.032608}, "th": {"chunks": 4695.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.291887, "cps": 36.679688, "mean_p": 0.380367, "boundary_entropy": 0.643625, "bpb": 1.168881}, "zh": {"chunks": 3407.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.576871, "cps": 26.617188, "mean_p": 0.435365, "boundary_entropy": 0.665335, "bpb": 3.293557}, "my": {"chunks": 3129.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.165136, "cps": 24.445312, "mean_p": 0.426446, "boundary_entropy": 0.675944, "bpb": 1.143151}}, "bpb": {"en": 2.11858, "de": 2.33989, "vi": 1.99832, "tr": 2.32878, "fi": 2.50924, "ru": 1.49128, "el": 1.55619, "hi": 1.24833, "ta": 1.03261, "th": 1.16888, "zh": 3.29356, "my": 1.14315}, "gini_cpc": 0.145257, "gini_cps": 0.09034, "bpic": 7.057392, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1058797835199716, "vi": 1.0873869797647202, "tr": 1.3474126420587764, "fi": 1.2633151445324875, "ru": 1.1791871542590355, "el": 1.1964554850169622, "hi": 1.1248461577629343, "ta": 1.1128637773606316, "th": 1.318719332750371, "zh": 2.6062527626480487, "my": 0.7460707760935487}, "t": 281.5} |
| {"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 2.12755, "de": 2.34883, "vi": 2.00174, "tr": 2.33549, "fi": 2.51834, "ru": 1.46757, "el": 1.54659, "hi": 1.24466, "ta": 1.03817, "th": 1.18738, "zh": 3.29165, "my": 1.15213}, "bytes_per_chunk": {"en": 4.51, "de": 4.131, "vi": 5.339, "tr": 3.571, "fi": 3.804, "ru": 6.898, "el": 7.206, "hi": 10.172, "ta": 10.638, "th": 9.75, "zh": 5.044, "my": 16.648}, "bpb_mean": 1.85501, "frozen_inventory": false, "t": 282.7} |
| {"event": "train", "step": 3600, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.65202, "bpb": 2.38336, "chunk_loss": 0.035082, "F": 0.20721, "G": 0.42938, "chunks": 212.19, "bytes_per_chunk": 4.826, "boundary_entropy": 0.66504, "router_noise": 0.0, "grad_norm": 0.8303, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.102, "n_target_mean": 7.8431, "alpha": 0.03, "t": 282.9} |
| {"event": "train", "step": 3650, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.44891, "bpb": 2.09034, "chunk_loss": 0.036648, "F": 0.20358, "G": 0.42417, "chunks": 208.47, "bytes_per_chunk": 4.912, "boundary_entropy": 0.66232, "router_noise": 0.0, "grad_norm": 0.8223, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.051, "n_target_mean": 8.856, "alpha": 0.03, "t": 285.5} |
| {"event": "train", "step": 3700, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.36158, "bpb": 1.96435, "chunk_loss": 0.035673, "F": 0.17578, "G": 0.41643, "chunks": 180.0, "bytes_per_chunk": 5.689, "boundary_entropy": 0.65925, "router_noise": 0.0, "grad_norm": 0.7648, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.055, "n_target_mean": 10.295, "alpha": 0.03, "t": 288.2} |
| {"event": "train", "step": 3750, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.40461, "bpb": 2.02643, "chunk_loss": 0.036353, "F": 0.20444, "G": 0.42453, "chunks": 209.34, "bytes_per_chunk": 4.891, "boundary_entropy": 0.66297, "router_noise": 0.0, "grad_norm": 0.7822, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.056, "n_target_mean": 9.0563, "alpha": 0.03, "t": 291.1} |
| {"event": "probe", "step": 3800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4105.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.235879, "cps": 32.070312, "mean_p": 0.435564, "boundary_entropy": 0.660437, "bpb": 2.113168}, "de": {"chunks": 5160.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.251303, "cps": 40.3125, "mean_p": 0.443365, "boundary_entropy": 0.668074, "bpb": 2.313828}, "vi": {"chunks": 4314.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.234011, "cps": 33.703125, "mean_p": 0.439481, "boundary_entropy": 0.670411, "bpb": 1.97367}, "tr": {"chunks": 5372.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.296746, "cps": 41.96875, "mean_p": 0.448328, "boundary_entropy": 0.663896, "bpb": 2.302516}, "fi": {"chunks": 4777.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253947, "cps": 37.320312, "mean_p": 0.44694, "boundary_entropy": 0.667394, "bpb": 2.494993}, "ru": {"chunks": 5446.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.287388, "cps": 42.546875, "mean_p": 0.40529, "boundary_entropy": 0.653022, "bpb": 1.471742}, "el": {"chunks": 5730.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.279199, "cps": 44.765625, "mean_p": 0.41938, "boundary_entropy": 0.659044, "bpb": 1.538934}, "hi": {"chunks": 3216.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.191144, "cps": 25.125, "mean_p": 0.396761, "boundary_entropy": 0.656115, "bpb": 1.2412}, "ta": {"chunks": 3702.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.198021, "cps": 28.921875, "mean_p": 0.360919, "boundary_entropy": 0.628938, "bpb": 1.01745}, "th": {"chunks": 4835.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.300591, "cps": 37.773438, "mean_p": 0.385248, "boundary_entropy": 0.646684, "bpb": 1.152467}, "zh": {"chunks": 3236.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.547917, "cps": 25.28125, "mean_p": 0.433183, "boundary_entropy": 0.664583, "bpb": 3.251635}, "my": {"chunks": 3664.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.193371, "cps": 28.625, "mean_p": 0.429938, "boundary_entropy": 0.67696, "bpb": 1.132959}}, "bpb": {"en": 2.11317, "de": 2.31383, "vi": 1.97367, "tr": 2.30252, "fi": 2.49499, "ru": 1.47174, "el": 1.53893, "hi": 1.2412, "ta": 1.01745, "th": 1.15247, "zh": 3.25164, "my": 1.13296}, "gini_cpc": 0.152419, "gini_cps": 0.108052, "bpic": 7.165842, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.065389109821313, "vi": 0.9920828852864543, "tr": 1.2580456816230148, "fi": 1.0765998540129065, "ru": 1.2183705179666986, "el": 1.1836539059026672, "hi": 0.8103486741968353, "ta": 0.8395023257829136, "th": 1.27434310039123, "zh": 2.322876011636631, "my": 0.8197908141186199}, "t": 294.9} |
| {"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 2.1223, "de": 2.32337, "vi": 1.97656, "tr": 2.30568, "fi": 2.5023, "ru": 1.44967, "el": 1.5301, "hi": 1.23556, "ta": 1.02362, "th": 1.16984, "zh": 3.24979, "my": 1.14167}, "bytes_per_chunk": {"en": 4.182, "de": 4.003, "vi": 5.517, "tr": 3.597, "fi": 4.168, "ru": 6.26, "el": 6.751, "hi": 12.738, "ta": 13.369, "th": 9.51, "zh": 5.088, "my": 14.099}, "bpb_mean": 1.83587, "frozen_inventory": false, "t": 296.1} |
| {"event": "train", "step": 3800, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.47704, "bpb": 2.13091, "chunk_loss": 0.035397, "F": 0.18784, "G": 0.42319, "chunks": 192.34, "bytes_per_chunk": 5.324, "boundary_entropy": 0.66304, "router_noise": 0.0, "grad_norm": 0.7559, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.106, "n_target_mean": 9.1031, "alpha": 0.03, "t": 296.4} |
| {"event": "train", "step": 3850, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.48249, "bpb": 2.13878, "chunk_loss": 0.034789, "F": 0.19006, "G": 0.41904, "chunks": 194.62, "bytes_per_chunk": 5.261, "boundary_entropy": 0.66126, "router_noise": 0.0, "grad_norm": 0.7741, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.053, "n_target_mean": 9.1437, "alpha": 0.03, "t": 299.0} |
| {"event": "train", "step": 3900, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.48239, "bpb": 2.13863, "chunk_loss": 0.036245, "F": 0.21484, "G": 0.43021, "chunks": 220.0, "bytes_per_chunk": 4.655, "boundary_entropy": 0.66508, "router_noise": 0.0, "grad_norm": 0.8074, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.05, "n_target_mean": 8.4141, "alpha": 0.03, "t": 301.5} |
| {"event": "train", "step": 3950, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.5532, "bpb": 2.2408, "chunk_loss": 0.035135, "F": 0.18973, "G": 0.42147, "chunks": 194.28, "bytes_per_chunk": 5.271, "boundary_entropy": 0.66233, "router_noise": 0.0, "grad_norm": 0.7493, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.053, "n_target_mean": 9.0542, "alpha": 0.03, "t": 304.2} |
| {"event": "probe", "step": 4000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3976.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.228466, "cps": 31.0625, "mean_p": 0.432686, "boundary_entropy": 0.659397, "bpb": 2.093626}, "de": {"chunks": 5227.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.254566, "cps": 40.835938, "mean_p": 0.441736, "boundary_entropy": 0.666559, "bpb": 2.306148}, "vi": {"chunks": 4153.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.225278, "cps": 32.445312, "mean_p": 0.437003, "boundary_entropy": 0.669288, "bpb": 1.973914}, "tr": {"chunks": 5557.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.306966, "cps": 43.414062, "mean_p": 0.448018, "boundary_entropy": 0.662581, "bpb": 2.286278}, "fi": {"chunks": 5008.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.266227, "cps": 39.125, "mean_p": 0.448204, "boundary_entropy": 0.666568, "bpb": 2.473421}, "ru": {"chunks": 5042.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.266069, "cps": 39.390625, "mean_p": 0.40246, "boundary_entropy": 0.652156, "bpb": 1.460541}, "el": {"chunks": 6326.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.30824, "cps": 49.421875, "mean_p": 0.417294, "boundary_entropy": 0.656896, "bpb": 1.518876}, "hi": {"chunks": 3950.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.23477, "cps": 30.859375, "mean_p": 0.401968, "boundary_entropy": 0.658462, "bpb": 1.229505}, "ta": {"chunks": 4898.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.261995, "cps": 38.265625, "mean_p": 0.363709, "boundary_entropy": 0.629175, "bpb": 1.00982}, "th": {"chunks": 4921.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.305937, "cps": 38.445312, "mean_p": 0.371202, "boundary_entropy": 0.637629, "bpb": 1.14095}, "zh": {"chunks": 3385.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.573146, "cps": 26.445312, "mean_p": 0.434457, "boundary_entropy": 0.664654, "bpb": 3.222369}, "my": {"chunks": 3827.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.201974, "cps": 29.898438, "mean_p": 0.430331, "boundary_entropy": 0.676845, "bpb": 1.130296}}, "bpb": {"en": 2.09363, "de": 2.30615, "vi": 1.97391, "tr": 2.28628, "fi": 2.47342, "ru": 1.46054, "el": 1.51888, "hi": 1.22951, "ta": 1.00982, "th": 1.14095, "zh": 3.22237, "my": 1.1303}, "gini_cpc": 0.140132, "gini_cps": 0.095933, "bpic": 6.820348, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1142376711934785, "vi": 0.9860445422739857, "tr": 1.3435925584996473, "fi": 1.1652797011723202, "ru": 1.1645854812250815, "el": 1.3491681708435086, "hi": 1.0275897584616167, "ta": 1.1467560915102815, "th": 1.339090855359074, "zh": 2.508666756720479, "my": 0.8840418620016489}, "t": 307.8} |
| {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.10079, "de": 2.3177, "vi": 1.97648, "tr": 2.29348, "fi": 2.4843, "ru": 1.43809, "el": 1.51066, "hi": 1.22609, "ta": 1.01613, "th": 1.15894, "zh": 3.22201, "my": 1.13892}, "bytes_per_chunk": {"en": 4.341, "de": 3.959, "vi": 5.723, "tr": 3.496, "fi": 3.993, "ru": 6.663, "el": 6.153, "hi": 10.596, "ta": 10.176, "th": 9.434, "zh": 4.955, "my": 13.535}, "bpb_mean": 1.82363, "frozen_inventory": false, "t": 309.0} |
| {"event": "train", "step": 4000, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.34212, "bpb": 1.93627, "chunk_loss": 0.035196, "F": 0.19156, "G": 0.4196, "chunks": 196.16, "bytes_per_chunk": 5.22, "boundary_entropy": 0.65986, "router_noise": 0.0, "grad_norm": 0.7575, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.101, "n_target_mean": 8.7169, "alpha": 0.03, "t": 309.2} |
| {"event": "train", "step": 4050, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.54665, "bpb": 2.23135, "chunk_loss": 0.034543, "F": 0.19855, "G": 0.42318, "chunks": 203.31, "bytes_per_chunk": 5.037, "boundary_entropy": 0.66152, "router_noise": 0.0, "grad_norm": 0.7373, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.052, "n_target_mean": 8.3032, "alpha": 0.03, "t": 311.8} |
| {"event": "train", "step": 4100, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.40605, "bpb": 2.02851, "chunk_loss": 0.035996, "F": 0.20111, "G": 0.42668, "chunks": 205.94, "bytes_per_chunk": 4.972, "boundary_entropy": 0.66396, "router_noise": 0.0, "grad_norm": 0.7264, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.049, "n_target_mean": 8.5769, "alpha": 0.03, "t": 314.2} |
| {"event": "train", "step": 4150, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.40243, "bpb": 2.02327, "chunk_loss": 0.035887, "F": 0.21118, "G": 0.42677, "chunks": 216.25, "bytes_per_chunk": 4.735, "boundary_entropy": 0.66261, "router_noise": 0.0, "grad_norm": 1.241, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.051, "n_target_mean": 8.3506, "alpha": 0.03, "t": 316.8} |
| {"event": "probe", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3995.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.229558, "cps": 31.210938, "mean_p": 0.433757, "boundary_entropy": 0.659548, "bpb": 2.079387}, "de": {"chunks": 5077.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.247261, "cps": 39.664062, "mean_p": 0.442259, "boundary_entropy": 0.667389, "bpb": 2.28513}, "vi": {"chunks": 4176.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226526, "cps": 32.625, "mean_p": 0.435889, "boundary_entropy": 0.669338, "bpb": 1.960784}, "tr": {"chunks": 5449.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.301, "cps": 42.570312, "mean_p": 0.446808, "boundary_entropy": 0.662328, "bpb": 2.263113}, "fi": {"chunks": 5265.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.279889, "cps": 41.132812, "mean_p": 0.451968, "boundary_entropy": 0.667365, "bpb": 2.448475}, "ru": {"chunks": 4955.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.261478, "cps": 38.710938, "mean_p": 0.401124, "boundary_entropy": 0.651148, "bpb": 1.45221}, "el": {"chunks": 5723.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.278858, "cps": 44.710938, "mean_p": 0.412938, "boundary_entropy": 0.654886, "bpb": 1.515338}, "hi": {"chunks": 3610.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.214562, "cps": 28.203125, "mean_p": 0.402285, "boundary_entropy": 0.659211, "bpb": 1.221334}, "ta": {"chunks": 4511.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.241294, "cps": 35.242188, "mean_p": 0.361232, "boundary_entropy": 0.627298, "bpb": 0.999038}, "th": {"chunks": 4570.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.284116, "cps": 35.703125, "mean_p": 0.371999, "boundary_entropy": 0.637718, "bpb": 1.130938}, "zh": {"chunks": 3486.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.590247, "cps": 27.234375, "mean_p": 0.435641, "boundary_entropy": 0.664923, "bpb": 3.215022}, "my": {"chunks": 3405.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.179702, "cps": 26.601562, "mean_p": 0.429156, "boundary_entropy": 0.676717, "bpb": 1.124672}}, "bpb": {"en": 2.07939, "de": 2.28513, "vi": 1.96078, "tr": 2.26311, "fi": 2.44848, "ru": 1.45221, "el": 1.51534, "hi": 1.22133, "ta": 0.99904, "th": 1.13094, "zh": 3.21502, "my": 1.12467}, "gini_cpc": 0.151838, "gini_cps": 0.096142, "bpic": 7.077957, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0771150467085953, "vi": 0.9867898746500661, "tr": 1.3112140465589166, "fi": 1.2192529881425729, "ru": 1.1390473580101776, "el": 1.2147593130500978, "hi": 0.9346725011018602, "ta": 1.0511257954497146, "th": 1.2376631809345144, "zh": 2.5712320726000626, "my": 0.7828184930339619}, "t": 320.4} |
| {"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 2.08721, "de": 2.2941, "vi": 1.96377, "tr": 2.26991, "fi": 2.45672, "ru": 1.43156, "el": 1.50734, "hi": 1.21606, "ta": 1.00442, "th": 1.14898, "zh": 3.21211, "my": 1.13354}, "bytes_per_chunk": {"en": 4.291, "de": 4.073, "vi": 5.599, "tr": 3.541, "fi": 3.775, "ru": 6.834, "el": 6.672, "hi": 11.529, "ta": 10.982, "th": 9.948, "zh": 4.761, "my": 14.955}, "bpb_mean": 1.81048, "frozen_inventory": false, "t": 321.6} |
| {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584, "t": 321.9} |
| {"event": "train", "step": 4200, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.20437, "bpb": 1.73755, "chunk_loss": 0.0, "F": 0.16559, "G": 0.41681, "chunks": 169.56, "bytes_per_chunk": 6.039, "boundary_entropy": 0.66146, "router_noise": 0.0, "grad_norm": 0.6761, "lr": 9.12e-05, "router_trainable": false, "sec_per_step": 0.103, "alpha": 0.0, "t": 321.9} |
| {"event": "train", "step": 4250, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.25585, "bpb": 1.81181, "chunk_loss": 0.0, "F": 0.19281, "G": 0.42011, "chunks": 197.44, "bytes_per_chunk": 5.186, "boundary_entropy": 0.66056, "router_noise": 0.0, "grad_norm": 0.7557, "lr": 8.81e-05, "router_trainable": false, "sec_per_step": 0.055, "alpha": 0.0, "t": 324.7} |
| {"event": "train", "step": 4300, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.48376, "bpb": 2.14062, "chunk_loss": 0.0, "F": 0.22049, "G": 0.43016, "chunks": 225.78, "bytes_per_chunk": 4.535, "boundary_entropy": 0.66347, "router_noise": 0.0, "grad_norm": 0.7835, "lr": 8.5e-05, "router_trainable": false, "sec_per_step": 0.053, "alpha": 0.0, "t": 327.3} |
| {"event": "train", "step": 4350, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.47556, "bpb": 2.12878, "chunk_loss": 0.0, "F": 0.21188, "G": 0.43179, "chunks": 216.97, "bytes_per_chunk": 4.72, "boundary_entropy": 0.66697, "router_noise": 0.0, "grad_norm": 0.8783, "lr": 8.21e-05, "router_trainable": false, "sec_per_step": 0.048, "alpha": 0.0, "t": 329.7} |
| {"event": "probe", "step": 4400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4187.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.240591, "cps": 32.710938, "mean_p": 0.437006, "boundary_entropy": 0.660399, "bpb": 2.069633}, "de": {"chunks": 5497.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.267715, "cps": 42.945312, "mean_p": 0.44599, "boundary_entropy": 0.667924, "bpb": 2.275156}, "vi": {"chunks": 4582.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.248549, "cps": 35.796875, "mean_p": 0.442781, "boundary_entropy": 0.670792, "bpb": 1.951393}, "tr": {"chunks": 5522.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.305032, "cps": 43.140625, "mean_p": 0.449641, "boundary_entropy": 0.663217, "bpb": 2.243963}, "fi": {"chunks": 5249.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.279039, "cps": 41.007812, "mean_p": 0.452165, "boundary_entropy": 0.667352, "bpb": 2.437332}, "ru": {"chunks": 5174.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.273034, "cps": 40.421875, "mean_p": 0.404838, "boundary_entropy": 0.653176, "bpb": 1.434533}, "el": {"chunks": 6185.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.301369, "cps": 48.320312, "mean_p": 0.41807, "boundary_entropy": 0.657943, "bpb": 1.497086}, "hi": {"chunks": 3685.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.219019, "cps": 28.789062, "mean_p": 0.406649, "boundary_entropy": 0.661833, "bpb": 1.212861}, "ta": {"chunks": 3716.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.19877, "cps": 29.03125, "mean_p": 0.367508, "boundary_entropy": 0.633895, "bpb": 0.992958}, "th": {"chunks": 4462.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.277401, "cps": 34.859375, "mean_p": 0.382292, "boundary_entropy": 0.645136, "bpb": 1.122896}, "zh": {"chunks": 3499.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.592448, "cps": 27.335938, "mean_p": 0.437534, "boundary_entropy": 0.666213, "bpb": 3.198113}, "my": {"chunks": 3619.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.190996, "cps": 28.273438, "mean_p": 0.434627, "boundary_entropy": 0.678525, "bpb": 1.112026}}, "bpb": {"en": 2.06963, "de": 2.27516, "vi": 1.95139, "tr": 2.24396, "fi": 2.43733, "ru": 1.43453, "el": 1.49709, "hi": 1.21286, "ta": 0.99296, "th": 1.1229, "zh": 3.19811, "my": 1.11203}, "gini_cpc": 0.153039, "gini_cps": 0.105494, "bpic": 6.930332, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1127420217294601, "vi": 1.0330779741160938, "tr": 1.267847475385687, "fi": 1.159807328583755, "ru": 1.134849757983153, "el": 1.2526219512599592, "hi": 0.910339900867718, "ta": 0.8261737558904575, "th": 1.1530009364064013, "zh": 2.4624740308944904, "my": 0.7938644720918345}, "t": 333.3} |
| {"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 2.07702, "de": 2.28486, "vi": 1.95333, "tr": 2.25033, "fi": 2.44482, "ru": 1.41323, "el": 1.48926, "hi": 1.20686, "ta": 0.99952, "th": 1.1416, "zh": 3.19507, "my": 1.12037}, "bytes_per_chunk": {"en": 4.118, "de": 3.773, "vi": 5.175, "tr": 3.515, "fi": 3.799, "ru": 6.589, "el": 6.228, "hi": 10.97, "ta": 13.094, "th": 10.248, "zh": 4.77, "my": 14.105}, "bpb_mean": 1.79802, "frozen_inventory": false, "t": 334.5} |
| {"event": "train", "step": 4400, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.4797, "bpb": 2.13476, "chunk_loss": 0.0, "F": 0.22397, "G": 0.42962, "chunks": 229.34, "bytes_per_chunk": 4.465, "boundary_entropy": 0.6642, "router_noise": 0.0, "grad_norm": 0.8025, "lr": 7.92e-05, "router_trainable": false, "sec_per_step": 0.1, "alpha": 0.0, "t": 334.7} |
| {"event": "train", "step": 4450, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.498, "bpb": 2.16115, "chunk_loss": 0.0, "F": 0.21426, "G": 0.4283, "chunks": 219.41, "bytes_per_chunk": 4.667, "boundary_entropy": 0.66467, "router_noise": 0.0, "grad_norm": 0.7366, "lr": 7.63e-05, "router_trainable": false, "sec_per_step": 0.051, "alpha": 0.0, "t": 337.3} |
| {"event": "train", "step": 4500, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.42087, "bpb": 2.04989, "chunk_loss": 0.0, "F": 0.22891, "G": 0.43728, "chunks": 234.41, "bytes_per_chunk": 4.368, "boundary_entropy": 0.66785, "router_noise": 0.0, "grad_norm": 0.7975, "lr": 7.36e-05, "router_trainable": false, "sec_per_step": 0.051, "alpha": 0.0, "t": 339.8} |
| {"event": "train", "step": 4550, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.22998, "bpb": 1.77449, "chunk_loss": 0.0, "F": 0.19751, "G": 0.42491, "chunks": 202.25, "bytes_per_chunk": 5.063, "boundary_entropy": 0.66258, "router_noise": 0.0, "grad_norm": 0.77, "lr": 7.09e-05, "router_trainable": false, "sec_per_step": 0.048, "alpha": 0.0, "t": 342.2} |
| {"event": "probe", "step": 4600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4276.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.245705, "cps": 33.40625, "mean_p": 0.439585, "boundary_entropy": 0.661803, "bpb": 2.057011}, "de": {"chunks": 5623.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.273852, "cps": 43.929688, "mean_p": 0.448476, "boundary_entropy": 0.669033, "bpb": 2.261999}, "vi": {"chunks": 4618.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.250502, "cps": 36.078125, "mean_p": 0.444329, "boundary_entropy": 0.671524, "bpb": 1.946576}, "tr": {"chunks": 5577.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.30807, "cps": 43.570312, "mean_p": 0.452261, "boundary_entropy": 0.664343, "bpb": 2.232789}, "fi": {"chunks": 5372.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.285578, "cps": 41.96875, "mean_p": 0.455848, "boundary_entropy": 0.668913, "bpb": 2.413796}, "ru": {"chunks": 5829.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.307599, "cps": 45.539062, "mean_p": 0.412229, "boundary_entropy": 0.656588, "bpb": 1.423944}, "el": {"chunks": 6584.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.320811, "cps": 51.4375, "mean_p": 0.423167, "boundary_entropy": 0.659876, "bpb": 1.487519}, "hi": {"chunks": 4657.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.27679, "cps": 36.382812, "mean_p": 0.415438, "boundary_entropy": 0.665555, "bpb": 1.206581}, "ta": {"chunks": 5434.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.290666, "cps": 42.453125, "mean_p": 0.376046, "boundary_entropy": 0.637857, "bpb": 0.987345}, "th": {"chunks": 5340.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.331986, "cps": 41.71875, "mean_p": 0.388521, "boundary_entropy": 0.648568, "bpb": 1.113371}, "zh": {"chunks": 3581.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.606333, "cps": 27.976562, "mean_p": 0.439713, "boundary_entropy": 0.667092, "bpb": 3.178445}, "my": {"chunks": 3960.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.208993, "cps": 30.9375, "mean_p": 0.43849, "boundary_entropy": 0.679401, "bpb": 1.105625}}, "bpb": {"en": 2.05701, "de": 2.262, "vi": 1.94658, "tr": 2.23279, "fi": 2.4138, "ru": 1.42394, "el": 1.48752, "hi": 1.20658, "ta": 0.98735, "th": 1.11337, "zh": 3.17844, "my": 1.10563}, "gini_cpc": 0.132817, "gini_cps": 0.090989, "bpic": 6.306897, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.114556528296392, "vi": 1.0195234285862167, "tr": 1.2538238213544397, "fi": 1.1622794098888958, "ru": 1.2519046849199436, "el": 1.305675938109913, "hi": 1.1265165817159808, "ta": 1.1829886782332009, "th": 1.3511594886031382, "zh": 2.467728075046638, "my": 0.8505860063473407}, "t": 345.6} |
| {"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 2.06557, "de": 2.27087, "vi": 1.94783, "tr": 2.23922, "fi": 2.42307, "ru": 1.40252, "el": 1.47833, "hi": 1.20159, "ta": 0.9934, "th": 1.13126, "zh": 3.17473, "my": 1.11425}, "bytes_per_chunk": {"en": 4.045, "de": 3.703, "vi": 5.133, "tr": 3.439, "fi": 3.73, "ru": 5.861, "el": 5.917, "hi": 8.959, "ta": 9.108, "th": 8.64, "zh": 4.611, "my": 12.989}, "bpb_mean": 1.78689, "frozen_inventory": false, "t": 346.9} |
| {"event": "train", "step": 4600, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.30921, "bpb": 1.88879, "chunk_loss": 0.0, "F": 0.20145, "G": 0.43004, "chunks": 206.28, "bytes_per_chunk": 4.964, "boundary_entropy": 0.66589, "router_noise": 0.0, "grad_norm": 0.7614, "lr": 6.82e-05, "router_trainable": false, "sec_per_step": 0.097, "alpha": 0.0, "t": 347.1} |
| {"event": "train", "step": 4650, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.40889, "bpb": 2.0326, "chunk_loss": 0.0, "F": 0.20349, "G": 0.42832, "chunks": 208.38, "bytes_per_chunk": 4.914, "boundary_entropy": 0.66427, "router_noise": 0.0, "grad_norm": 0.7495, "lr": 6.57e-05, "router_trainable": false, "sec_per_step": 0.05, "alpha": 0.0, "t": 349.6} |
| {"event": "train", "step": 4700, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.32815, "bpb": 1.91611, "chunk_loss": 0.0, "F": 0.23096, "G": 0.43425, "chunks": 236.5, "bytes_per_chunk": 4.33, "boundary_entropy": 0.66565, "router_noise": 0.0, "grad_norm": 0.8327, "lr": 6.32e-05, "router_trainable": false, "sec_per_step": 0.053, "alpha": 0.0, "t": 352.2} |
| {"event": "train", "step": 4750, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.34884, "bpb": 1.94596, "chunk_loss": 0.0, "F": 0.21249, "G": 0.4376, "chunks": 217.59, "bytes_per_chunk": 4.706, "boundary_entropy": 0.66863, "router_noise": 0.0, "grad_norm": 0.7714, "lr": 6.08e-05, "router_trainable": false, "sec_per_step": 0.055, "alpha": 0.0, "t": 355.0} |
| {"event": "probe", "step": 4800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4309.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.247601, "cps": 33.664062, "mean_p": 0.440391, "boundary_entropy": 0.66201, "bpb": 2.049923}, "de": {"chunks": 5536.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.269615, "cps": 43.25, "mean_p": 0.448551, "boundary_entropy": 0.669424, "bpb": 2.249904}, "vi": {"chunks": 4865.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.2639, "cps": 38.007812, "mean_p": 0.447845, "boundary_entropy": 0.672438, "bpb": 1.940187}, "tr": {"chunks": 5842.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.322709, "cps": 45.640625, "mean_p": 0.454192, "boundary_entropy": 0.664354, "bpb": 2.218142}, "fi": {"chunks": 5471.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.29084, "cps": 42.742188, "mean_p": 0.457228, "boundary_entropy": 0.669283, "bpb": 2.399281}, "ru": {"chunks": 5440.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.287071, "cps": 42.5, "mean_p": 0.411156, "boundary_entropy": 0.656451, "bpb": 1.418667}, "el": {"chunks": 6343.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.309068, "cps": 49.554688, "mean_p": 0.422333, "boundary_entropy": 0.659383, "bpb": 1.473966}, "hi": {"chunks": 3641.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.216404, "cps": 28.445312, "mean_p": 0.407716, "boundary_entropy": 0.662372, "bpb": 1.201549}, "ta": {"chunks": 4928.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.2636, "cps": 38.5, "mean_p": 0.375062, "boundary_entropy": 0.637953, "bpb": 0.979721}, "th": {"chunks": 5585.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.347218, "cps": 43.632812, "mean_p": 0.386748, "boundary_entropy": 0.646802, "bpb": 1.106743}, "zh": {"chunks": 3568.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.604131, "cps": 27.875, "mean_p": 0.439265, "boundary_entropy": 0.666945, "bpb": 3.168381}, "my": {"chunks": 3779.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.199441, "cps": 29.523438, "mean_p": 0.437649, "boundary_entropy": 0.679319, "bpb": 1.103277}}, "bpb": {"en": 2.04992, "de": 2.2499, "vi": 1.94019, "tr": 2.21814, "fi": 2.39928, "ru": 1.41867, "el": 1.47397, "hi": 1.20155, "ta": 0.97972, "th": 1.10674, "zh": 3.16838, "my": 1.10328}, "gini_cpc": 0.149112, "gini_cps": 0.100383, "bpic": 6.471091, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0889082805611516, "vi": 1.0658284994356808, "tr": 1.303342710538791, "fi": 1.1746337025564075, "ru": 1.1594107197857342, "el": 1.2482497632376344, "hi": 0.8740036214330075, "ta": 1.0646156737569679, "th": 1.402328428552394, "zh": 2.4399393389606505, "my": 0.8054918340348661}, "t": 358.8} |
| {"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 2.05785, "de": 2.25932, "vi": 1.94076, "tr": 2.2235, "fi": 2.40782, "ru": 1.39708, "el": 1.46473, "hi": 1.19689, "ta": 0.98569, "th": 1.12492, "zh": 3.16353, "my": 1.11123}, "bytes_per_chunk": {"en": 4.014, "de": 3.735, "vi": 4.835, "tr": 3.286, "fi": 3.652, "ru": 6.207, "el": 6.086, "hi": 11.452, "ta": 9.93, "th": 8.174, "zh": 4.638, "my": 13.446}, "bpb_mean": 1.77778, "frozen_inventory": false, "t": 360.0} |
| {"event": "train", "step": 4800, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.26063, "bpb": 1.8187, "chunk_loss": 0.0, "F": 0.19992, "G": 0.42493, "chunks": 204.72, "bytes_per_chunk": 5.002, "boundary_entropy": 0.66235, "router_noise": 0.0, "grad_norm": 0.7343, "lr": 5.85e-05, "router_trainable": false, "sec_per_step": 0.105, "alpha": 0.0, "t": 360.3} |
| {"event": "train", "step": 4850, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.4965, "bpb": 2.159, "chunk_loss": 0.0, "F": 0.22934, "G": 0.43597, "chunks": 234.84, "bytes_per_chunk": 4.36, "boundary_entropy": 0.66701, "router_noise": 0.0, "grad_norm": 0.7994, "lr": 5.62e-05, "router_trainable": false, "sec_per_step": 0.049, "alpha": 0.0, "t": 362.7} |
| {"event": "train", "step": 4900, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.40837, "bpb": 2.03185, "chunk_loss": 0.0, "F": 0.2009, "G": 0.42646, "chunks": 205.72, "bytes_per_chunk": 4.978, "boundary_entropy": 0.66479, "router_noise": 0.0, "grad_norm": 0.7254, "lr": 5.41e-05, "router_trainable": false, "sec_per_step": 0.051, "alpha": 0.0, "t": 365.3} |
| {"event": "train", "step": 4950, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.52603, "bpb": 2.20159, "chunk_loss": 0.0, "F": 0.23007, "G": 0.43691, "chunks": 235.59, "bytes_per_chunk": 4.346, "boundary_entropy": 0.66784, "router_noise": 0.0, "grad_norm": 0.7968, "lr": 5.2e-05, "router_trainable": false, "sec_per_step": 0.049, "alpha": 0.0, "t": 367.7} |
| {"event": "probe", "step": 5000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4269.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.245303, "cps": 33.351562, "mean_p": 0.439888, "boundary_entropy": 0.662017, "bpb": 2.041342}, "de": {"chunks": 5557.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.270638, "cps": 43.414062, "mean_p": 0.44866, "boundary_entropy": 0.669442, "bpb": 2.236525}, "vi": {"chunks": 4767.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.258584, "cps": 37.242188, "mean_p": 0.446658, "boundary_entropy": 0.672336, "bpb": 1.934893}, "tr": {"chunks": 5799.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.320334, "cps": 45.304688, "mean_p": 0.455415, "boundary_entropy": 0.665026, "bpb": 2.209792}, "fi": {"chunks": 5216.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.277285, "cps": 40.75, "mean_p": 0.454596, "boundary_entropy": 0.668606, "bpb": 2.389947}, "ru": {"chunks": 5583.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.294617, "cps": 43.617188, "mean_p": 0.411453, "boundary_entropy": 0.656862, "bpb": 1.404517}, "el": {"chunks": 6379.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.310822, "cps": 49.835938, "mean_p": 0.422492, "boundary_entropy": 0.660107, "bpb": 1.469135}, "hi": {"chunks": 2974.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.176761, "cps": 23.234375, "mean_p": 0.406704, "boundary_entropy": 0.662765, "bpb": 1.196147}, "ta": {"chunks": 3679.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.196791, "cps": 28.742188, "mean_p": 0.371893, "boundary_entropy": 0.637388, "bpb": 0.97989}, "th": {"chunks": 4602.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.286105, "cps": 35.953125, "mean_p": 0.384846, "boundary_entropy": 0.646845, "bpb": 1.102167}, "zh": {"chunks": 3622.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.613275, "cps": 28.296875, "mean_p": 0.438922, "boundary_entropy": 0.666849, "bpb": 3.168381}, "my": {"chunks": 3768.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.19886, "cps": 29.4375, "mean_p": 0.438206, "boundary_entropy": 0.679326, "bpb": 1.096833}}, "bpb": {"en": 2.04134, "de": 2.23653, "vi": 1.93489, "tr": 2.20979, "fi": 2.38995, "ru": 1.40452, "el": 1.46913, "hi": 1.19615, "ta": 0.97989, "th": 1.10217, "zh": 3.16838, "my": 1.09683}, "gini_cpc": 0.169912, "gini_cps": 0.122042, "bpic": 6.827021, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1032805323511414, "vi": 1.054144082997735, "tr": 1.3058717375380955, "fi": 1.1303779172881943, "ru": 1.2010369782521226, "el": 1.2670965907026686, "hi": 0.7205827422340888, "ta": 0.8022362528114796, "th": 1.166335553545367, "zh": 2.5000746052170344, "my": 0.8106725793609643}, "t": 371.2} |
| {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.04866, "de": 2.24579, "vi": 1.93627, "tr": 2.21706, "fi": 2.39804, "ru": 1.38223, "el": 1.45971, "hi": 1.19103, "ta": 0.98551, "th": 1.12091, "zh": 3.16359, "my": 1.10607}, "bytes_per_chunk": {"en": 4.036, "de": 3.739, "vi": 4.962, "tr": 3.291, "fi": 3.826, "ru": 6.071, "el": 6.096, "hi": 13.621, "ta": 13.246, "th": 9.965, "zh": 4.627, "my": 13.146}, "bpb_mean": 1.77124, "frozen_inventory": false, "t": 372.4} |
| {"event": "train", "step": 5000, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.29104, "bpb": 1.86258, "chunk_loss": 0.0, "F": 0.20447, "G": 0.42534, "chunks": 209.38, "bytes_per_chunk": 4.891, "boundary_entropy": 0.66303, "router_noise": 0.0, "grad_norm": 0.7899, "lr": 5e-05, "router_trainable": false, "sec_per_step": 0.101, "alpha": 0.0, "t": 372.8} |
| {"event": "train", "step": 5050, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.26597, "bpb": 1.82641, "chunk_loss": 0.0, "F": 0.21442, "G": 0.42984, "chunks": 219.56, "bytes_per_chunk": 4.664, "boundary_entropy": 0.66414, "router_noise": 0.0, "grad_norm": 0.7606, "lr": 4.81e-05, "router_trainable": false, "sec_per_step": 0.05, "alpha": 0.0, "t": 375.3} |
| {"event": "train", "step": 5100, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.30601, "bpb": 1.88417, "chunk_loss": 0.0, "F": 0.20837, "G": 0.43041, "chunks": 213.38, "bytes_per_chunk": 4.799, "boundary_entropy": 0.66452, "router_noise": 0.0, "grad_norm": 0.7836, "lr": 4.63e-05, "router_trainable": false, "sec_per_step": 0.05, "alpha": 0.0, "t": 377.8} |
| {"event": "train", "step": 5150, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.34381, "bpb": 1.9387, "chunk_loss": 0.0, "F": 0.21609, "G": 0.42979, "chunks": 221.28, "bytes_per_chunk": 4.628, "boundary_entropy": 0.66505, "router_noise": 0.0, "grad_norm": 0.7728, "lr": 4.45e-05, "router_trainable": false, "sec_per_step": 0.052, "alpha": 0.0, "t": 380.4} |
| {"event": "probe", "step": 5200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4463.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.25645, "cps": 34.867188, "mean_p": 0.442404, "boundary_entropy": 0.662398, "bpb": 2.035102}, "de": {"chunks": 5804.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.282667, "cps": 45.34375, "mean_p": 0.451366, "boundary_entropy": 0.669842, "bpb": 2.227145}, "vi": {"chunks": 4894.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.265473, "cps": 38.234375, "mean_p": 0.448997, "boundary_entropy": 0.672816, "bpb": 1.930746}, "tr": {"chunks": 5922.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.327128, "cps": 46.265625, "mean_p": 0.456327, "boundary_entropy": 0.664941, "bpb": 2.201446}, "fi": {"chunks": 5352.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.284514, "cps": 41.8125, "mean_p": 0.455758, "boundary_entropy": 0.668588, "bpb": 2.37739}, "ru": {"chunks": 5553.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.293034, "cps": 43.382812, "mean_p": 0.414083, "boundary_entropy": 0.658311, "bpb": 1.399701}, "el": {"chunks": 6380.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.310871, "cps": 49.84375, "mean_p": 0.42427, "boundary_entropy": 0.660764, "bpb": 1.46281}, "hi": {"chunks": 3806.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.226211, "cps": 29.734375, "mean_p": 0.409234, "boundary_entropy": 0.662926, "bpb": 1.189979}, "ta": {"chunks": 3962.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.211928, "cps": 30.953125, "mean_p": 0.374489, "boundary_entropy": 0.638776, "bpb": 0.97714}, "th": {"chunks": 4996.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.3106, "cps": 39.03125, "mean_p": 0.388871, "boundary_entropy": 0.649081, "bpb": 1.093892}, "zh": {"chunks": 3552.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.601422, "cps": 27.75, "mean_p": 0.439583, "boundary_entropy": 0.667141, "bpb": 3.145537}, "my": {"chunks": 4278.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.225776, "cps": 33.421875, "mean_p": 0.444496, "boundary_entropy": 0.680983, "bpb": 1.097013}}, "bpb": {"en": 2.0351, "de": 2.22714, "vi": 1.93075, "tr": 2.20145, "fi": 2.37739, "ru": 1.3997, "el": 1.46281, "hi": 1.18998, "ta": 0.97714, "th": 1.09389, "zh": 3.14554, "my": 1.09701}, "gini_cpc": 0.142389, "gini_cps": 0.102029, "bpic": 6.508955, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.102230006796577, "vi": 1.0351852051047334, "tr": 1.2756016871565303, "fi": 1.1094339576794023, "ru": 1.1426564948858307, "el": 1.212207779778597, "hi": 0.8820860307917269, "ta": 0.8263922488414294, "th": 1.211151852579051, "zh": 2.3451830564723806, "my": 0.8803890605991675}, "t": 383.9} |
| {"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 2.04165, "de": 2.23663, "vi": 1.93307, "tr": 2.20667, "fi": 2.38667, "ru": 1.37776, "el": 1.4542, "hi": 1.1847, "ta": 0.98329, "th": 1.11287, "zh": 3.14249, "my": 1.10515}, "bytes_per_chunk": {"en": 3.919, "de": 3.563, "vi": 4.855, "tr": 3.251, "fi": 3.718, "ru": 6.092, "el": 6.076, "hi": 10.946, "ta": 12.22, "th": 9.069, "zh": 4.587, "my": 11.825}, "bpb_mean": 1.76376, "frozen_inventory": false, "t": 385.2} |
| {"event": "train", "step": 5200, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.29313, "bpb": 1.86559, "chunk_loss": 0.0, "F": 0.20596, "G": 0.42916, "chunks": 210.91, "bytes_per_chunk": 4.855, "boundary_entropy": 0.6652, "router_noise": 0.0, "grad_norm": 0.8568, "lr": 4.29e-05, "router_trainable": false, "sec_per_step": 0.102, "alpha": 0.0, "t": 385.4} |
| {"event": "train", "step": 5250, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.37962, "bpb": 1.99036, "chunk_loss": 0.0, "F": 0.21643, "G": 0.43279, "chunks": 221.62, "bytes_per_chunk": 4.62, "boundary_entropy": 0.66614, "router_noise": 0.0, "grad_norm": 0.7508, "lr": 4.14e-05, "router_trainable": false, "sec_per_step": 0.061, "alpha": 0.0, "t": 388.5} |
| {"event": "train", "step": 5300, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.37919, "bpb": 1.98975, "chunk_loss": 0.0, "F": 0.22089, "G": 0.43198, "chunks": 226.19, "bytes_per_chunk": 4.527, "boundary_entropy": 0.66571, "router_noise": 0.0, "grad_norm": 0.7901, "lr": 3.99e-05, "router_trainable": false, "sec_per_step": 0.052, "alpha": 0.0, "t": 391.1} |
| {"event": "train", "step": 5350, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.43023, "bpb": 2.06339, "chunk_loss": 0.0, "F": 0.22455, "G": 0.43317, "chunks": 229.94, "bytes_per_chunk": 4.453, "boundary_entropy": 0.66619, "router_noise": 0.0, "grad_norm": 0.8297, "lr": 3.86e-05, "router_trainable": false, "sec_per_step": 0.048, "alpha": 0.0, "t": 393.5} |
| {"event": "probe", "step": 5400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4446.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.255473, "cps": 34.734375, "mean_p": 0.442627, "boundary_entropy": 0.662616, "bpb": 2.027142}, "de": {"chunks": 5840.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.28442, "cps": 45.625, "mean_p": 0.451417, "boundary_entropy": 0.669953, "bpb": 2.220942}, "vi": {"chunks": 5028.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.272742, "cps": 39.28125, "mean_p": 0.449502, "boundary_entropy": 0.672924, "bpb": 1.922867}, "tr": {"chunks": 5882.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.324919, "cps": 45.953125, "mean_p": 0.456679, "boundary_entropy": 0.665722, "bpb": 2.192997}, "fi": {"chunks": 5424.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.288342, "cps": 42.375, "mean_p": 0.45772, "boundary_entropy": 0.669377, "bpb": 2.368064}, "ru": {"chunks": 5747.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.303272, "cps": 44.898438, "mean_p": 0.413323, "boundary_entropy": 0.657728, "bpb": 1.393721}, "el": {"chunks": 6335.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.308678, "cps": 49.492188, "mean_p": 0.423943, "boundary_entropy": 0.660242, "bpb": 1.451928}, "hi": {"chunks": 3722.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.221218, "cps": 29.078125, "mean_p": 0.410705, "boundary_entropy": 0.663876, "bpb": 1.183024}, "ta": {"chunks": 4503.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.240867, "cps": 35.179688, "mean_p": 0.378281, "boundary_entropy": 0.640853, "bpb": 0.967752}, "th": {"chunks": 5167.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.321231, "cps": 40.367188, "mean_p": 0.385787, "boundary_entropy": 0.646875, "bpb": 1.087075}, "zh": {"chunks": 3691.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.624958, "cps": 28.835938, "mean_p": 0.439387, "boundary_entropy": 0.666867, "bpb": 3.140092}, "my": {"chunks": 4575.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.24145, "cps": 35.742188, "mean_p": 0.444778, "boundary_entropy": 0.680999, "bpb": 1.092071}}, "bpb": {"en": 2.02714, "de": 2.22094, "vi": 1.92287, "tr": 2.193, "fi": 2.36806, "ru": 1.39372, "el": 1.45193, "hi": 1.18302, "ta": 0.96775, "th": 1.08708, "zh": 3.14009, "my": 1.09207}, "gini_cpc": 0.138248, "gini_cps": 0.09276, "bpic": 6.358201, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1133074154090385, "vi": 1.067595633635632, "tr": 1.2718301922504773, "fi": 1.128658246951295, "ru": 1.1870981950512571, "el": 1.2082601048703265, "hi": 0.8659163854315179, "ta": 0.9428251074948052, "th": 1.2573959527594, "zh": 2.446274928901874, "my": 0.945110056160521}, "t": 397.0} |
| {"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 2.03534, "de": 2.2289, "vi": 1.92366, "tr": 2.19909, "fi": 2.37673, "ru": 1.37171, "el": 1.44333, "hi": 1.17776, "ta": 0.97435, "th": 1.10688, "zh": 3.13636, "my": 1.10044}, "bytes_per_chunk": {"en": 3.869, "de": 3.555, "vi": 4.688, "tr": 3.257, "fi": 3.67, "ru": 5.899, "el": 6.082, "hi": 11.079, "ta": 10.863, "th": 8.911, "zh": 4.557, "my": 11.018}, "bpb_mean": 1.75621, "frozen_inventory": false, "t": 398.2} |
| {"event": "train", "step": 5400, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.13769, "bpb": 1.64134, "chunk_loss": 0.0, "F": 0.18716, "G": 0.42725, "chunks": 191.66, "bytes_per_chunk": 5.343, "boundary_entropy": 0.66545, "router_noise": 0.0, "grad_norm": 0.7289, "lr": 3.73e-05, "router_trainable": false, "sec_per_step": 0.1, "alpha": 0.0, "t": 398.4} |
| {"event": "train", "step": 5450, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.19886, "bpb": 1.72959, "chunk_loss": 0.0, "F": 0.19458, "G": 0.42737, "chunks": 199.25, "bytes_per_chunk": 5.139, "boundary_entropy": 0.6647, "router_noise": 0.0, "grad_norm": 0.7639, "lr": 3.62e-05, "router_trainable": false, "sec_per_step": 0.053, "alpha": 0.0, "t": 401.1} |
| {"event": "train", "step": 5500, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.39836, "bpb": 2.0174, "chunk_loss": 0.0, "F": 0.23071, "G": 0.43657, "chunks": 236.25, "bytes_per_chunk": 4.334, "boundary_entropy": 0.66661, "router_noise": 0.0, "grad_norm": 0.843, "lr": 3.51e-05, "router_trainable": false, "sec_per_step": 0.05, "alpha": 0.0, "t": 403.6} |
| {"event": "train", "step": 5550, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.41168, "bpb": 2.03663, "chunk_loss": 0.0, "F": 0.23294, "G": 0.43724, "chunks": 238.53, "bytes_per_chunk": 4.293, "boundary_entropy": 0.66761, "router_noise": 0.0, "grad_norm": 0.8276, "lr": 3.41e-05, "router_trainable": false, "sec_per_step": 0.053, "alpha": 0.0, "t": 406.2} |
| {"event": "probe", "step": 5600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4427.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.254381, "cps": 34.585938, "mean_p": 0.442309, "boundary_entropy": 0.662256, "bpb": 2.021522}, "de": {"chunks": 5822.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.283544, "cps": 45.484375, "mean_p": 0.451277, "boundary_entropy": 0.669721, "bpb": 2.215395}, "vi": {"chunks": 4993.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.270844, "cps": 39.007812, "mean_p": 0.44972, "boundary_entropy": 0.672998, "bpb": 1.921788}, "tr": {"chunks": 5873.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.324421, "cps": 45.882812, "mean_p": 0.455882, "boundary_entropy": 0.665238, "bpb": 2.182123}, "fi": {"chunks": 5377.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.285843, "cps": 42.007812, "mean_p": 0.456643, "boundary_entropy": 0.66913, "bpb": 2.35869}, "ru": {"chunks": 5743.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.303061, "cps": 44.867188, "mean_p": 0.414924, "boundary_entropy": 0.658786, "bpb": 1.389192}, "el": {"chunks": 6341.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.30897, "cps": 49.539062, "mean_p": 0.424951, "boundary_entropy": 0.66115, "bpb": 1.444934}, "hi": {"chunks": 3895.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.231501, "cps": 30.429688, "mean_p": 0.412691, "boundary_entropy": 0.664641, "bpb": 1.177871}, "ta": {"chunks": 4806.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.257074, "cps": 37.546875, "mean_p": 0.380075, "boundary_entropy": 0.641793, "bpb": 0.963182}, "th": {"chunks": 4911.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.305316, "cps": 38.367188, "mean_p": 0.388292, "boundary_entropy": 0.649102, "bpb": 1.084236}, "zh": {"chunks": 3603.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.610058, "cps": 28.148438, "mean_p": 0.440553, "boundary_entropy": 0.667495, "bpb": 3.130509}, "my": {"chunks": 4404.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.232426, "cps": 34.40625, "mean_p": 0.443015, "boundary_entropy": 0.680506, "bpb": 1.089353}}, "bpb": {"en": 2.02152, "de": 2.21539, "vi": 1.92179, "tr": 2.18212, "fi": 2.35869, "ru": 1.38919, "el": 1.44493, "hi": 1.17787, "ta": 0.96318, "th": 1.08424, "zh": 3.13051, "my": 1.08935}, "gini_cpc": 0.129845, "gini_cps": 0.091676, "bpic": 6.375629, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.114639404703152, "vi": 1.064714141328555, "tr": 1.275334321970404, "fi": 1.1236802592219008, "ru": 1.1913632524810276, "el": 1.214595048318634, "hi": 0.9100536321257071, "ta": 1.0105851103452137, "th": 1.2002272066758515, "zh": 2.3982001052403628, "my": 0.913689286179731}, "t": 409.8} |
| {"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 2.02995, "de": 2.22423, "vi": 1.92271, "tr": 2.18809, "fi": 2.36809, "ru": 1.36633, "el": 1.43622, "hi": 1.17289, "ta": 0.96979, "th": 1.10356, "zh": 3.12686, "my": 1.09756}, "bytes_per_chunk": {"en": 3.905, "de": 3.574, "vi": 4.702, "tr": 3.28, "fi": 3.698, "ru": 5.919, "el": 6.086, "hi": 10.733, "ta": 10.159, "th": 9.229, "zh": 4.59, "my": 11.408}, "bpb_mean": 1.75052, "frozen_inventory": false, "t": 411.0} |
| {"event": "train", "step": 5600, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.47343, "bpb": 2.12571, "chunk_loss": 0.0, "F": 0.22772, "G": 0.43467, "chunks": 233.19, "bytes_per_chunk": 4.391, "boundary_entropy": 0.66685, "router_noise": 0.0, "grad_norm": 0.8231, "lr": 3.33e-05, "router_trainable": false, "sec_per_step": 0.103, "alpha": 0.0, "t": 411.4} |
| {"event": "train", "step": 5650, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.2783, "bpb": 1.8442, "chunk_loss": 0.0, "F": 0.20569, "G": 0.42779, "chunks": 210.62, "bytes_per_chunk": 4.862, "boundary_entropy": 0.6644, "router_noise": 0.0, "grad_norm": 0.7598, "lr": 3.25e-05, "router_trainable": false, "sec_per_step": 0.051, "alpha": 0.0, "t": 413.9} |
| {"event": "train", "step": 5700, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.32687, "bpb": 1.91426, "chunk_loss": 0.0, "F": 0.22644, "G": 0.43457, "chunks": 231.88, "bytes_per_chunk": 4.416, "boundary_entropy": 0.66635, "router_noise": 0.0, "grad_norm": 0.806, "lr": 3.18e-05, "router_trainable": false, "sec_per_step": 0.055, "alpha": 0.0, "t": 416.7} |
| {"event": "train", "step": 5750, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.20996, "bpb": 1.7456, "chunk_loss": 0.0, "F": 0.21201, "G": 0.43015, "chunks": 217.09, "bytes_per_chunk": 4.717, "boundary_entropy": 0.66466, "router_noise": 0.0, "grad_norm": 0.775, "lr": 3.13e-05, "router_trainable": false, "sec_per_step": 0.052, "alpha": 0.0, "t": 419.3} |
| {"event": "probe", "step": 5800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4567.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.262426, "cps": 35.679688, "mean_p": 0.444253, "boundary_entropy": 0.662821, "bpb": 2.017582}, "de": {"chunks": 5884.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.286563, "cps": 45.96875, "mean_p": 0.452091, "boundary_entropy": 0.669769, "bpb": 2.210798}, "vi": {"chunks": 4885.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.264985, "cps": 38.164062, "mean_p": 0.44964, "boundary_entropy": 0.673094, "bpb": 1.915787}, "tr": {"chunks": 5899.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.325858, "cps": 46.085938, "mean_p": 0.456679, "boundary_entropy": 0.665464, "bpb": 2.175182}, "fi": {"chunks": 5485.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.291585, "cps": 42.851562, "mean_p": 0.457622, "boundary_entropy": 0.668917, "bpb": 2.352667}, "ru": {"chunks": 5900.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.311346, "cps": 46.09375, "mean_p": 0.41522, "boundary_entropy": 0.658394, "bpb": 1.385192}, "el": {"chunks": 6436.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.313599, "cps": 50.28125, "mean_p": 0.425608, "boundary_entropy": 0.661421, "bpb": 1.440938}, "hi": {"chunks": 3902.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.231917, "cps": 30.484375, "mean_p": 0.412834, "boundary_entropy": 0.665085, "bpb": 1.17475}, "ta": {"chunks": 5109.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.273282, "cps": 39.914062, "mean_p": 0.379121, "boundary_entropy": 0.640786, "bpb": 0.960477}, "th": {"chunks": 5182.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.322164, "cps": 40.484375, "mean_p": 0.390809, "boundary_entropy": 0.650639, "bpb": 1.081482}, "zh": {"chunks": 3679.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.622926, "cps": 28.742188, "mean_p": 0.441153, "boundary_entropy": 0.667711, "bpb": 3.129099}, "my": {"chunks": 4403.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.232373, "cps": 34.398438, "mean_p": 0.443823, "boundary_entropy": 0.680814, "bpb": 1.088052}}, "bpb": {"en": 2.01758, "de": 2.2108, "vi": 1.91579, "tr": 2.17518, "fi": 2.35267, "ru": 1.38519, "el": 1.44094, "hi": 1.17475, "ta": 0.96048, "th": 1.08148, "zh": 3.1291, "my": 1.08805}, "gini_cpc": 0.132007, "gini_cps": 0.091361, "bpic": 6.257537, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.091976684293297, "vi": 1.00975156440328, "tr": 1.2417122148314124, "fi": 1.1111120487866832, "ru": 1.1864130249530156, "el": 1.195001082967887, "hi": 0.8837416038763778, "ta": 1.041366354078086, "th": 1.227635537858292, "zh": 2.37371980752985, "my": 0.8854793100211016}, "t": 422.9} |
| {"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 2.02557, "de": 2.22017, "vi": 1.91657, "tr": 2.18145, "fi": 2.36314, "ru": 1.36278, "el": 1.4324, "hi": 1.16964, "ta": 0.96705, "th": 1.10028, "zh": 3.12522, "my": 1.09627}, "bytes_per_chunk": {"en": 3.788, "de": 3.509, "vi": 4.773, "tr": 3.258, "fi": 3.63, "ru": 5.711, "el": 5.983, "hi": 10.575, "ta": 9.587, "th": 8.812, "zh": 4.491, "my": 11.591}, "bpb_mean": 1.74671, "frozen_inventory": false, "t": 424.1} |
| {"event": "train", "step": 5800, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.26309, "bpb": 1.82225, "chunk_loss": 0.0, "F": 0.21286, "G": 0.42841, "chunks": 217.97, "bytes_per_chunk": 4.698, "boundary_entropy": 0.66392, "router_noise": 0.0, "grad_norm": 0.7601, "lr": 3.08e-05, "router_trainable": false, "sec_per_step": 0.102, "alpha": 0.0, "t": 424.4} |
| {"event": "train", "step": 5850, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.37528, "bpb": 1.98412, "chunk_loss": 0.0, "F": 0.22516, "G": 0.43638, "chunks": 230.56, "bytes_per_chunk": 4.441, "boundary_entropy": 0.66792, "router_noise": 0.0, "grad_norm": 0.7471, "lr": 3.05e-05, "router_trainable": false, "sec_per_step": 0.05, "alpha": 0.0, "t": 426.9} |
| {"event": "train", "step": 5900, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.32354, "bpb": 1.90947, "chunk_loss": 0.0, "F": 0.22244, "G": 0.43383, "chunks": 227.78, "bytes_per_chunk": 4.496, "boundary_entropy": 0.66581, "router_noise": 0.0, "grad_norm": 0.7821, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.048, "alpha": 0.0, "t": 429.3} |
| {"event": "train", "step": 5950, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.28804, "bpb": 1.85825, "chunk_loss": 0.0, "F": 0.21338, "G": 0.43065, "chunks": 218.5, "bytes_per_chunk": 4.686, "boundary_entropy": 0.66523, "router_noise": 0.0, "grad_norm": 0.7816, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.05, "alpha": 0.0, "t": 431.8} |
| {"event": "train", "step": 5999, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.34911, "bpb": 1.94636, "chunk_loss": 0.0, "F": 0.21832, "G": 0.43033, "chunks": 223.56, "bytes_per_chunk": 4.58, "boundary_entropy": 0.66475, "router_noise": 0.0, "grad_norm": 0.837, "lr": 3e-05, "router_trainable": false, "sec_per_step": 0.049, "alpha": 0.0, "t": 434.2} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4544.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.261104, "cps": 35.5, "mean_p": 0.44422, "boundary_entropy": 0.662641, "bpb": 2.014128}, "de": {"chunks": 6037.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.294015, "cps": 47.164062, "mean_p": 0.453599, "boundary_entropy": 0.670051, "bpb": 2.204613}, "vi": {"chunks": 5105.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276919, "cps": 39.882812, "mean_p": 0.450809, "boundary_entropy": 0.673336, "bpb": 1.915194}, "tr": {"chunks": 5989.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.330829, "cps": 46.789062, "mean_p": 0.457722, "boundary_entropy": 0.665343, "bpb": 2.171662}, "fi": {"chunks": 5622.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.298868, "cps": 43.921875, "mean_p": 0.459332, "boundary_entropy": 0.669446, "bpb": 2.352162}, "ru": {"chunks": 5798.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.305963, "cps": 45.296875, "mean_p": 0.414723, "boundary_entropy": 0.658467, "bpb": 1.381736}, "el": {"chunks": 6311.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307509, "cps": 49.304688, "mean_p": 0.425384, "boundary_entropy": 0.661103, "bpb": 1.435054}, "hi": {"chunks": 3509.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.208559, "cps": 27.414062, "mean_p": 0.409166, "boundary_entropy": 0.663293, "bpb": 1.176524}, "ta": {"chunks": 4759.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.25456, "cps": 37.179688, "mean_p": 0.380081, "boundary_entropy": 0.642033, "bpb": 0.960608}, "th": {"chunks": 4915.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.305564, "cps": 38.398438, "mean_p": 0.389172, "boundary_entropy": 0.649809, "bpb": 1.078642}, "zh": {"chunks": 3712.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.628513, "cps": 29.0, "mean_p": 0.439657, "boundary_entropy": 0.666801, "bpb": 3.120519}, "my": {"chunks": 4756.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.251003, "cps": 37.15625, "mean_p": 0.443974, "boundary_entropy": 0.680665, "bpb": 1.085661}}, "bpb": {"en": 2.01413, "de": 2.20461, "vi": 1.91519, "tr": 2.17166, "fi": 2.35216, "ru": 1.38174, "el": 1.43505, "hi": 1.17652, "ta": 0.96061, "th": 1.07864, "zh": 3.12052, "my": 1.08566}, "gini_cpc": 0.134815, "gini_cps": 0.095336, "bpic": 6.285618, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1260419396155827, "vi": 1.0605677136455838, "tr": 1.2670377638559638, "fi": 1.1446290275543005, "ru": 1.171803509049017, "el": 1.177722934780147, "hi": 0.7987559252244523, "ta": 0.9749358329032769, "th": 1.170275880883686, "zh": 2.407134306005352, "my": 0.9613117869662561}, "t": 435.2} |
| {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.02313, "de": 2.21417, "vi": 1.9161, "tr": 2.17762, "fi": 2.36062, "ru": 1.35985, "el": 1.42679, "hi": 1.1715, "ta": 0.96741, "th": 1.09794, "zh": 3.11605, "my": 1.09413}, "bytes_per_chunk": {"en": 3.811, "de": 3.427, "vi": 4.57, "tr": 3.225, "fi": 3.557, "ru": 5.827, "el": 6.081, "hi": 11.772, "ta": 10.368, "th": 9.188, "zh": 4.493, "my": 10.773}, "bpb_mean": 1.74378, "frozen_inventory": false, "t": 436.5} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s0/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s0/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s0/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s0/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s0/probe/reps/th0.70"], "t": 441.6} |
| {"event": "final", "run": "parity_B_ste_s0", "arch": "hnet", "objective": "parity", "condition": "B", "step": 6000, "params": 22516992, "seed": 0, "per_lang": {"en": {"cpc": 0.26110440728610007, "cps": 35.5, "bpb": 2.0320761994873697, "bytes_per_chunk": 3.8016953208275024}, "de": {"cpc": 0.29401451322261724, "cps": 47.1640625, "bpb": 2.206460846043454, "bytes_per_chunk": 3.425196644628531}, "vi": {"cpc": 0.2769189042582045, "cps": 39.8828125, "bpb": 1.9191335373327487, "bytes_per_chunk": 4.586144156752974}, "tr": {"cpc": 0.33082914434071703, "cps": 46.7890625, "bpb": 2.1688446609108367, "bytes_per_chunk": 3.227639023128967}, "fi": {"cpc": 0.2988676838020307, "cps": 43.921875, "bpb": 2.378471777303414, "bytes_per_chunk": 3.554856426077228}, "ru": {"cpc": 0.3059630606860158, "cps": 45.296875, "bpb": 1.3591301434361727, "bytes_per_chunk": 5.815406799401032}, "el": {"cpc": 0.3075086488330166, "cps": 49.3046875, "bpb": 1.4236893295078978, "bytes_per_chunk": 6.0818746012412275}, "hi": {"cpc": 0.20855869242199107, "cps": 27.4140625, "bpb": 1.1790276918015519, "bytes_per_chunk": 11.712530438085025}, "ta": {"cpc": 0.2545600427921904, "cps": 37.1796875, "bpb": 0.9676912290931334, "bytes_per_chunk": 10.479262857028642}, "th": {"cpc": 0.30556419023935344, "cps": 38.3984375, "bpb": 1.0954840449550585, "bytes_per_chunk": 9.267951210889164}, "zh": {"cpc": 0.6285133762275652, "cps": 29.0, "bpb": 3.125566476443265, "bytes_per_chunk": 4.495888179050723}, "my": {"cpc": 0.251002744352966, "cps": 37.15625, "bpb": 1.1037527612046865, "bytes_per_chunk": 10.623655042451015}}, "bpb_mean": 1.746610724793299, "gini_cpc": 0.13481524300909165, "gini_cps": 0.09533577913534352, "bpic": 6.2856183566175865, "mean_boundary_entropy": 0.6647466421127319, "dechunk": "ste", "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": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584}], "probe_overhead_sec_total": 0.0, "probe_overhead_seq_frac": 0.0, "wall_sec": 445.2848217487335, "t": 445.3} |
|
|
