{"event": "start", "run": "pilot_B_s0", "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}, "t": 2.0}
{"event": "probe_schedule", "n": 40, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 200, 256, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 2.0}
{"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": 8.160238}, "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": 8.156329}, "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": 8.087291}, "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": 8.171434}, "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": 8.214758}, "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": 8.099397}, "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": 8.158044}, "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": 8.262162}, "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": 8.051803}, "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": 8.248459}, "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": 8.123118}, "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": 7.913994}}, "bpb": {"en": 8.16024, "de": 8.15633, "vi": 8.08729, "tr": 8.17143, "fi": 8.21476, "ru": 8.0994, "el": 8.15804, "hi": 8.26216, "ta": 8.0518, "th": 8.24846, "zh": 8.12312, "my": 7.91399}, "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": 3.9}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.63916, "bpb": 8.13559, "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": 7.8061, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.006, "alpha": 0.03, "t": 4.3}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6329.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.363673, "cps": 49.445312, "mean_p": 0.496549, "boundary_entropy": 0.686989, "bpb": 8.138007}, "de": {"chunks": 8593.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.418497, "cps": 67.132812, "mean_p": 0.498781, "boundary_entropy": 0.687819, "bpb": 8.136762}, "vi": {"chunks": 13022.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.706374, "cps": 101.734375, "mean_p": 0.505542, "boundary_entropy": 0.688387, "bpb": 8.072257}, "tr": {"chunks": 5241.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.28951, "cps": 40.945312, "mean_p": 0.489422, "boundary_entropy": 0.68728, "bpb": 8.156677}, "fi": {"chunks": 8838.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.469831, "cps": 69.046875, "mean_p": 0.500506, "boundary_entropy": 0.687509, "bpb": 8.196361}, "ru": {"chunks": 10500.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.55409, "cps": 82.03125, "mean_p": 0.493306, "boundary_entropy": 0.689882, "bpb": 8.059413}, "el": {"chunks": 15046.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.733129, "cps": 117.546875, "mean_p": 0.497468, "boundary_entropy": 0.690234, "bpb": 8.140788}, "hi": {"chunks": 2501.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.148648, "cps": 19.539062, "mean_p": 0.484439, "boundary_entropy": 0.690243, "bpb": 8.249415}, "ta": {"chunks": 37852.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.024712, "cps": 295.71875, "mean_p": 0.512695, "boundary_entropy": 0.690391, "bpb": 8.028644}, "th": {"chunks": 33856.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.104818, "cps": 264.5, "mean_p": 0.511769, "boundary_entropy": 0.690558, "bpb": 8.229671}, "zh": {"chunks": 11820.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.001355, "cps": 92.34375, "mean_p": 0.520021, "boundary_entropy": 0.685675, "bpb": 8.120637}, "my": {"chunks": 48157.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.541535, "cps": 376.226562, "mean_p": 0.521508, "boundary_entropy": 0.69021, "bpb": 7.893431}}, "bpb": {"en": 8.13801, "de": 8.13676, "vi": 8.07226, "tr": 8.15668, "fi": 8.19636, "ru": 8.05941, "el": 8.14079, "hi": 8.24942, "ta": 8.02864, "th": 8.22967, "zh": 8.12064, "my": 7.89343}, "gini_cpc": 0.425855, "gini_cps": 0.427309, "bpic": 1.902213, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1507511806716928, "vi": 1.9423324845233707, "tr": 0.7960725204339101, "fi": 1.291906663601748, "ru": 1.5235934932844624, "el": 2.01590116971762, "hi": 0.4087404731680415, "ta": 5.567399504338443, "th": 5.787667929592518, "zh": 5.5031716409026075, "my": 6.9885177512042596}, "t": 5.6}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6249.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.359076, "cps": 48.820312, "mean_p": 0.496201, "boundary_entropy": 0.686976, "bpb": 8.088591}, "de": {"chunks": 8517.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.414796, "cps": 66.539062, "mean_p": 0.498576, "boundary_entropy": 0.687814, "bpb": 8.092818}, "vi": {"chunks": 12838.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.696393, "cps": 100.296875, "mean_p": 0.505085, "boundary_entropy": 0.68839, "bpb": 8.038697}, "tr": {"chunks": 5170.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.285588, "cps": 40.390625, "mean_p": 0.489117, "boundary_entropy": 0.687262, "bpb": 8.12198}, "fi": {"chunks": 8750.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.465153, "cps": 68.359375, "mean_p": 0.50014, "boundary_entropy": 0.687503, "bpb": 8.154229}, "ru": {"chunks": 10060.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.530871, "cps": 78.59375, "mean_p": 0.492759, "boundary_entropy": 0.689864, "bpb": 7.993239}, "el": {"chunks": 15091.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.735321, "cps": 117.898438, "mean_p": 0.497513, "boundary_entropy": 0.690235, "bpb": 8.120892}, "hi": {"chunks": 2407.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.143061, "cps": 18.804688, "mean_p": 0.484164, "boundary_entropy": 0.690224, "bpb": 8.217173}, "ta": {"chunks": 37473.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.00444, "cps": 292.757812, "mean_p": 0.512238, "boundary_entropy": 0.690413, "bpb": 7.987914}, "th": {"chunks": 33649.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.091949, "cps": 262.882812, "mean_p": 0.511607, "boundary_entropy": 0.690564, "bpb": 8.202697}, "zh": {"chunks": 11810.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.999661, "cps": 92.265625, "mean_p": 0.519978, "boundary_entropy": 0.685678, "bpb": 8.11415}, "my": {"chunks": 48089.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.537946, "cps": 375.695312, "mean_p": 0.521422, "boundary_entropy": 0.690216, "bpb": 7.860119}}, "bpb": {"en": 8.08859, "de": 8.09282, "vi": 8.0387, "tr": 8.12198, "fi": 8.15423, "ru": 7.99324, "el": 8.12089, "hi": 8.21717, "ta": 7.98791, "th": 8.2027, "zh": 8.11415, "my": 7.86012}, "gini_cpc": 0.428202, "gini_cps": 0.429742, "bpic": 1.917917, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1551751440996156, "vi": 1.939401936531163, "tr": 0.7953413959258948, "fi": 1.2954175161247852, "ru": 1.4784354309354406, "el": 2.0478152341240556, "hi": 0.3984140202402667, "ta": 5.582215381794599, "th": 5.825922357089102, "zh": 5.568908092683926, "my": 7.06799063784341}, "t": 7.2}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5919.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.340114, "cps": 46.242188, "mean_p": 0.494973, "boundary_entropy": 0.686928, "bpb": 7.918697}, "de": {"chunks": 8279.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.403205, "cps": 64.679688, "mean_p": 0.497821, "boundary_entropy": 0.687798, "bpb": 7.939057}, "vi": {"chunks": 12273.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.665745, "cps": 95.882812, "mean_p": 0.503598, "boundary_entropy": 0.688396, "bpb": 7.926934}, "tr": {"chunks": 4899.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.270618, "cps": 38.273438, "mean_p": 0.48808, "boundary_entropy": 0.687201, "bpb": 8.003685}, "fi": {"chunks": 8331.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.442879, "cps": 65.085938, "mean_p": 0.498782, "boundary_entropy": 0.687477, "bpb": 8.007523}, "ru": {"chunks": 8696.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.458892, "cps": 67.9375, "mean_p": 0.490852, "boundary_entropy": 0.689793, "bpb": 7.771387}, "el": {"chunks": 15246.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.742874, "cps": 119.109375, "mean_p": 0.497599, "boundary_entropy": 0.690236, "bpb": 8.033879}, "hi": {"chunks": 2072.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.12315, "cps": 16.1875, "mean_p": 0.482797, "boundary_entropy": 0.690125, "bpb": 8.096838}, "ta": {"chunks": 35903.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.92046, "cps": 280.492188, "mean_p": 0.510441, "boundary_entropy": 0.690491, "bpb": 7.841708}, "th": {"chunks": 32624.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.028225, "cps": 254.875, "mean_p": 0.510752, "boundary_entropy": 0.690595, "bpb": 8.088119}, "zh": {"chunks": 11776.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.993905, "cps": 92.0, "mean_p": 0.519759, "boundary_entropy": 0.68569, "bpb": 8.087292}, "my": {"chunks": 47875.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.526652, "cps": 374.023438, "mean_p": 0.521127, "boundary_entropy": 0.690237, "bpb": 7.783285}}, "bpb": {"en": 7.9187, "de": 7.93906, "vi": 7.92693, "tr": 8.00368, "fi": 8.00752, "ru": 7.77139, "el": 8.03388, "hi": 8.09684, "ta": 7.84171, "th": 8.08812, "zh": 8.08729, "my": 7.78329}, "gini_cpc": 0.437247, "gini_cps": 0.438191, "bpic": 1.979344, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1854991738294076, "vi": 1.9574170751854676, "tr": 0.7956694220790258, "fi": 1.3021500479211365, "ru": 1.3492303351177573, "el": 2.1841921932049635, "hi": 0.36208493714175544, "ta": 5.646522327972154, "th": 5.963372296282688, "zh": 5.862463267658942, "my": 7.428843188191524}, "t": 8.8}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5087.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.292306, "cps": 39.742188, "mean_p": 0.49168, "boundary_entropy": 0.686769, "bpb": 7.437776}, "de": {"chunks": 7673.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.373691, "cps": 59.945312, "mean_p": 0.495826, "boundary_entropy": 0.687749, "bpb": 7.496227}, "vi": {"chunks": 10391.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.563656, "cps": 81.179688, "mean_p": 0.498905, "boundary_entropy": 0.688358, "bpb": 7.585993}, "tr": {"chunks": 4092.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.22604, "cps": 31.96875, "mean_p": 0.48508, "boundary_entropy": 0.687, "bpb": 7.643273}, "fi": {"chunks": 7124.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.378715, "cps": 55.65625, "mean_p": 0.494859, "boundary_entropy": 0.687364, "bpb": 7.577047}, "ru": {"chunks": 5217.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.275303, "cps": 40.757812, "mean_p": 0.485327, "boundary_entropy": 0.689504, "bpb": 7.189153}, "el": {"chunks": 14857.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.72392, "cps": 116.070312, "mean_p": 0.497132, "boundary_entropy": 0.69023, "bpb": 7.711154}, "hi": {"chunks": 1134.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.0674, "cps": 8.859375, "mean_p": 0.477107, "boundary_entropy": 0.689636, "bpb": 7.695903}, "ta": {"chunks": 28681.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.534154, "cps": 224.070312, "mean_p": 0.503574, "boundary_entropy": 0.690676, "bpb": 7.366186}, "th": {"chunks": 27408.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.703948, "cps": 214.125, "mean_p": 0.506243, "boundary_entropy": 0.690718, "bpb": 7.547635}, "zh": {"chunks": 11598.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.963766, "cps": 90.609375, "mean_p": 0.518919, "boundary_entropy": 0.685739, "bpb": 7.998385}, "my": {"chunks": 47100.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.48575, "cps": 367.96875, "mean_p": 0.520001, "boundary_entropy": 0.690311, "bpb": 7.472415}}, "bpb": {"en": 7.43778, "de": 7.49623, "vi": 7.58599, "tr": 7.64327, "fi": 7.57705, "ru": 7.18915, "el": 7.71115, "hi": 7.6959, "ta": 7.36619, "th": 7.54763, "zh": 7.99838, "my": 7.47241}, "gini_cpc": 0.469739, "gini_cps": 0.463481, "bpic": 2.252738, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2784247609349235, "vi": 1.928308809945356, "tr": 0.7732990135591435, "fi": 1.2956103535863672, "ru": 0.9418332206934434, "el": 2.476581707208474, "hi": 0.23057932538466638, "ta": 5.248451672116723, "th": 5.829330287135028, "zh": 6.718186321300932, "my": 8.503934640485445}, "t": 10.5}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3942.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.226513, "cps": 30.796875, "mean_p": 0.488114, "boundary_entropy": 0.686664, "bpb": 6.812404}, "de": {"chunks": 6926.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.337311, "cps": 54.109375, "mean_p": 0.4939, "boundary_entropy": 0.687798, "bpb": 6.904711}, "vi": {"chunks": 7190.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.390019, "cps": 56.171875, "mean_p": 0.490799, "boundary_entropy": 0.688163, "bpb": 7.158126}, "tr": {"chunks": 3500.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193338, "cps": 27.34375, "mean_p": 0.483515, "boundary_entropy": 0.686947, "bpb": 7.13155}, "fi": {"chunks": 5536.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.294296, "cps": 43.25, "mean_p": 0.490073, "boundary_entropy": 0.687262, "bpb": 7.002122}, "ru": {"chunks": 2627.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.138628, "cps": 20.523438, "mean_p": 0.478881, "boundary_entropy": 0.689015, "bpb": 6.36065}, "el": {"chunks": 5602.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.272962, "cps": 43.765625, "mean_p": 0.487001, "boundary_entropy": 0.68987, "bpb": 7.040936}, "hi": {"chunks": 404.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.024012, "cps": 3.15625, "mean_p": 0.467006, "boundary_entropy": 0.688448, "bpb": 7.04958}, "ta": {"chunks": 14608.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.781385, "cps": 114.125, "mean_p": 0.490177, "boundary_entropy": 0.690532, "bpb": 6.622467}, "th": {"chunks": 17488.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.087224, "cps": 136.625, "mean_p": 0.496602, "boundary_entropy": 0.690725, "bpb": 6.369985}, "zh": {"chunks": 10875.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.841348, "cps": 84.960938, "mean_p": 0.515366, "boundary_entropy": 0.685938, "bpb": 7.770743}, "my": {"chunks": 42531.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.244617, "cps": 332.273438, "mean_p": 0.515706, "boundary_entropy": 0.690552, "bpb": 6.672967}}, "bpb": {"en": 6.8124, "de": 6.90471, "vi": 7.15813, "tr": 7.13155, "fi": 7.00212, "ru": 6.36065, "el": 7.04094, "hi": 7.04958, "ta": 6.62247, "th": 6.36999, "zh": 7.77074, "my": 6.67297}, "gini_cpc": 0.523271, "gini_cps": 0.49208, "bpic": 3.165752, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.48914703220068, "vi": 1.7218418079235984, "tr": 0.8535421906323118, "fi": 1.2992469269130478, "ru": 0.6120097749826306, "el": 1.2050630205810595, "hi": 0.10600681652158377, "ta": 3.4496321833746317, "th": 4.79983799913795, "zh": 8.129116044824007, "my": 9.909453823634594}, "t": 12.3}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4867.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.279664, "cps": 38.023438, "mean_p": 0.495232, "boundary_entropy": 0.687434, "bpb": 6.360912}, "de": {"chunks": 8528.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.415331, "cps": 66.625, "mean_p": 0.49897, "boundary_entropy": 0.68833, "bpb": 6.444318}, "vi": {"chunks": 5665.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.307296, "cps": 44.257812, "mean_p": 0.489065, "boundary_entropy": 0.688362, "bpb": 6.904357}, "tr": {"chunks": 6808.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.37607, "cps": 53.1875, "mean_p": 0.496493, "boundary_entropy": 0.687979, "bpb": 6.71869}, "fi": {"chunks": 7152.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.380203, "cps": 55.875, "mean_p": 0.497707, "boundary_entropy": 0.688012, "bpb": 6.521184}, "ru": {"chunks": 5316.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.280528, "cps": 41.53125, "mean_p": 0.487921, "boundary_entropy": 0.689825, "bpb": 5.791064}, "el": {"chunks": 485.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.023632, "cps": 3.789062, "mean_p": 0.469255, "boundary_entropy": 0.688346, "bpb": 6.574903}, "hi": {"chunks": 2652.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.157623, "cps": 20.71875, "mean_p": 0.485627, "boundary_entropy": 0.690382, "bpb": 5.947555}, "ta": {"chunks": 5459.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.292003, "cps": 42.648438, "mean_p": 0.482358, "boundary_entropy": 0.690233, "bpb": 5.809846}, "th": {"chunks": 3722.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.231396, "cps": 29.078125, "mean_p": 0.487888, "boundary_entropy": 0.690626, "bpb": 5.624242}, "zh": {"chunks": 7412.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.254995, "cps": 57.90625, "mean_p": 0.501045, "boundary_entropy": 0.686375, "bpb": 7.331292}, "my": {"chunks": 25416.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.341355, "cps": 198.5625, "mean_p": 0.496457, "boundary_entropy": 0.690915, "bpb": 5.630121}}, "bpb": {"en": 6.36091, "de": 6.44432, "vi": 6.90436, "tr": 6.71869, "fi": 6.52118, "ru": 5.79106, "el": 6.5749, "hi": 5.94756, "ta": 5.80985, "th": 5.62424, "zh": 7.33129, "my": 5.63012}, "gini_cpc": 0.412968, "gini_cps": 0.368405, "bpic": 4.597171, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.485106361730681, "vi": 1.0988022655692689, "tr": 1.34471970716706, "fi": 1.3594974468215109, "ru": 1.003086838126351, "el": 0.0845013539060212, "hi": 0.5636132842373862, "ta": 1.0441199616677106, "th": 0.8274048790139756, "zh": 4.487502897075761, "my": 4.79630287236436}, "t": 14.6}
{"event": "train", "step": 50, "phase": 1, "objective": "global", "lm_loss": 4.22406, "bpb": 6.09403, "chunk_loss": 0.030799, "F": 0.17557, "G": 0.4874, "chunks": 179.78, "bytes_per_chunk": 5.696, "boundary_entropy": 0.69167, "router_noise": 0.0, "grad_norm": 1.5882, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 15.8}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4813.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.276562, "cps": 37.601562, "mean_p": 0.496143, "boundary_entropy": 0.687593, "bpb": 5.819841}, "de": {"chunks": 6263.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.305021, "cps": 48.929688, "mean_p": 0.495948, "boundary_entropy": 0.688416, "bpb": 5.934262}, "vi": {"chunks": 2898.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.157201, "cps": 22.640625, "mean_p": 0.480991, "boundary_entropy": 0.687911, "bpb": 6.312732}, "tr": {"chunks": 5113.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.282439, "cps": 39.945312, "mean_p": 0.494035, "boundary_entropy": 0.688061, "bpb": 6.226768}, "fi": {"chunks": 5481.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.291372, "cps": 42.820312, "mean_p": 0.495207, "boundary_entropy": 0.688105, "bpb": 5.973342}, "ru": {"chunks": 7500.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.395778, "cps": 58.59375, "mean_p": 0.493466, "boundary_entropy": 0.690198, "bpb": 5.415808}, "el": {"chunks": 4211.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.205184, "cps": 32.898438, "mean_p": 0.490326, "boundary_entropy": 0.690334, "bpb": 5.660888}, "hi": {"chunks": 432.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.025676, "cps": 3.375, "mean_p": 0.473833, "boundary_entropy": 0.689308, "bpb": 5.442111}, "ta": {"chunks": 2740.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.146563, "cps": 21.40625, "mean_p": 0.483453, "boundary_entropy": 0.690377, "bpb": 5.206809}, "th": {"chunks": 1448.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.090022, "cps": 11.3125, "mean_p": 0.483885, "boundary_entropy": 0.690415, "bpb": 5.097859}, "zh": {"chunks": 4678.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.792076, "cps": 36.546875, "mean_p": 0.493144, "boundary_entropy": 0.686649, "bpb": 6.993413}, "my": {"chunks": 22022.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.162233, "cps": 172.046875, "mean_p": 0.494073, "boundary_entropy": 0.690997, "bpb": 4.996166}}, "bpb": {"en": 5.81984, "de": 5.93426, "vi": 6.31273, "tr": 6.22677, "fi": 5.97334, "ru": 5.41581, "el": 5.66089, "hi": 5.44211, "ta": 5.20681, "th": 5.09786, "zh": 6.99341, "my": 4.99617}, "gini_cpc": 0.43519, "gini_cps": 0.409739, "bpic": 5.677318, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1029053998899412, "vi": 0.5684123399857688, "tr": 1.0212533633321226, "fi": 1.053552488280996, "ru": 1.4310681219016, "el": 0.7419124427693075, "hi": 0.09284038492826356, "ta": 0.5299481151359314, "th": 0.3255035692693495, "zh": 2.864013319271819, "my": 4.202441153478277}, "t": 17.6}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.38692, "bpb": 4.88629, "chunk_loss": 0.03055, "F": 0.17398, "G": 0.47917, "chunks": 178.16, "bytes_per_chunk": 5.748, "boundary_entropy": 0.69024, "router_noise": 0.0, "grad_norm": 2.1107, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.077, "alpha": 0.03, "t": 19.6}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3041.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17474, "cps": 23.757812, "mean_p": 0.449608, "boundary_entropy": 0.675944, "bpb": 4.506768}, "de": {"chunks": 3468.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.168899, "cps": 27.09375, "mean_p": 0.451452, "boundary_entropy": 0.67813, "bpb": 4.725655}, "vi": {"chunks": 6349.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.344399, "cps": 49.601562, "mean_p": 0.46628, "boundary_entropy": 0.682715, "bpb": 5.058324}, "tr": {"chunks": 3468.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.19157, "cps": 27.09375, "mean_p": 0.457071, "boundary_entropy": 0.679975, "bpb": 5.045065}, "fi": {"chunks": 4100.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.217958, "cps": 32.03125, "mean_p": 0.454248, "boundary_entropy": 0.678173, "bpb": 4.760448}, "ru": {"chunks": 5358.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.282744, "cps": 41.859375, "mean_p": 0.462153, "boundary_entropy": 0.684323, "bpb": 3.685568}, "el": {"chunks": 8650.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.421478, "cps": 67.578125, "mean_p": 0.480302, "boundary_entropy": 0.688059, "bpb": 4.661783}, "hi": {"chunks": 9548.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.567489, "cps": 74.59375, "mean_p": 0.490744, "boundary_entropy": 0.689721, "bpb": 3.707132}, "ta": {"chunks": 4443.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.237657, "cps": 34.710938, "mean_p": 0.4807, "boundary_entropy": 0.689278, "bpb": 3.522447}, "th": {"chunks": 2326.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.144607, "cps": 18.171875, "mean_p": 0.458887, "boundary_entropy": 0.685161, "bpb": 3.264181}, "zh": {"chunks": 3167.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.536234, "cps": 24.742188, "mean_p": 0.482388, "boundary_entropy": 0.685378, "bpb": 6.340058}, "my": {"chunks": 4419.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.233217, "cps": 34.523438, "mean_p": 0.471731, "boundary_entropy": 0.688819, "bpb": 4.051642}}, "bpb": {"en": 4.50677, "de": 4.72566, "vi": 5.05832, "tr": 5.04506, "fi": 4.76045, "ru": 3.68557, "el": 4.66178, "hi": 3.70713, "ta": 3.52245, "th": 3.26418, "zh": 6.34006, "my": 4.05164}, "gini_cpc": 0.252008, "gini_cps": 0.235544, "bpic": 6.578689, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9665723817065727, "vi": 1.9709240327940174, "tr": 1.0963172244147965, "fi": 1.2473251332317943, "ru": 1.6180844899825517, "el": 2.412031428778674, "hi": 3.2476187302507147, "ta": 1.3600614900276589, "th": 0.8275540057306133, "zh": 3.068755732832016, "my": 1.3346528726363511}, "t": 21.8}
{"event": "train", "step": 150, "phase": 1, "objective": "global", "lm_loss": 2.89879, "bpb": 4.18207, "chunk_loss": 0.032574, "F": 0.20758, "G": 0.43579, "chunks": 212.56, "bytes_per_chunk": 4.817, "boundary_entropy": 0.67133, "router_noise": 0.0, "grad_norm": 1.8225, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.068, "alpha": 0.03, "t": 23.0}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2141.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.123025, "cps": 16.726562, "mean_p": 0.353537, "boundary_entropy": 0.609553, "bpb": 3.882107}, "de": {"chunks": 2542.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.123801, "cps": 19.859375, "mean_p": 0.357638, "boundary_entropy": 0.617001, "bpb": 4.107755}, "vi": {"chunks": 4404.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.238893, "cps": 34.40625, "mean_p": 0.392837, "boundary_entropy": 0.613356, "bpb": 3.645944}, "tr": {"chunks": 3077.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.169972, "cps": 24.039062, "mean_p": 0.385157, "boundary_entropy": 0.634279, "bpb": 4.17754}, "fi": {"chunks": 3131.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.166445, "cps": 24.460938, "mean_p": 0.358926, "boundary_entropy": 0.615895, "bpb": 4.127333}, "ru": {"chunks": 4516.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.238311, "cps": 35.28125, "mean_p": 0.459818, "boundary_entropy": 0.6787, "bpb": 2.842053}, "el": {"chunks": 4461.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.217366, "cps": 34.851562, "mean_p": 0.393344, "boundary_entropy": 0.652353, "bpb": 3.085802}, "hi": {"chunks": 10341.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.614621, "cps": 80.789062, "mean_p": 0.405144, "boundary_entropy": 0.647673, "bpb": 2.394829}, "ta": {"chunks": 8461.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.452581, "cps": 66.101562, "mean_p": 0.477205, "boundary_entropy": 0.681258, "bpb": 2.260396}, "th": {"chunks": 12121.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.753559, "cps": 94.695312, "mean_p": 0.453474, "boundary_entropy": 0.680611, "bpb": 2.308649}, "zh": {"chunks": 4055.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.68659, "cps": 31.679688, "mean_p": 0.471351, "boundary_entropy": 0.680784, "bpb": 5.310439}, "my": {"chunks": 9357.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.493825, "cps": 73.101562, "mean_p": 0.446802, "boundary_entropy": 0.681032, "bpb": 2.418897}}, "bpb": {"en": 3.88211, "de": 4.10775, "vi": 3.64594, "tr": 4.17754, "fi": 4.12733, "ru": 2.84205, "el": 3.0858, "hi": 2.39483, "ta": 2.2604, "th": 2.30865, "zh": 5.31044, "my": 2.4189}, "gini_cpc": 0.340106, "gini_cps": 0.306117, "bpic": 5.593904, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.006307227656199, "vi": 1.9418318550337879, "tr": 1.3816065952771213, "fi": 1.3529403118128458, "ru": 1.9371005830382815, "el": 1.7668465425480526, "hi": 4.9959135896974995, "ta": 3.678778828608758, "th": 6.125264384661298, "zh": 5.580908070434946, "my": 4.0140308533386735}, "t": 26.2}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.88548, "de": 4.108, "vi": 3.64981, "tr": 4.18745, "fi": 4.13024, "ru": 2.82626, "el": 3.07818, "hi": 2.39058, "ta": 2.27857, "th": 2.34111, "zh": 5.31998, "my": 2.43695}, "bytes_per_chunk": {"en": 6.916, "de": 7.738, "vi": 5.397, "tr": 7.071, "fi": 6.292, "ru": 7.503, "el": 7.622, "hi": 4.21, "ta": 5.878, "th": 3.229, "zh": 5.31, "my": 5.834}, "bpb_mean": 3.38605, "t": 27.5}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.49356, "bpb": 3.59745, "chunk_loss": 0.029979, "F": 0.16238, "G": 0.39498, "chunks": 166.28, "bytes_per_chunk": 6.158, "boundary_entropy": 0.64199, "router_noise": 0.0, "grad_norm": 1.2602, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.09, "alpha": 0.03, "t": 27.5}
{"event": "train", "step": 250, "phase": 1, "objective": "global", "lm_loss": 2.56871, "bpb": 3.70586, "chunk_loss": 0.031576, "F": 0.19559, "G": 0.39769, "chunks": 200.28, "bytes_per_chunk": 5.113, "boundary_entropy": 0.63843, "router_noise": 0.0, "grad_norm": 1.1057, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.043, "alpha": 0.03, "t": 29.7}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3079.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.176924, "cps": 24.054688, "mean_p": 0.374766, "boundary_entropy": 0.615729, "bpb": 3.630169}, "de": {"chunks": 3742.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182243, "cps": 29.234375, "mean_p": 0.381047, "boundary_entropy": 0.62577, "bpb": 3.90177}, "vi": {"chunks": 4093.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222023, "cps": 31.976562, "mean_p": 0.416546, "boundary_entropy": 0.636167, "bpb": 3.198279}, "tr": {"chunks": 3847.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212506, "cps": 30.054688, "mean_p": 0.392113, "boundary_entropy": 0.627631, "bpb": 3.862419}, "fi": {"chunks": 2621.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.139333, "cps": 20.476562, "mean_p": 0.356335, "boundary_entropy": 0.60893, "bpb": 3.882719}, "ru": {"chunks": 3984.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.210237, "cps": 31.125, "mean_p": 0.443114, "boundary_entropy": 0.670671, "bpb": 2.644537}, "el": {"chunks": 5144.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.250646, "cps": 40.1875, "mean_p": 0.400605, "boundary_entropy": 0.65472, "bpb": 2.776612}, "hi": {"chunks": 8984.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.533967, "cps": 70.1875, "mean_p": 0.335429, "boundary_entropy": 0.576792, "bpb": 2.123784}, "ta": {"chunks": 9102.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.486868, "cps": 71.109375, "mean_p": 0.392048, "boundary_entropy": 0.639116, "bpb": 1.962437}, "th": {"chunks": 4298.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.267205, "cps": 33.578125, "mean_p": 0.424869, "boundary_entropy": 0.669126, "bpb": 2.074395}, "zh": {"chunks": 4727.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.800373, "cps": 36.929688, "mean_p": 0.471205, "boundary_entropy": 0.67585, "bpb": 4.964646}, "my": {"chunks": 12069.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.636954, "cps": 94.289062, "mean_p": 0.412172, "boundary_entropy": 0.66075, "bpb": 2.15975}}, "bpb": {"en": 3.63017, "de": 3.90177, "vi": 3.19828, "tr": 3.86242, "fi": 3.88272, "ru": 2.64454, "el": 2.77661, "hi": 2.12378, "ta": 1.96244, "th": 2.0744, "zh": 4.96465, "my": 2.15975}, "gini_cpc": 0.312547, "gini_cps": 0.258472, "bpic": 5.842305, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0300677904928177, "vi": 1.2549113115937223, "tr": 1.201119080835428, "fi": 0.7875344646855426, "ru": 1.1882957578832176, "el": 1.4166890791195115, "hi": 3.018068563006746, "ta": 2.7518565620786872, "th": 1.5102880190624204, "zh": 4.523833277590116, "my": 3.600164478115967}, "t": 30.9}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.36283, "bpb": 3.40884, "chunk_loss": 0.030214, "F": 0.1665, "G": 0.36528, "chunks": 170.5, "bytes_per_chunk": 6.006, "boundary_entropy": 0.60681, "router_noise": 0.0, "grad_norm": 1.6009, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.064, "alpha": 0.03, "t": 32.9}
{"event": "train", "step": 350, "phase": 1, "objective": "global", "lm_loss": 2.05709, "bpb": 2.96776, "chunk_loss": 0.032784, "F": 0.2149, "G": 0.37733, "chunks": 220.06, "bytes_per_chunk": 4.653, "boundary_entropy": 0.60275, "router_noise": 0.0, "grad_norm": 0.9298, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.044, "alpha": 0.03, "t": 35.1}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3478.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199851, "cps": 27.171875, "mean_p": 0.327037, "boundary_entropy": 0.545817, "bpb": 3.259406}, "de": {"chunks": 4889.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.238105, "cps": 38.195312, "mean_p": 0.347505, "boundary_entropy": 0.55682, "bpb": 3.549075}, "vi": {"chunks": 5880.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.318959, "cps": 45.9375, "mean_p": 0.362692, "boundary_entropy": 0.576165, "bpb": 2.803122}, "tr": {"chunks": 4416.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.243937, "cps": 34.5, "mean_p": 0.345501, "boundary_entropy": 0.556652, "bpb": 3.441417}, "fi": {"chunks": 3762.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.199989, "cps": 29.390625, "mean_p": 0.341134, "boundary_entropy": 0.573498, "bpb": 3.572055}, "ru": {"chunks": 6028.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.3181, "cps": 47.09375, "mean_p": 0.308115, "boundary_entropy": 0.493165, "bpb": 2.504723}, "el": {"chunks": 5083.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.247673, "cps": 39.710938, "mean_p": 0.351197, "boundary_entropy": 0.592182, "bpb": 2.553749}, "hi": {"chunks": 6111.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.36321, "cps": 47.742188, "mean_p": 0.357095, "boundary_entropy": 0.604039, "bpb": 1.801231}, "ta": {"chunks": 5237.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.280128, "cps": 40.914062, "mean_p": 0.363827, "boundary_entropy": 0.617317, "bpb": 1.573517}, "th": {"chunks": 4320.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.268573, "cps": 33.75, "mean_p": 0.349852, "boundary_entropy": 0.615988, "bpb": 1.90794}, "zh": {"chunks": 2803.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.474602, "cps": 21.898438, "mean_p": 0.343453, "boundary_entropy": 0.582683, "bpb": 4.152922}, "my": {"chunks": 6335.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.334336, "cps": 49.492188, "mean_p": 0.328839, "boundary_entropy": 0.576599, "bpb": 1.916809}}, "bpb": {"en": 3.25941, "de": 3.54907, "vi": 2.80312, "tr": 3.44142, "fi": 3.57206, "ru": 2.50472, "el": 2.55375, "hi": 1.80123, "ta": 1.57352, "th": 1.90794, "zh": 4.15292, "my": 1.91681}, "gini_cpc": 0.137703, "gini_cps": 0.126935, "bpic": 6.578126, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1914125557907078, "vi": 1.595984711058374, "tr": 1.2205991293129783, "fi": 1.0006943559414556, "ru": 1.59169030817153, "el": 1.239292459067159, "hi": 1.817405145653383, "ta": 1.4016889411931892, "th": 1.3438698918893892, "zh": 2.374784456192833, "my": 1.6729300632474604}, "t": 38.2}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 3.2678, "de": 3.54756, "vi": 2.80995, "tr": 3.44829, "fi": 3.56654, "ru": 2.48791, "el": 2.54775, "hi": 1.79785, "ta": 1.58924, "th": 1.93709, "zh": 4.16108, "my": 1.93184}, "bytes_per_chunk": {"en": 4.947, "de": 4.311, "vi": 4.102, "tr": 4.392, "fi": 5.323, "ru": 5.749, "el": 7.041, "hi": 6.094, "ta": 7.28, "th": 7.865, "zh": 4.134, "my": 7.409}, "bpb_mean": 2.75774, "t": 39.5}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.05591, "bpb": 2.96605, "chunk_loss": 0.030982, "F": 0.19235, "G": 0.33756, "chunks": 196.97, "bytes_per_chunk": 5.199, "boundary_entropy": 0.55949, "router_noise": 0.0, "grad_norm": 1.1088, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 39.5}
{"event": "train", "step": 450, "phase": 1, "objective": "global", "lm_loss": 2.12577, "bpb": 3.06683, "chunk_loss": 0.030558, "F": 0.17996, "G": 0.31689, "chunks": 184.28, "bytes_per_chunk": 5.557, "boundary_entropy": 0.54191, "router_noise": 0.0, "grad_norm": 0.9219, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 41.6}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.9561, "bpb": 2.82206, "chunk_loss": 0.031012, "F": 0.19492, "G": 0.32568, "chunks": 199.59, "bytes_per_chunk": 5.13, "boundary_entropy": 0.54119, "router_noise": 0.0, "grad_norm": 0.8449, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 43.7}
{"event": "train", "step": 550, "phase": 1, "objective": "global", "lm_loss": 1.91221, "bpb": 2.75874, "chunk_loss": 0.030906, "F": 0.19272, "G": 0.31654, "chunks": 197.34, "bytes_per_chunk": 5.189, "boundary_entropy": 0.52476, "router_noise": 0.0, "grad_norm": 0.941, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 45.7}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3190.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.183302, "cps": 24.921875, "mean_p": 0.287417, "boundary_entropy": 0.476116, "bpb": 2.943914}, "de": {"chunks": 3894.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189646, "cps": 30.421875, "mean_p": 0.291087, "boundary_entropy": 0.467947, "bpb": 3.280041}, "vi": {"chunks": 5265.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.285598, "cps": 41.132812, "mean_p": 0.313482, "boundary_entropy": 0.510108, "bpb": 2.580668}, "tr": {"chunks": 4446.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.245595, "cps": 34.734375, "mean_p": 0.335779, "boundary_entropy": 0.519941, "bpb": 3.163078}, "fi": {"chunks": 3675.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195364, "cps": 28.710938, "mean_p": 0.31324, "boundary_entropy": 0.520439, "bpb": 3.375743}, "ru": {"chunks": 5732.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.30248, "cps": 44.78125, "mean_p": 0.315501, "boundary_entropy": 0.542988, "bpb": 2.250124}, "el": {"chunks": 6319.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307898, "cps": 49.367188, "mean_p": 0.34169, "boundary_entropy": 0.581352, "bpb": 2.263219}, "hi": {"chunks": 7165.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.425854, "cps": 55.976562, "mean_p": 0.348189, "boundary_entropy": 0.566629, "bpb": 1.624164}, "ta": {"chunks": 7215.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.385932, "cps": 56.367188, "mean_p": 0.348772, "boundary_entropy": 0.585639, "bpb": 1.429799}, "th": {"chunks": 8050.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.500466, "cps": 62.890625, "mean_p": 0.299971, "boundary_entropy": 0.515278, "bpb": 1.732225}, "zh": {"chunks": 3003.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.508466, "cps": 23.460938, "mean_p": 0.347091, "boundary_entropy": 0.58071, "bpb": 3.864899}, "my": {"chunks": 9799.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.517152, "cps": 76.554688, "mean_p": 0.400051, "boundary_entropy": 0.634055, "bpb": 1.676381}}, "bpb": {"en": 2.94391, "de": 3.28004, "vi": 2.58067, "tr": 3.16308, "fi": 3.37574, "ru": 2.25012, "el": 2.26322, "hi": 1.62416, "ta": 1.4298, "th": 1.73222, "zh": 3.8649, "my": 1.67638}, "gini_cpc": 0.203511, "gini_cps": 0.203825, "bpic": 5.664413, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.034610727558833, "vi": 1.5580761176479592, "tr": 1.3398381639579993, "fi": 1.0658078070441344, "ru": 1.6501765245945028, "el": 1.6797356799932548, "hi": 2.3232425810598687, "ta": 2.105446949505383, "th": 2.7302866920737503, "zh": 2.7739289092331583, "my": 2.8213165647874434}, "t": 48.7}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.95221, "de": 3.27888, "vi": 2.58891, "tr": 3.17349, "fi": 3.36481, "ru": 2.23386, "el": 2.25572, "hi": 1.6189, "ta": 1.44333, "th": 1.76358, "zh": 3.87626, "my": 1.68893}, "bytes_per_chunk": {"en": 5.343, "de": 5.228, "vi": 5.067, "tr": 3.999, "fi": 5.414, "ru": 6.064, "el": 5.782, "hi": 5.738, "ta": 6.085, "th": 4.66, "zh": 5.525, "my": 6.662}, "bpb_mean": 2.51991, "t": 50.0}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.94497, "bpb": 2.806, "chunk_loss": 0.030497, "F": 0.18268, "G": 0.31321, "chunks": 187.06, "bytes_per_chunk": 5.474, "boundary_entropy": 0.51665, "router_noise": 0.0, "grad_norm": 0.8441, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 50.0}
{"event": "train", "step": 650, "phase": 1, "objective": "global", "lm_loss": 1.74157, "bpb": 2.51255, "chunk_loss": 0.030335, "F": 0.17621, "G": 0.31956, "chunks": 180.44, "bytes_per_chunk": 5.675, "boundary_entropy": 0.52535, "router_noise": 0.0, "grad_norm": 0.9793, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 52.1}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.75562, "bpb": 2.53282, "chunk_loss": 0.031057, "F": 0.1962, "G": 0.31341, "chunks": 200.91, "bytes_per_chunk": 5.097, "boundary_entropy": 0.50889, "router_noise": 0.0, "grad_norm": 0.7299, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 54.1}
{"event": "train", "step": 750, "phase": 1, "objective": "global", "lm_loss": 1.69285, "bpb": 2.44226, "chunk_loss": 0.030271, "F": 0.17645, "G": 0.31638, "chunks": 180.69, "bytes_per_chunk": 5.667, "boundary_entropy": 0.51853, "router_noise": 0.0, "grad_norm": 0.6616, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 56.1}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3224.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185255, "cps": 25.1875, "mean_p": 0.275189, "boundary_entropy": 0.43411, "bpb": 2.738482}, "de": {"chunks": 3637.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177129, "cps": 28.414062, "mean_p": 0.270544, "boundary_entropy": 0.432144, "bpb": 3.080824}, "vi": {"chunks": 4454.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.241606, "cps": 34.796875, "mean_p": 0.278887, "boundary_entropy": 0.462797, "bpb": 2.426329}, "tr": {"chunks": 3461.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191184, "cps": 27.039062, "mean_p": 0.287877, "boundary_entropy": 0.470958, "bpb": 3.019397}, "fi": {"chunks": 3557.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189091, "cps": 27.789062, "mean_p": 0.284032, "boundary_entropy": 0.467012, "bpb": 3.241286}, "ru": {"chunks": 5746.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.303219, "cps": 44.890625, "mean_p": 0.348359, "boundary_entropy": 0.572798, "bpb": 2.053949}, "el": {"chunks": 6310.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.30746, "cps": 49.296875, "mean_p": 0.308043, "boundary_entropy": 0.53018, "bpb": 2.143338}, "hi": {"chunks": 7782.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.462526, "cps": 60.796875, "mean_p": 0.314823, "boundary_entropy": 0.504653, "bpb": 1.541684}, "ta": {"chunks": 9157.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.48981, "cps": 71.539062, "mean_p": 0.356137, "boundary_entropy": 0.57578, "bpb": 1.263861}, "th": {"chunks": 8880.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.552067, "cps": 69.375, "mean_p": 0.365705, "boundary_entropy": 0.599187, "bpb": 1.602363}, "zh": {"chunks": 2807.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.475279, "cps": 21.929688, "mean_p": 0.339029, "boundary_entropy": 0.568984, "bpb": 3.738933}, "my": {"chunks": 13162.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.694638, "cps": 102.828125, "mean_p": 0.41557, "boundary_entropy": 0.646188, "bpb": 1.470195}}, "bpb": {"en": 2.73848, "de": 3.08082, "vi": 2.42633, "tr": 3.0194, "fi": 3.24129, "ru": 2.05395, "el": 2.14334, "hi": 1.54168, "ta": 1.26386, "th": 1.60236, "zh": 3.73893, "my": 1.4702}, "gini_cpc": 0.25739, "gini_cps": 0.272098, "bpic": 5.31722, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9561366838034373, "vi": 1.3041758616721215, "tr": 1.0320010399548545, "fi": 1.0207069427594566, "ru": 1.6367618520725167, "el": 1.6596541687961541, "hi": 2.49669355760474, "ta": 2.6439718791332987, "th": 2.980032411560483, "zh": 2.565535668819781, "my": 3.7496229395744702}, "t": 59.1}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.75042, "de": 3.08256, "vi": 2.43383, "tr": 3.02906, "fi": 3.23888, "ru": 2.0394, "el": 2.13168, "hi": 1.53741, "ta": 1.27592, "th": 1.62955, "zh": 3.75145, "my": 1.48132}, "bytes_per_chunk": {"en": 5.495, "de": 5.666, "vi": 5.397, "tr": 5.443, "fi": 5.454, "ru": 6.264, "el": 5.865, "hi": 5.345, "ta": 5.16, "th": 5.078, "zh": 6.123, "my": 4.456}, "bpb_mean": 2.36512, "t": 60.3}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.81383, "bpb": 2.6168, "chunk_loss": 0.030394, "F": 0.17923, "G": 0.30388, "chunks": 183.53, "bytes_per_chunk": 5.579, "boundary_entropy": 0.49495, "router_noise": 0.0, "grad_norm": 0.7938, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 60.4}
{"event": "train", "step": 850, "phase": 1, "objective": "global", "lm_loss": 1.86651, "bpb": 2.69281, "chunk_loss": 0.030197, "F": 0.17487, "G": 0.30422, "chunks": 179.06, "bytes_per_chunk": 5.719, "boundary_entropy": 0.49085, "router_noise": 0.0, "grad_norm": 0.8959, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 62.4}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.68304, "bpb": 2.42812, "chunk_loss": 0.030212, "F": 0.17645, "G": 0.2955, "chunks": 180.69, "bytes_per_chunk": 5.667, "boundary_entropy": 0.4692, "router_noise": 0.0, "grad_norm": 0.997, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 64.4}
{"event": "train", "step": 950, "phase": 1, "objective": "global", "lm_loss": 1.80903, "bpb": 2.60987, "chunk_loss": 0.030433, "F": 0.181, "G": 0.30507, "chunks": 185.34, "bytes_per_chunk": 5.525, "boundary_entropy": 0.47531, "router_noise": 0.0, "grad_norm": 0.7984, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 66.4}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3202.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.183991, "cps": 25.015625, "mean_p": 0.265362, "boundary_entropy": 0.395661, "bpb": 2.599345}, "de": {"chunks": 3917.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.190766, "cps": 30.601562, "mean_p": 0.275968, "boundary_entropy": 0.415144, "bpb": 2.959533}, "vi": {"chunks": 4715.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.255763, "cps": 36.835938, "mean_p": 0.274656, "boundary_entropy": 0.448243, "bpb": 2.321187}, "tr": {"chunks": 3831.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211622, "cps": 29.929688, "mean_p": 0.283341, "boundary_entropy": 0.439333, "bpb": 2.921124}, "fi": {"chunks": 3949.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.20993, "cps": 30.851562, "mean_p": 0.287003, "boundary_entropy": 0.439155, "bpb": 3.143469}, "ru": {"chunks": 5506.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.290554, "cps": 43.015625, "mean_p": 0.356929, "boundary_entropy": 0.578868, "bpb": 1.952767}, "el": {"chunks": 6369.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.310335, "cps": 49.757812, "mean_p": 0.294016, "boundary_entropy": 0.517443, "bpb": 2.029748}, "hi": {"chunks": 6130.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.364339, "cps": 47.890625, "mean_p": 0.326225, "boundary_entropy": 0.557746, "bpb": 1.509147}, "ta": {"chunks": 6768.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.362022, "cps": 52.875, "mean_p": 0.35587, "boundary_entropy": 0.604522, "bpb": 1.217685}, "th": {"chunks": 6986.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.434318, "cps": 54.578125, "mean_p": 0.356491, "boundary_entropy": 0.606548, "bpb": 1.52653}, "zh": {"chunks": 2961.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.501355, "cps": 23.132812, "mean_p": 0.341726, "boundary_entropy": 0.560933, "bpb": 3.624524}, "my": {"chunks": 9451.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.498786, "cps": 73.835938, "mean_p": 0.393679, "boundary_entropy": 0.64323, "bpb": 1.382696}}, "bpb": {"en": 2.59934, "de": 2.95953, "vi": 2.32119, "tr": 2.92112, "fi": 3.14347, "ru": 1.95277, "el": 2.02975, "hi": 1.50915, "ta": 1.21768, "th": 1.52653, "zh": 3.62452, "my": 1.3827}, "gini_cpc": 0.195257, "gini_cps": 0.190221, "bpic": 6.016791, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0368214108377753, "vi": 1.3900849702865734, "tr": 1.150176243559867, "fi": 1.1409800291276406, "ru": 1.579173273959712, "el": 1.6866819460611566, "hi": 1.980196069878317, "ta": 1.9676038929266428, "th": 2.360534263536331, "zh": 2.724882359547971, "my": 2.7109229843285885}, "t": 69.4}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.6104, "de": 2.96335, "vi": 2.32522, "tr": 2.93073, "fi": 3.14418, "ru": 1.93668, "el": 2.01885, "hi": 1.50485, "ta": 1.22924, "th": 1.55203, "zh": 3.63429, "my": 1.39512}, "bytes_per_chunk": {"en": 5.418, "de": 5.199, "vi": 5.085, "tr": 5.024, "fi": 4.898, "ru": 6.206, "el": 5.972, "hi": 6.92, "ta": 6.935, "th": 6.18, "zh": 5.639, "my": 5.99}, "bpb_mean": 2.27041, "t": 70.7}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.74102, "bpb": 2.51177, "chunk_loss": 0.030249, "F": 0.17755, "G": 0.30869, "chunks": 181.81, "bytes_per_chunk": 5.632, "boundary_entropy": 0.49309, "router_noise": 0.0, "grad_norm": 0.8015, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 70.7}
{"event": "train", "step": 1050, "phase": 1, "objective": "global", "lm_loss": 1.60811, "bpb": 2.32002, "chunk_loss": 0.030752, "F": 0.18787, "G": 0.33245, "chunks": 192.38, "bytes_per_chunk": 5.323, "boundary_entropy": 0.50902, "router_noise": 0.0, "grad_norm": 0.7552, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 72.7}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.80504, "bpb": 2.60412, "chunk_loss": 0.03069, "F": 0.1886, "G": 0.31654, "chunks": 193.12, "bytes_per_chunk": 5.302, "boundary_entropy": 0.48924, "router_noise": 0.0, "grad_norm": 0.7482, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 74.7}
{"event": "train", "step": 1150, "phase": 1, "objective": "global", "lm_loss": 1.6576, "bpb": 2.3914, "chunk_loss": 0.030594, "F": 0.18671, "G": 0.29975, "chunks": 191.19, "bytes_per_chunk": 5.356, "boundary_entropy": 0.46078, "router_noise": 0.0, "grad_norm": 0.8046, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 76.7}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3165.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.181865, "cps": 24.726562, "mean_p": 0.250051, "boundary_entropy": 0.363208, "bpb": 2.496172}, "de": {"chunks": 3783.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18424, "cps": 29.554688, "mean_p": 0.261259, "boundary_entropy": 0.388363, "bpb": 2.830076}, "vi": {"chunks": 4091.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221915, "cps": 31.960938, "mean_p": 0.276024, "boundary_entropy": 0.449767, "bpb": 2.284307}, "tr": {"chunks": 3994.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.220626, "cps": 31.203125, "mean_p": 0.274413, "boundary_entropy": 0.403092, "bpb": 2.79925}, "fi": {"chunks": 4119.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.218968, "cps": 32.179688, "mean_p": 0.283161, "boundary_entropy": 0.407525, "bpb": 3.066593}, "ru": {"chunks": 5990.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.316095, "cps": 46.796875, "mean_p": 0.367601, "boundary_entropy": 0.576954, "bpb": 1.86376}, "el": {"chunks": 7342.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.357745, "cps": 57.359375, "mean_p": 0.297863, "boundary_entropy": 0.503817, "bpb": 1.964215}, "hi": {"chunks": 8395.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.49896, "cps": 65.585938, "mean_p": 0.356454, "boundary_entropy": 0.580158, "bpb": 1.468597}, "ta": {"chunks": 8112.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.433913, "cps": 63.375, "mean_p": 0.351372, "boundary_entropy": 0.596314, "bpb": 1.185331}, "th": {"chunks": 9133.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.567796, "cps": 71.351562, "mean_p": 0.372769, "boundary_entropy": 0.613032, "bpb": 1.470302}, "zh": {"chunks": 2628.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.444971, "cps": 20.53125, "mean_p": 0.33725, "boundary_entropy": 0.551684, "bpb": 3.580966}, "my": {"chunks": 9345.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.493192, "cps": 73.007812, "mean_p": 0.382732, "boundary_entropy": 0.632108, "bpb": 1.328005}}, "bpb": {"en": 2.49617, "de": 2.83008, "vi": 2.28431, "tr": 2.79925, "fi": 3.06659, "ru": 1.86376, "el": 1.96421, "hi": 1.4686, "ta": 1.18533, "th": 1.4703, "zh": 3.58097, "my": 1.32801}, "gini_cpc": 0.217309, "gini_cps": 0.22837, "bpic": 5.474999, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0130580688167448, "vi": 1.220216078780704, "tr": 1.2131315984651285, "fi": 1.2040106109136002, "ru": 1.7380730032887743, "el": 1.9670886876889715, "hi": 2.7435699245790586, "ta": 2.3859035223702008, "th": 3.1220711651814192, "zh": 2.446709025924315, "my": 2.7118541941177003}, "t": 79.7}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.50914, "de": 2.8354, "vi": 2.28804, "tr": 2.80933, "fi": 3.0718, "ru": 1.84668, "el": 1.95438, "hi": 1.46635, "ta": 1.19372, "th": 1.49514, "zh": 3.58993, "my": 1.3382}, "bytes_per_chunk": {"en": 5.494, "de": 5.413, "vi": 5.723, "tr": 4.732, "fi": 4.639, "ru": 5.858, "el": 5.051, "hi": 4.912, "ta": 6.068, "th": 5.012, "zh": 6.435, "my": 5.667}, "bpb_mean": 2.19984, "t": 80.9}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.72036, "bpb": 2.48196, "chunk_loss": 0.030535, "F": 0.18729, "G": 0.29043, "chunks": 191.78, "bytes_per_chunk": 5.339, "boundary_entropy": 0.44217, "router_noise": 0.0, "grad_norm": 0.8033, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 81.0}
{"event": "train", "step": 1250, "phase": 1, "objective": "global", "lm_loss": 1.84156, "bpb": 2.65681, "chunk_loss": 0.030405, "F": 0.1832, "G": 0.2948, "chunks": 187.59, "bytes_per_chunk": 5.459, "boundary_entropy": 0.44766, "router_noise": 0.0, "grad_norm": 0.9228, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 83.0}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.51745, "bpb": 2.18921, "chunk_loss": 0.030091, "F": 0.17371, "G": 0.31035, "chunks": 177.88, "bytes_per_chunk": 5.757, "boundary_entropy": 0.48402, "router_noise": 0.0, "grad_norm": 0.7648, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 85.0}
{"event": "train", "step": 1350, "phase": 1, "objective": "global", "lm_loss": 1.5112, "bpb": 2.1802, "chunk_loss": 0.030295, "F": 0.17883, "G": 0.30753, "chunks": 183.12, "bytes_per_chunk": 5.592, "boundary_entropy": 0.47382, "router_noise": 0.0, "grad_norm": 0.7331, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 87.0}
{"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3310.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190197, "cps": 25.859375, "mean_p": 0.248933, "boundary_entropy": 0.351456, "bpb": 2.43428}, "de": {"chunks": 3869.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188428, "cps": 30.226562, "mean_p": 0.259038, "boundary_entropy": 0.375182, "bpb": 2.760649}, "vi": {"chunks": 4424.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.239978, "cps": 34.5625, "mean_p": 0.281803, "boundary_entropy": 0.446666, "bpb": 2.202778}, "tr": {"chunks": 3612.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199525, "cps": 28.21875, "mean_p": 0.248075, "boundary_entropy": 0.376451, "bpb": 2.730484}, "fi": {"chunks": 3356.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178406, "cps": 26.21875, "mean_p": 0.244001, "boundary_entropy": 0.371758, "bpb": 3.02309}, "ru": {"chunks": 5941.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.313509, "cps": 46.414062, "mean_p": 0.36333, "boundary_entropy": 0.566782, "bpb": 1.803863}, "el": {"chunks": 6461.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.314818, "cps": 50.476562, "mean_p": 0.284412, "boundary_entropy": 0.507632, "bpb": 1.879895}, "hi": {"chunks": 7401.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.439881, "cps": 57.820312, "mean_p": 0.373641, "boundary_entropy": 0.612516, "bpb": 1.421689}, "ta": {"chunks": 8154.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.436159, "cps": 63.703125, "mean_p": 0.40487, "boundary_entropy": 0.636231, "bpb": 1.139825}, "th": {"chunks": 8030.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.499223, "cps": 62.734375, "mean_p": 0.390498, "boundary_entropy": 0.627816, "bpb": 1.409892}, "zh": {"chunks": 2377.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.402472, "cps": 18.570312, "mean_p": 0.331093, "boundary_entropy": 0.551902, "bpb": 3.541631}, "my": {"chunks": 9163.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.483587, "cps": 71.585938, "mean_p": 0.413352, "boundary_entropy": 0.654802, "bpb": 1.286485}}, "bpb": {"en": 2.43428, "de": 2.76065, "vi": 2.20278, "tr": 2.73048, "fi": 3.02309, "ru": 1.80386, "el": 1.87989, "hi": 1.42169, "ta": 1.13982, "th": 1.40989, "zh": 3.54163, "my": 1.28648}, "gini_cpc": 0.205968, "gini_cps": 0.225925, "bpic": 5.806242, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9907006523872925, "vi": 1.2617348616884505, "tr": 1.0490430639177328, "fi": 0.9380072500486233, "ru": 1.6483387352629355, "el": 1.655217320836559, "hi": 2.31276474100277, "ta": 2.2931969951737585, "th": 2.6247660887637148, "zh": 2.116078942710726, "my": 2.5425554722194055}, "t": 90.0}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.44313, "de": 2.76423, "vi": 2.20645, "tr": 2.73917, "fi": 3.02289, "ru": 1.78594, "el": 1.86903, "hi": 1.41805, "ta": 1.14908, "th": 1.43184, "zh": 3.54748, "my": 1.29748}, "bytes_per_chunk": {"en": 5.345, "de": 5.342, "vi": 5.781, "tr": 5.289, "fi": 5.693, "ru": 5.775, "el": 5.759, "hi": 5.54, "ta": 6.012, "th": 5.73, "zh": 7.11, "my": 5.705}, "bpb_mean": 2.13956, "t": 91.2}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.57429, "bpb": 2.27122, "chunk_loss": 0.030446, "F": 0.18466, "G": 0.30575, "chunks": 189.09, "bytes_per_chunk": 5.415, "boundary_entropy": 0.46222, "router_noise": 0.0, "grad_norm": 0.6872, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 91.3}
{"event": "train", "step": 1450, "phase": 1, "objective": "global", "lm_loss": 1.66756, "bpb": 2.40579, "chunk_loss": 0.030087, "F": 0.17157, "G": 0.29581, "chunks": 175.69, "bytes_per_chunk": 5.829, "boundary_entropy": 0.46047, "router_noise": 0.0, "grad_norm": 0.7207, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 93.3}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.74314, "bpb": 2.51482, "chunk_loss": 0.030231, "F": 0.17612, "G": 0.31108, "chunks": 180.34, "bytes_per_chunk": 5.678, "boundary_entropy": 0.48478, "router_noise": 0.0, "grad_norm": 0.7436, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 95.3}
{"event": "train", "step": 1550, "phase": 1, "objective": "global", "lm_loss": 1.60176, "bpb": 2.31084, "chunk_loss": 0.030229, "F": 0.1795, "G": 0.29894, "chunks": 183.81, "bytes_per_chunk": 5.571, "boundary_entropy": 0.45214, "router_noise": 0.0, "grad_norm": 0.7673, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 97.3}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3371.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193702, "cps": 26.335938, "mean_p": 0.247309, "boundary_entropy": 0.338562, "bpb": 2.374931}, "de": {"chunks": 4028.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.196172, "cps": 31.46875, "mean_p": 0.26183, "boundary_entropy": 0.368463, "bpb": 2.67142}, "vi": {"chunks": 3898.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211446, "cps": 30.453125, "mean_p": 0.272899, "boundary_entropy": 0.439448, "bpb": 2.170272}, "tr": {"chunks": 3855.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212948, "cps": 30.117188, "mean_p": 0.258292, "boundary_entropy": 0.37214, "bpb": 2.666044}, "fi": {"chunks": 3864.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.205412, "cps": 30.1875, "mean_p": 0.268149, "boundary_entropy": 0.381919, "bpb": 2.89236}, "ru": {"chunks": 6229.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.328707, "cps": 48.664062, "mean_p": 0.356575, "boundary_entropy": 0.548595, "bpb": 1.745093}, "el": {"chunks": 6824.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.332505, "cps": 53.3125, "mean_p": 0.286483, "boundary_entropy": 0.495416, "bpb": 1.840002}, "hi": {"chunks": 7372.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.438158, "cps": 57.59375, "mean_p": 0.390406, "boundary_entropy": 0.623168, "bpb": 1.400659}, "ta": {"chunks": 9561.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.51142, "cps": 74.695312, "mean_p": 0.405416, "boundary_entropy": 0.622279, "bpb": 1.121443}, "th": {"chunks": 6736.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.418775, "cps": 52.625, "mean_p": 0.370108, "boundary_entropy": 0.61897, "bpb": 1.383473}, "zh": {"chunks": 2142.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.362682, "cps": 16.734375, "mean_p": 0.32562, "boundary_entropy": 0.547516, "bpb": 3.518135}, "my": {"chunks": 9283.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.48992, "cps": 72.523438, "mean_p": 0.410817, "boundary_entropy": 0.652476, "bpb": 1.257703}}, "bpb": {"en": 2.37493, "de": 2.67142, "vi": 2.17027, "tr": 2.66604, "fi": 2.89236, "ru": 1.74509, "el": 1.84, "hi": 1.40066, "ta": 1.12144, "th": 1.38347, "zh": 3.51814, "my": 1.2577}, "gini_cpc": 0.197716, "gini_cps": 0.22948, "bpic": 5.714173, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0127503976891834, "vi": 1.0916013409739271, "tr": 1.0993581456241701, "fi": 1.0604509902489725, "ru": 1.6969712486371025, "el": 1.7165779939136232, "hi": 2.2620157333571944, "ta": 2.6402388447822305, "th": 2.1619536600419, "zh": 1.8723687820349322, "my": 2.529241749985769}, "t": 100.3}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.38278, "de": 2.67652, "vi": 2.17624, "tr": 2.67544, "fi": 2.89893, "ru": 1.72747, "el": 1.82979, "hi": 1.39774, "ta": 1.13162, "th": 1.40114, "zh": 3.5229, "my": 1.26778}, "bytes_per_chunk": {"en": 5.251, "de": 5.099, "vi": 6.181, "tr": 5.025, "fi": 4.924, "ru": 5.665, "el": 5.426, "hi": 5.631, "ta": 5.365, "th": 6.704, "zh": 7.394, "my": 5.608}, "bpb_mean": 2.0907, "t": 101.5}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.59957, "bpb": 2.30769, "chunk_loss": 0.030595, "F": 0.18582, "G": 0.3055, "chunks": 190.28, "bytes_per_chunk": 5.382, "boundary_entropy": 0.45624, "router_noise": 0.0, "grad_norm": 0.74, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 101.6}
{"event": "train", "step": 1650, "phase": 1, "objective": "global", "lm_loss": 1.58134, "bpb": 2.28139, "chunk_loss": 0.030401, "F": 0.18161, "G": 0.29362, "chunks": 185.97, "bytes_per_chunk": 5.506, "boundary_entropy": 0.43935, "router_noise": 0.0, "grad_norm": 0.7962, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 103.6}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.54453, "bpb": 2.22829, "chunk_loss": 0.030343, "F": 0.17957, "G": 0.30112, "chunks": 183.88, "bytes_per_chunk": 5.569, "boundary_entropy": 0.45431, "router_noise": 0.0, "grad_norm": 0.7751, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 105.6}
{"event": "train", "step": 1750, "phase": 1, "objective": "global", "lm_loss": 1.56163, "bpb": 2.25296, "chunk_loss": 0.030815, "F": 0.19116, "G": 0.2968, "chunks": 195.75, "bytes_per_chunk": 5.231, "boundary_entropy": 0.43719, "router_noise": 0.0, "grad_norm": 0.7014, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 107.6}
{"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3378.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194104, "cps": 26.390625, "mean_p": 0.236387, "boundary_entropy": 0.317953, "bpb": 2.327439}, "de": {"chunks": 3914.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.19062, "cps": 30.578125, "mean_p": 0.242654, "boundary_entropy": 0.334058, "bpb": 2.632062}, "vi": {"chunks": 4440.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.240846, "cps": 34.6875, "mean_p": 0.281409, "boundary_entropy": 0.433924, "bpb": 2.132089}, "tr": {"chunks": 3793.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.209523, "cps": 29.632812, "mean_p": 0.245464, "boundary_entropy": 0.357791, "bpb": 2.622393}, "fi": {"chunks": 3593.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.191005, "cps": 28.070312, "mean_p": 0.246001, "boundary_entropy": 0.348149, "bpb": 2.844663}, "ru": {"chunks": 5700.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.300792, "cps": 44.53125, "mean_p": 0.340952, "boundary_entropy": 0.544681, "bpb": 1.701638}, "el": {"chunks": 6056.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.295084, "cps": 47.3125, "mean_p": 0.269557, "boundary_entropy": 0.480519, "bpb": 1.803027}, "hi": {"chunks": 8220.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.488559, "cps": 64.21875, "mean_p": 0.370151, "boundary_entropy": 0.605467, "bpb": 1.362877}, "ta": {"chunks": 8186.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.437871, "cps": 63.953125, "mean_p": 0.368239, "boundary_entropy": 0.605559, "bpb": 1.091758}, "th": {"chunks": 7543.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.468946, "cps": 58.929688, "mean_p": 0.394986, "boundary_entropy": 0.629551, "bpb": 1.33608}, "zh": {"chunks": 2586.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.43786, "cps": 20.203125, "mean_p": 0.317908, "boundary_entropy": 0.525396, "bpb": 3.4728}, "my": {"chunks": 7126.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.376082, "cps": 55.671875, "mean_p": 0.397814, "boundary_entropy": 0.649577, "bpb": 1.24805}}, "bpb": {"en": 2.32744, "de": 2.63206, "vi": 2.13209, "tr": 2.62239, "fi": 2.84466, "ru": 1.70164, "el": 1.80303, "hi": 1.36288, "ta": 1.09176, "th": 1.33608, "zh": 3.4728, "my": 1.24805}, "gini_cpc": 0.196589, "gini_cps": 0.203051, "bpic": 5.946866, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.982048392575864, "vi": 1.2408071949786612, "tr": 1.079435672467857, "fi": 0.9840333303338161, "ru": 1.5496374960750223, "el": 1.5202306920909858, "hi": 2.5169884322735085, "ta": 2.255852739379504, "th": 2.415947638295887, "zh": 2.2557946009086103, "my": 1.9375232242603955}, "t": 110.6}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.33496, "de": 2.63503, "vi": 2.13633, "tr": 2.63136, "fi": 2.84581, "ru": 1.68104, "el": 1.79595, "hi": 1.35914, "ta": 1.10129, "th": 1.357, "zh": 3.47642, "my": 1.25612}, "bytes_per_chunk": {"en": 5.239, "de": 5.265, "vi": 5.385, "tr": 5.11, "fi": 5.161, "ru": 5.999, "el": 6.014, "hi": 5.558, "ta": 6.103, "th": 6.136, "zh": 6.184, "my": 7.008}, "bpb_mean": 2.05087, "t": 111.8}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.53112, "bpb": 2.20894, "chunk_loss": 0.030308, "F": 0.18057, "G": 0.29501, "chunks": 184.91, "bytes_per_chunk": 5.538, "boundary_entropy": 0.45493, "router_noise": 0.0, "grad_norm": 0.7419, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 111.9}
{"event": "train", "step": 1850, "phase": 1, "objective": "global", "lm_loss": 1.57603, "bpb": 2.27373, "chunk_loss": 0.030487, "F": 0.18585, "G": 0.28174, "chunks": 190.31, "bytes_per_chunk": 5.381, "boundary_entropy": 0.41379, "router_noise": 0.0, "grad_norm": 0.6978, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 113.9}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.63464, "bpb": 2.35829, "chunk_loss": 0.031109, "F": 0.19287, "G": 0.29915, "chunks": 197.5, "bytes_per_chunk": 5.185, "boundary_entropy": 0.43125, "router_noise": 0.0, "grad_norm": 0.8166, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 115.9}
{"event": "train", "step": 1950, "phase": 1, "objective": "global", "lm_loss": 1.47319, "bpb": 2.12537, "chunk_loss": 0.029932, "F": 0.16876, "G": 0.30187, "chunks": 172.81, "bytes_per_chunk": 5.925, "boundary_entropy": 0.46072, "router_noise": 0.0, "grad_norm": 0.693, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 118.0}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3387.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194622, "cps": 26.460938, "mean_p": 0.238373, "boundary_entropy": 0.312568, "bpb": 2.294749}, "de": {"chunks": 4053.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.19739, "cps": 31.664062, "mean_p": 0.248696, "boundary_entropy": 0.326944, "bpb": 2.567433}, "vi": {"chunks": 4315.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.234066, "cps": 33.710938, "mean_p": 0.2706, "boundary_entropy": 0.411051, "bpb": 2.108605}, "tr": {"chunks": 3783.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208971, "cps": 29.554688, "mean_p": 0.242939, "boundary_entropy": 0.332447, "bpb": 2.564256}, "fi": {"chunks": 3616.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.192228, "cps": 28.25, "mean_p": 0.251529, "boundary_entropy": 0.348698, "bpb": 2.781837}, "ru": {"chunks": 5925.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.312665, "cps": 46.289062, "mean_p": 0.338505, "boundary_entropy": 0.531669, "bpb": 1.649358}, "el": {"chunks": 6992.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.340691, "cps": 54.625, "mean_p": 0.293325, "boundary_entropy": 0.494309, "bpb": 1.71888}, "hi": {"chunks": 8903.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.529153, "cps": 69.554688, "mean_p": 0.406677, "boundary_entropy": 0.621597, "bpb": 1.337019}, "ta": {"chunks": 8302.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.444076, "cps": 64.859375, "mean_p": 0.374714, "boundary_entropy": 0.609814, "bpb": 1.081025}, "th": {"chunks": 7738.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.481069, "cps": 60.453125, "mean_p": 0.400719, "boundary_entropy": 0.633742, "bpb": 1.31414}, "zh": {"chunks": 2662.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.450728, "cps": 20.796875, "mean_p": 0.327433, "boundary_entropy": 0.533584, "bpb": 3.437395}, "my": {"chunks": 8674.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.457779, "cps": 67.765625, "mean_p": 0.407053, "boundary_entropy": 0.649325, "bpb": 1.224017}}, "bpb": {"en": 2.29475, "de": 2.56743, "vi": 2.1086, "tr": 2.56426, "fi": 2.78184, "ru": 1.64936, "el": 1.71888, "hi": 1.33702, "ta": 1.08102, "th": 1.31414, "zh": 3.4374, "my": 1.22402}, "gini_cpc": 0.205555, "gini_cps": 0.21853, "bpic": 5.614938, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0142222179276574, "vi": 1.2026702874605746, "tr": 1.073729075240014, "fi": 0.9877009282443573, "ru": 1.6065271295727184, "el": 1.7505297584204926, "hi": 2.718881151645725, "ta": 2.2817401430970468, "th": 2.4718185304872557, "zh": 2.3159198869084805, "my": 2.3521497369239346}, "t": 120.9}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.30228, "de": 2.57327, "vi": 2.11293, "tr": 2.57002, "fi": 2.78967, "ru": 1.6252, "el": 1.71095, "hi": 1.33428, "ta": 1.08929, "th": 1.33365, "zh": 3.43965, "my": 1.23207}, "bytes_per_chunk": {"en": 5.243, "de": 5.07, "vi": 5.56, "tr": 5.246, "fi": 5.142, "ru": 5.908, "el": 5.198, "hi": 5.064, "ta": 6.076, "th": 5.873, "zh": 5.999, "my": 5.774}, "bpb_mean": 2.00944, "t": 122.2}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.59477, "bpb": 2.30077, "chunk_loss": 0.030405, "F": 0.18625, "G": 0.28652, "chunks": 190.72, "bytes_per_chunk": 5.369, "boundary_entropy": 0.41542, "router_noise": 0.0, "grad_norm": 0.7412, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 122.2}
{"event": "train", "step": 2050, "phase": 1, "objective": "global", "lm_loss": 1.60667, "bpb": 2.31793, "chunk_loss": 0.030415, "F": 0.18311, "G": 0.30016, "chunks": 187.5, "bytes_per_chunk": 5.461, "boundary_entropy": 0.45102, "router_noise": 0.0, "grad_norm": 0.7581, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 124.3}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.54529, "bpb": 2.22938, "chunk_loss": 0.03046, "F": 0.18491, "G": 0.29143, "chunks": 189.34, "bytes_per_chunk": 5.408, "boundary_entropy": 0.42314, "router_noise": 0.0, "grad_norm": 0.7281, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 126.3}
{"event": "train", "step": 2150, "phase": 1, "objective": "global", "lm_loss": 1.50323, "bpb": 2.1687, "chunk_loss": 0.03029, "F": 0.17963, "G": 0.29092, "chunks": 183.94, "bytes_per_chunk": 5.567, "boundary_entropy": 0.42792, "router_noise": 0.0, "grad_norm": 0.7171, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 128.3}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3465.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199104, "cps": 27.070312, "mean_p": 0.242076, "boundary_entropy": 0.306609, "bpb": 2.259391}, "de": {"chunks": 4082.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.198802, "cps": 31.890625, "mean_p": 0.248968, "boundary_entropy": 0.329824, "bpb": 2.518016}, "vi": {"chunks": 4219.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.228858, "cps": 32.960938, "mean_p": 0.265443, "boundary_entropy": 0.403563, "bpb": 2.087305}, "tr": {"chunks": 3880.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214329, "cps": 30.3125, "mean_p": 0.24579, "boundary_entropy": 0.336018, "bpb": 2.515674}, "fi": {"chunks": 3912.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.207963, "cps": 30.5625, "mean_p": 0.259307, "boundary_entropy": 0.343095, "bpb": 2.723074}, "ru": {"chunks": 6067.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.320158, "cps": 47.398438, "mean_p": 0.341349, "boundary_entropy": 0.524694, "bpb": 1.609205}, "el": {"chunks": 6459.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.31472, "cps": 50.460938, "mean_p": 0.285087, "boundary_entropy": 0.48417, "bpb": 1.693565}, "hi": {"chunks": 7297.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.4337, "cps": 57.007812, "mean_p": 0.388841, "boundary_entropy": 0.62259, "bpb": 1.318843}, "ta": {"chunks": 7537.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.403156, "cps": 58.882812, "mean_p": 0.355408, "boundary_entropy": 0.600724, "bpb": 1.071211}, "th": {"chunks": 6655.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.41374, "cps": 51.992188, "mean_p": 0.376221, "boundary_entropy": 0.623283, "bpb": 1.284498}, "zh": {"chunks": 2843.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.481375, "cps": 22.210938, "mean_p": 0.317796, "boundary_entropy": 0.518512, "bpb": 3.445939}, "my": {"chunks": 8526.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.449968, "cps": 66.609375, "mean_p": 0.423008, "boundary_entropy": 0.659421, "bpb": 1.199957}}, "bpb": {"en": 2.25939, "de": 2.51802, "vi": 2.08731, "tr": 2.51567, "fi": 2.72307, "ru": 1.60921, "el": 1.69356, "hi": 1.31884, "ta": 1.07121, "th": 1.2845, "zh": 3.44594, "my": 1.19996}, "gini_cpc": 0.182616, "gini_cps": 0.187057, "bpic": 5.909596, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9984848379432707, "vi": 1.1494425364889158, "tr": 1.0764703083103349, "fi": 1.0444985560396756, "ru": 1.6079985836502988, "el": 1.5806849584659848, "hi": 2.178262197436046, "ta": 2.024854991669648, "th": 2.078011160945572, "zh": 2.417710509379998, "my": 2.2599708292551863}, "t": 131.2}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.26474, "de": 2.52356, "vi": 2.09008, "tr": 2.52111, "fi": 2.7343, "ru": 1.5873, "el": 1.68489, "hi": 1.31672, "ta": 1.07735, "th": 1.30113, "zh": 3.44691, "my": 1.20878}, "bytes_per_chunk": {"en": 5.058, "de": 5.02, "vi": 5.739, "tr": 5.012, "fi": 4.773, "ru": 5.672, "el": 5.611, "hi": 5.772, "ta": 6.682, "th": 6.748, "zh": 5.716, "my": 5.742}, "bpb_mean": 1.97974, "t": 132.5}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.44649, "bpb": 2.08685, "chunk_loss": 0.030425, "F": 0.18762, "G": 0.2967, "chunks": 192.12, "bytes_per_chunk": 5.33, "boundary_entropy": 0.43061, "router_noise": 0.0, "grad_norm": 0.6895, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 132.6}
{"event": "train", "step": 2250, "phase": 1, "objective": "global", "lm_loss": 1.52982, "bpb": 2.20706, "chunk_loss": 0.030216, "F": 0.17816, "G": 0.28872, "chunks": 182.44, "bytes_per_chunk": 5.613, "boundary_entropy": 0.42507, "router_noise": 0.0, "grad_norm": 0.7054, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 134.6}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.47429, "bpb": 2.12695, "chunk_loss": 0.030443, "F": 0.18402, "G": 0.29355, "chunks": 188.44, "bytes_per_chunk": 5.434, "boundary_entropy": 0.43383, "router_noise": 0.0, "grad_norm": 0.7343, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 136.6}
{"event": "train", "step": 2350, "phase": 1, "objective": "global", "lm_loss": 1.49187, "bpb": 2.15232, "chunk_loss": 0.030064, "F": 0.17581, "G": 0.28642, "chunks": 180.03, "bytes_per_chunk": 5.688, "boundary_entropy": 0.41899, "router_noise": 0.0, "grad_norm": 0.7051, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 138.6}
{"event": "probe", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3355.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192783, "cps": 26.210938, "mean_p": 0.235606, "boundary_entropy": 0.295782, "bpb": 2.231672}, "de": {"chunks": 3992.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.194419, "cps": 31.1875, "mean_p": 0.244042, "boundary_entropy": 0.317335, "bpb": 2.484918}, "vi": {"chunks": 4292.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.232818, "cps": 33.53125, "mean_p": 0.266036, "boundary_entropy": 0.398476, "bpb": 2.072862}, "tr": {"chunks": 3873.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213942, "cps": 30.257812, "mean_p": 0.246567, "boundary_entropy": 0.337418, "bpb": 2.484938}, "fi": {"chunks": 3897.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.207166, "cps": 30.445312, "mean_p": 0.260423, "boundary_entropy": 0.334599, "bpb": 2.68148}, "ru": {"chunks": 6014.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.317361, "cps": 46.984375, "mean_p": 0.343441, "boundary_entropy": 0.521449, "bpb": 1.585791}, "el": {"chunks": 6227.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.303416, "cps": 48.648438, "mean_p": 0.282205, "boundary_entropy": 0.477261, "bpb": 1.664767}, "hi": {"chunks": 6350.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.377415, "cps": 49.609375, "mean_p": 0.384726, "boundary_entropy": 0.616247, "bpb": 1.305664}, "ta": {"chunks": 9381.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.501792, "cps": 73.289062, "mean_p": 0.402934, "boundary_entropy": 0.619737, "bpb": 1.047454}, "th": {"chunks": 7182.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.446503, "cps": 56.109375, "mean_p": 0.396984, "boundary_entropy": 0.634515, "bpb": 1.265646}, "zh": {"chunks": 2611.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.442093, "cps": 20.398438, "mean_p": 0.312153, "boundary_entropy": 0.514514, "bpb": 3.377257}, "my": {"chunks": 8116.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.42833, "cps": 63.40625, "mean_p": 0.408214, "boundary_entropy": 0.648616, "bpb": 1.183507}}, "bpb": {"en": 2.23167, "de": 2.48492, "vi": 2.07286, "tr": 2.48494, "fi": 2.68148, "ru": 1.58579, "el": 1.66477, "hi": 1.30566, "ta": 1.04745, "th": 1.26565, "zh": 3.37726, "my": 1.18351}, "gini_cpc": 0.190771, "gini_cps": 0.205716, "bpic": 5.878098, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0084856459120037, "vi": 1.207669691351213, "tr": 1.1097586562337898, "fi": 1.0746081204191256, "ru": 1.6462121592236216, "el": 1.5738727506698347, "hi": 1.9577185146473626, "ta": 2.60288668718853, "th": 2.3160926653161042, "zh": 2.293216325513017, "my": 2.2218270916056397}, "t": 141.5}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.23965, "de": 2.48905, "vi": 2.07732, "tr": 2.49273, "fi": 2.69171, "ru": 1.56231, "el": 1.6578, "hi": 1.30286, "ta": 1.05544, "th": 1.28401, "zh": 3.38223, "my": 1.19316}, "bytes_per_chunk": {"en": 5.215, "de": 5.102, "vi": 5.602, "tr": 5.004, "fi": 4.692, "ru": 5.721, "el": 5.747, "hi": 6.75, "ta": 5.342, "th": 6.332, "zh": 6.134, "my": 6.507}, "bpb_mean": 1.95236, "t": 142.8}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 142.8}
{"event": "train", "step": 2400, "phase": 2, "objective": "parity_a", "lm_loss": 1.36903, "bpb": 1.97509, "chunk_loss": 0.036776, "F": 0.18716, "G": 0.29713, "chunks": 191.66, "bytes_per_chunk": 5.343, "boundary_entropy": 0.42602, "router_noise": 0.0, "grad_norm": 0.7212, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.086, "n_target_mean": 10.027, "alpha": 0.03, "t": 142.9}
{"event": "train", "step": 2450, "phase": 2, "objective": "parity_a", "lm_loss": 1.47716, "bpb": 2.13109, "chunk_loss": 0.031132, "F": 0.14682, "G": 0.24194, "chunks": 150.34, "bytes_per_chunk": 6.811, "boundary_entropy": 0.39522, "router_noise": 0.0, "grad_norm": 0.6785, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.8405, "alpha": 0.03, "t": 144.9}
{"event": "train", "step": 2500, "phase": 2, "objective": "parity_a", "lm_loss": 1.5122, "bpb": 2.18165, "chunk_loss": 0.030517, "F": 0.15219, "G": 0.23096, "chunks": 155.84, "bytes_per_chunk": 6.571, "boundary_entropy": 0.35865, "router_noise": 0.0, "grad_norm": 0.748, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.0154, "alpha": 0.03, "t": 146.9}
{"event": "train", "step": 2550, "phase": 2, "objective": "parity_a", "lm_loss": 1.56698, "bpb": 2.26067, "chunk_loss": 0.030605, "F": 0.15912, "G": 0.22881, "chunks": 162.94, "bytes_per_chunk": 6.285, "boundary_entropy": 0.33709, "router_noise": 0.0, "grad_norm": 0.7171, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.354, "alpha": 0.03, "t": 148.9}
{"event": "probe", "step": 2600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3388.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194679, "cps": 26.46875, "mean_p": 0.230488, "boundary_entropy": 0.274236, "bpb": 2.206092}, "de": {"chunks": 3818.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.185945, "cps": 29.828125, "mean_p": 0.222947, "boundary_entropy": 0.277195, "bpb": 2.450771}, "vi": {"chunks": 3366.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.182587, "cps": 26.296875, "mean_p": 0.210457, "boundary_entropy": 0.340869, "bpb": 2.054248}, "tr": {"chunks": 3585.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198033, "cps": 28.007812, "mean_p": 0.223971, "boundary_entropy": 0.289732, "bpb": 2.444137}, "fi": {"chunks": 3646.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.193823, "cps": 28.484375, "mean_p": 0.236341, "boundary_entropy": 0.299205, "bpb": 2.659041}, "ru": {"chunks": 3493.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.184327, "cps": 27.289062, "mean_p": 0.19547, "boundary_entropy": 0.356118, "bpb": 1.606774}, "el": {"chunks": 3655.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.178093, "cps": 28.554688, "mean_p": 0.184035, "boundary_entropy": 0.366727, "bpb": 1.699571}, "hi": {"chunks": 3260.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.193759, "cps": 25.46875, "mean_p": 0.229093, "boundary_entropy": 0.462154, "bpb": 1.337356}, "ta": {"chunks": 2782.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.14881, "cps": 21.734375, "mean_p": 0.201868, "boundary_entropy": 0.440634, "bpb": 1.087279}, "th": {"chunks": 3312.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.205906, "cps": 25.875, "mean_p": 0.217341, "boundary_entropy": 0.453714, "bpb": 1.270923}, "zh": {"chunks": 2962.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.501524, "cps": 23.140625, "mean_p": 0.311929, "boundary_entropy": 0.506705, "bpb": 3.34636}, "my": {"chunks": 2610.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.137745, "cps": 20.390625, "mean_p": 0.289024, "boundary_entropy": 0.566963, "bpb": 1.194682}}, "bpb": {"en": 2.20609, "de": 2.45077, "vi": 2.05425, "tr": 2.44414, "fi": 2.65904, "ru": 1.60677, "el": 1.69957, "hi": 1.33736, "ta": 1.08728, "th": 1.27092, "zh": 3.34636, "my": 1.19468}, "gini_cpc": 0.158062, "gini_cps": 0.05877, "bpic": 9.624119, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9551338470838344, "vi": 0.9378895311360731, "tr": 1.0172303918004029, "fi": 0.995601401802319, "ru": 0.9468258139078479, "el": 0.9148023144016553, "hi": 0.9952753446742371, "ta": 0.7643853848502636, "th": 1.0576695014935047, "zh": 2.5761570187962457, "my": 0.7075511640757193}, "t": 151.8}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.21162, "de": 2.45631, "vi": 2.05773, "tr": 2.4495, "fi": 2.66791, "ru": 1.58301, "el": 1.6905, "hi": 1.33317, "ta": 1.09311, "th": 1.28788, "zh": 3.34923, "my": 1.20286}, "bytes_per_chunk": {"en": 5.171, "de": 5.357, "vi": 7.158, "tr": 5.287, "fi": 5.069, "ru": 9.596, "el": 9.924, "hi": 12.811, "ta": 17.639, "th": 12.975, "zh": 5.743, "my": 17.85}, "bpb_mean": 1.94857, "t": 153.1}
{"event": "train", "step": 2600, "phase": 2, "objective": "parity_a", "lm_loss": 1.59881, "bpb": 2.30659, "chunk_loss": 0.030659, "F": 0.16522, "G": 0.23702, "chunks": 169.19, "bytes_per_chunk": 6.052, "boundary_entropy": 0.34625, "router_noise": 0.0, "grad_norm": 0.749, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.085, "n_target_mean": 8.0814, "alpha": 0.03, "t": 153.2}
{"event": "train", "step": 2650, "phase": 2, "objective": "parity_a", "lm_loss": 1.50426, "bpb": 2.17019, "chunk_loss": 0.030826, "F": 0.14612, "G": 0.22629, "chunks": 149.62, "bytes_per_chunk": 6.844, "boundary_entropy": 0.34772, "router_noise": 0.0, "grad_norm": 0.6549, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.2749, "alpha": 0.03, "t": 155.2}
{"event": "train", "step": 2700, "phase": 2, "objective": "parity_a", "lm_loss": 1.52074, "bpb": 2.19397, "chunk_loss": 0.030491, "F": 0.15472, "G": 0.22858, "chunks": 158.44, "bytes_per_chunk": 6.463, "boundary_entropy": 0.33623, "router_noise": 0.0, "grad_norm": 0.704, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.6096, "alpha": 0.03, "t": 157.2}
{"event": "train", "step": 2750, "phase": 2, "objective": "parity_a", "lm_loss": 1.40661, "bpb": 2.0293, "chunk_loss": 0.030674, "F": 0.15924, "G": 0.22423, "chunks": 163.06, "bytes_per_chunk": 6.28, "boundary_entropy": 0.31322, "router_noise": 0.0, "grad_norm": 0.7092, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.5736, "alpha": 0.03, "t": 159.2}
{"event": "probe", "step": 2800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3278.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188358, "cps": 25.609375, "mean_p": 0.218832, "boundary_entropy": 0.263589, "bpb": 2.193201}, "de": {"chunks": 3723.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181318, "cps": 29.085938, "mean_p": 0.214532, "boundary_entropy": 0.26092, "bpb": 2.428037}, "vi": {"chunks": 3145.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.170599, "cps": 24.570312, "mean_p": 0.202708, "boundary_entropy": 0.331956, "bpb": 2.04535}, "tr": {"chunks": 3461.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191184, "cps": 27.039062, "mean_p": 0.216212, "boundary_entropy": 0.283277, "bpb": 2.429825}, "fi": {"chunks": 3526.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187444, "cps": 27.546875, "mean_p": 0.226131, "boundary_entropy": 0.285984, "bpb": 2.628986}, "ru": {"chunks": 3133.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.16533, "cps": 24.476562, "mean_p": 0.169402, "boundary_entropy": 0.313376, "bpb": 1.573239}, "el": {"chunks": 3262.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.158944, "cps": 25.484375, "mean_p": 0.168272, "boundary_entropy": 0.344091, "bpb": 1.662571}, "hi": {"chunks": 3144.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.186865, "cps": 24.5625, "mean_p": 0.2125, "boundary_entropy": 0.433641, "bpb": 1.322638}, "ta": {"chunks": 2959.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.158278, "cps": 23.117188, "mean_p": 0.201807, "boundary_entropy": 0.433172, "bpb": 1.112256}, "th": {"chunks": 2573.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.159963, "cps": 20.101562, "mean_p": 0.194755, "boundary_entropy": 0.423797, "bpb": 1.267827}, "zh": {"chunks": 3000.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.507958, "cps": 23.4375, "mean_p": 0.305872, "boundary_entropy": 0.490163, "bpb": 3.341119}, "my": {"chunks": 2146.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.113257, "cps": 16.765625, "mean_p": 0.268346, "boundary_entropy": 0.548512, "bpb": 1.181481}}, "bpb": {"en": 2.1932, "de": 2.42804, "vi": 2.04535, "tr": 2.42982, "fi": 2.62899, "ru": 1.57324, "el": 1.66257, "hi": 1.32264, "ta": 1.11226, "th": 1.26783, "zh": 3.34112, "my": 1.18148}, "gini_cpc": 0.177833, "gini_cps": 0.069647, "bpic": 10.275261, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9626220378826187, "vi": 0.9057173324542986, "tr": 1.0150004126950736, "fi": 0.9951432360999947, "ru": 0.8777409321920664, "el": 0.8438364528411806, "hi": 0.9920707277205777, "ta": 0.8403005864181464, "th": 0.8492467468525866, "zh": 2.696764254797605, "my": 0.6012865820885084}, "t": 162.1}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.19962, "de": 2.43483, "vi": 2.04668, "tr": 2.43537, "fi": 2.63775, "ru": 1.55026, "el": 1.65369, "hi": 1.31839, "ta": 1.11815, "th": 1.28428, "zh": 3.34039, "my": 1.1898}, "bytes_per_chunk": {"en": 5.448, "de": 5.472, "vi": 7.528, "tr": 5.451, "fi": 5.163, "ru": 10.493, "el": 11.045, "hi": 12.983, "ta": 16.916, "th": 16.836, "zh": 5.474, "my": 23.137}, "bpb_mean": 1.9341, "t": 163.4}
{"event": "train", "step": 2800, "phase": 2, "objective": "parity_a", "lm_loss": 1.47775, "bpb": 2.13195, "chunk_loss": 0.030279, "F": 0.14532, "G": 0.21801, "chunks": 148.81, "bytes_per_chunk": 6.881, "boundary_entropy": 0.32742, "router_noise": 0.0, "grad_norm": 0.8044, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 9.2272, "alpha": 0.03, "t": 163.4}
{"event": "train", "step": 2850, "phase": 2, "objective": "parity_a", "lm_loss": 1.44263, "bpb": 2.08127, "chunk_loss": 0.030463, "F": 0.14694, "G": 0.21315, "chunks": 150.47, "bytes_per_chunk": 6.805, "boundary_entropy": 0.31909, "router_noise": 0.0, "grad_norm": 0.6859, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.3574, "alpha": 0.03, "t": 165.4}
{"event": "train", "step": 2900, "phase": 2, "objective": "parity_a", "lm_loss": 1.46141, "bpb": 2.10837, "chunk_loss": 0.030308, "F": 0.14859, "G": 0.21153, "chunks": 152.16, "bytes_per_chunk": 6.73, "boundary_entropy": 0.30876, "router_noise": 0.0, "grad_norm": 0.7245, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.9121, "alpha": 0.03, "t": 167.4}
{"event": "train", "step": 2950, "phase": 2, "objective": "parity_a", "lm_loss": 1.36356, "bpb": 1.96719, "chunk_loss": 0.030646, "F": 0.14212, "G": 0.21963, "chunks": 145.53, "bytes_per_chunk": 7.036, "boundary_entropy": 0.33761, "router_noise": 0.0, "grad_norm": 0.6836, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.8268, "alpha": 0.03, "t": 169.4}
{"event": "probe", "step": 3000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3314.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190427, "cps": 25.890625, "mean_p": 0.2182, "boundary_entropy": 0.249511, "bpb": 2.164874}, "de": {"chunks": 3675.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17898, "cps": 28.710938, "mean_p": 0.205822, "boundary_entropy": 0.241527, "bpb": 2.396343}, "vi": {"chunks": 3200.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.173583, "cps": 25.0, "mean_p": 0.197059, "boundary_entropy": 0.310851, "bpb": 2.028981}, "tr": {"chunks": 3321.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18345, "cps": 25.945312, "mean_p": 0.205819, "boundary_entropy": 0.263782, "bpb": 2.390457}, "fi": {"chunks": 3488.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.185423, "cps": 27.25, "mean_p": 0.221878, "boundary_entropy": 0.275628, "bpb": 2.5799}, "ru": {"chunks": 3286.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.173404, "cps": 25.671875, "mean_p": 0.161023, "boundary_entropy": 0.287982, "bpb": 1.548433}, "el": {"chunks": 2572.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.125323, "cps": 20.09375, "mean_p": 0.151504, "boundary_entropy": 0.32374, "bpb": 1.641226}, "hi": {"chunks": 2589.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.153878, "cps": 20.226562, "mean_p": 0.192562, "boundary_entropy": 0.410616, "bpb": 1.305956}, "ta": {"chunks": 2505.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.133993, "cps": 19.570312, "mean_p": 0.179317, "boundary_entropy": 0.403233, "bpb": 1.073916}, "th": {"chunks": 2829.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.175878, "cps": 22.101562, "mean_p": 0.192485, "boundary_entropy": 0.418918, "bpb": 1.267057}, "zh": {"chunks": 3002.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.508297, "cps": 23.453125, "mean_p": 0.308675, "boundary_entropy": 0.494739, "bpb": 3.314805}, "my": {"chunks": 2427.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.128087, "cps": 18.960938, "mean_p": 0.263204, "boundary_entropy": 0.533337, "bpb": 1.175761}}, "bpb": {"en": 2.16487, "de": 2.39634, "vi": 2.02898, "tr": 2.39046, "fi": 2.5799, "ru": 1.54843, "el": 1.64123, "hi": 1.30596, "ta": 1.07392, "th": 1.26706, "zh": 3.3148, "my": 1.17576}, "gini_cpc": 0.191162, "gini_cps": 0.075978, "bpic": 10.599343, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9398889686417033, "vi": 0.9115457120787958, "tr": 0.963362951175482, "fi": 0.9737247498999594, "ru": 0.9106048537984691, "el": 0.658114917714088, "hi": 0.8080692743020962, "ta": 0.7036454388112189, "th": 0.9235990942055153, "zh": 2.6692476029102896, "my": 0.6726327614667924}, "t": 172.3}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.17284, "de": 2.39968, "vi": 2.02904, "tr": 2.39873, "fi": 2.5896, "ru": 1.5278, "el": 1.63256, "hi": 1.30285, "ta": 1.08033, "th": 1.28429, "zh": 3.31451, "my": 1.18444}, "bytes_per_chunk": {"en": 5.38, "de": 5.625, "vi": 7.202, "tr": 5.696, "fi": 5.217, "ru": 10.14, "el": 14.001, "hi": 15.297, "ta": 20.351, "th": 16.115, "zh": 5.681, "my": 20.139}, "bpb_mean": 1.90972, "t": 173.6}
{"event": "train", "step": 3000, "phase": 2, "objective": "parity_a", "lm_loss": 1.444, "bpb": 2.08326, "chunk_loss": 0.030184, "F": 0.13898, "G": 0.2113, "chunks": 142.31, "bytes_per_chunk": 7.195, "boundary_entropy": 0.32971, "router_noise": 0.0, "grad_norm": 0.7176, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 9.5059, "alpha": 0.03, "t": 173.6}
{"event": "train", "step": 3050, "phase": 2, "objective": "parity_a", "lm_loss": 1.40116, "bpb": 2.02145, "chunk_loss": 0.030398, "F": 0.14276, "G": 0.2141, "chunks": 146.19, "bytes_per_chunk": 7.005, "boundary_entropy": 0.32277, "router_noise": 0.0, "grad_norm": 0.6609, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.6662, "alpha": 0.03, "t": 175.6}
{"event": "train", "step": 3100, "phase": 2, "objective": "parity_a", "lm_loss": 1.41289, "bpb": 2.03838, "chunk_loss": 0.030376, "F": 0.15585, "G": 0.21126, "chunks": 159.59, "bytes_per_chunk": 6.416, "boundary_entropy": 0.29962, "router_noise": 0.0, "grad_norm": 0.7271, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.4014, "alpha": 0.03, "t": 177.6}
{"event": "train", "step": 3150, "phase": 2, "objective": "parity_a", "lm_loss": 1.35922, "bpb": 1.96094, "chunk_loss": 0.030182, "F": 0.13461, "G": 0.20292, "chunks": 137.84, "bytes_per_chunk": 7.429, "boundary_entropy": 0.31182, "router_noise": 0.0, "grad_norm": 0.7018, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.7269, "alpha": 0.03, "t": 179.6}
{"event": "probe", "step": 3200, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3295.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.189335, "cps": 25.742188, "mean_p": 0.220492, "boundary_entropy": 0.25176, "bpb": 2.151858}, "de": {"chunks": 3724.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181367, "cps": 29.09375, "mean_p": 0.209344, "boundary_entropy": 0.242306, "bpb": 2.367886}, "vi": {"chunks": 2897.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.157147, "cps": 22.632812, "mean_p": 0.184754, "boundary_entropy": 0.303056, "bpb": 2.017577}, "tr": {"chunks": 3358.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185494, "cps": 26.234375, "mean_p": 0.207339, "boundary_entropy": 0.257591, "bpb": 2.372435}, "fi": {"chunks": 3560.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189251, "cps": 27.8125, "mean_p": 0.224027, "boundary_entropy": 0.268798, "bpb": 2.55567}, "ru": {"chunks": 3235.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.170712, "cps": 25.273438, "mean_p": 0.158571, "boundary_entropy": 0.277486, "bpb": 1.520018}, "el": {"chunks": 2881.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.140379, "cps": 22.507812, "mean_p": 0.151938, "boundary_entropy": 0.316462, "bpb": 1.616861}, "hi": {"chunks": 2311.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.137355, "cps": 18.054688, "mean_p": 0.184776, "boundary_entropy": 0.39599, "bpb": 1.304425}, "ta": {"chunks": 2304.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.123242, "cps": 18.0, "mean_p": 0.170907, "boundary_entropy": 0.391064, "bpb": 1.057871}, "th": {"chunks": 2812.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.174821, "cps": 21.96875, "mean_p": 0.198343, "boundary_entropy": 0.429249, "bpb": 1.254658}, "zh": {"chunks": 3163.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.535557, "cps": 24.710938, "mean_p": 0.30012, "boundary_entropy": 0.471788, "bpb": 3.277078}, "my": {"chunks": 3142.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.165822, "cps": 24.546875, "mean_p": 0.288657, "boundary_entropy": 0.556511, "bpb": 1.163403}}, "bpb": {"en": 2.15186, "de": 2.36789, "vi": 2.01758, "tr": 2.37243, "fi": 2.55567, "ru": 1.52002, "el": 1.61686, "hi": 1.30442, "ta": 1.05787, "th": 1.25466, "zh": 3.27708, "my": 1.1634}, "gini_cpc": 0.196338, "gini_cps": 0.076759, "bpic": 10.462379, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9579127777358584, "vi": 0.8299922831027113, "tr": 0.979712934173588, "fi": 0.9995552758697269, "ru": 0.9016412490340766, "el": 0.7414316385251026, "hi": 0.7254601708653978, "ta": 0.650917138426486, "th": 0.9233427688545723, "zh": 2.828618975995708, "my": 0.8758132392046214}, "t": 182.6}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.15948, "de": 2.37325, "vi": 2.0209, "tr": 2.37948, "fi": 2.56313, "ru": 1.49815, "el": 1.6082, "hi": 1.3009, "ta": 1.06517, "th": 1.2725, "zh": 3.27824, "my": 1.17149}, "bytes_per_chunk": {"en": 5.381, "de": 5.536, "vi": 8.083, "tr": 5.623, "fi": 5.118, "ru": 10.182, "el": 12.629, "hi": 18.539, "ta": 22.088, "th": 15.985, "zh": 5.331, "my": 16.414}, "bpb_mean": 1.89091, "t": 183.8}
{"event": "train", "step": 3200, "phase": 2, "objective": "parity_a", "lm_loss": 1.39009, "bpb": 2.00548, "chunk_loss": 0.030131, "F": 0.13315, "G": 0.20331, "chunks": 136.34, "bytes_per_chunk": 7.51, "boundary_entropy": 0.31989, "router_noise": 0.0, "grad_norm": 0.6532, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.085, "n_target_mean": 9.469, "alpha": 0.03, "t": 183.9}
{"event": "train", "step": 3250, "phase": 2, "objective": "parity_a", "lm_loss": 1.44881, "bpb": 2.09019, "chunk_loss": 0.030301, "F": 0.1463, "G": 0.20422, "chunks": 149.81, "bytes_per_chunk": 6.835, "boundary_entropy": 0.29945, "router_noise": 0.0, "grad_norm": 0.6722, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.943, "alpha": 0.03, "t": 185.9}
{"event": "train", "step": 3300, "phase": 2, "objective": "parity_a", "lm_loss": 1.41631, "bpb": 2.0433, "chunk_loss": 0.03028, "F": 0.15329, "G": 0.20984, "chunks": 156.97, "bytes_per_chunk": 6.524, "boundary_entropy": 0.29382, "router_noise": 0.0, "grad_norm": 0.7521, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.4233, "alpha": 0.03, "t": 187.9}
{"event": "train", "step": 3350, "phase": 2, "objective": "parity_a", "lm_loss": 1.47398, "bpb": 2.12651, "chunk_loss": 0.03041, "F": 0.15152, "G": 0.21581, "chunks": 155.16, "bytes_per_chunk": 6.6, "boundary_entropy": 0.31531, "router_noise": 0.0, "grad_norm": 0.6923, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 8.9185, "alpha": 0.03, "t": 189.9}
{"event": "probe", "step": 3400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3245.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186462, "cps": 25.351562, "mean_p": 0.215472, "boundary_entropy": 0.240124, "bpb": 2.130903}, "de": {"chunks": 3717.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181026, "cps": 29.039062, "mean_p": 0.208588, "boundary_entropy": 0.233806, "bpb": 2.344561}, "vi": {"chunks": 2876.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.156008, "cps": 22.46875, "mean_p": 0.187832, "boundary_entropy": 0.302764, "bpb": 1.994424}, "tr": {"chunks": 3378.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186599, "cps": 26.390625, "mean_p": 0.205034, "boundary_entropy": 0.248966, "bpb": 2.343073}, "fi": {"chunks": 3504.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186274, "cps": 27.375, "mean_p": 0.216208, "boundary_entropy": 0.257345, "bpb": 2.542118}, "ru": {"chunks": 3361.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.177361, "cps": 26.257812, "mean_p": 0.158144, "boundary_entropy": 0.269203, "bpb": 1.498304}, "el": {"chunks": 2880.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.14033, "cps": 22.5, "mean_p": 0.151389, "boundary_entropy": 0.310867, "bpb": 1.613827}, "hi": {"chunks": 2659.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.158039, "cps": 20.773438, "mean_p": 0.187701, "boundary_entropy": 0.400566, "bpb": 1.289548}, "ta": {"chunks": 2999.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.160417, "cps": 23.429688, "mean_p": 0.181962, "boundary_entropy": 0.393248, "bpb": 1.04788}, "th": {"chunks": 3234.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.201057, "cps": 25.265625, "mean_p": 0.202243, "boundary_entropy": 0.424068, "bpb": 1.213945}, "zh": {"chunks": 3416.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.578395, "cps": 26.6875, "mean_p": 0.300485, "boundary_entropy": 0.462973, "bpb": 3.253036}, "my": {"chunks": 2036.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.107452, "cps": 15.90625, "mean_p": 0.239142, "boundary_entropy": 0.504379, "bpb": 1.157201}}, "bpb": {"en": 2.1309, "de": 2.34456, "vi": 1.99442, "tr": 2.34307, "fi": 2.54212, "ru": 1.4983, "el": 1.61383, "hi": 1.28955, "ta": 1.04788, "th": 1.21394, "zh": 3.25304, "my": 1.1572}, "gini_cpc": 0.212324, "gini_cps": 0.074751, "bpic": 10.287656, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9708442728556691, "vi": 0.8366718529226286, "tr": 1.0007336719935156, "fi": 0.9989911322088942, "ru": 0.9511931569168723, "el": 0.7525945372875275, "hi": 0.8475643543501603, "ta": 0.8603207848311906, "th": 1.0782721031806661, "zh": 3.1019431807093882, "my": 0.5762670879640427}, "t": 192.8}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.13821, "de": 2.35136, "vi": 1.99748, "tr": 2.34883, "fi": 2.5524, "ru": 1.47687, "el": 1.60522, "hi": 1.2875, "ta": 1.05323, "th": 1.23102, "zh": 3.25464, "my": 1.16668}, "bytes_per_chunk": {"en": 5.402, "de": 5.53, "vi": 8.138, "tr": 5.602, "fi": 5.207, "ru": 9.944, "el": 12.8, "hi": 15.809, "ta": 17.498, "th": 14.095, "zh": 5.044, "my": 24.004}, "bpb_mean": 1.87195, "t": 194.1}
{"event": "train", "step": 3400, "phase": 2, "objective": "parity_a", "lm_loss": 1.3996, "bpb": 2.0192, "chunk_loss": 0.030469, "F": 0.14188, "G": 0.21291, "chunks": 145.28, "bytes_per_chunk": 7.048, "boundary_entropy": 0.3265, "router_noise": 0.0, "grad_norm": 0.6697, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 9.6778, "alpha": 0.03, "t": 194.1}
{"event": "train", "step": 3450, "phase": 2, "objective": "parity_a", "lm_loss": 1.2641, "bpb": 1.82371, "chunk_loss": 0.030115, "F": 0.13278, "G": 0.19906, "chunks": 135.97, "bytes_per_chunk": 7.531, "boundary_entropy": 0.29861, "router_noise": 0.0, "grad_norm": 0.6229, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.8297, "alpha": 0.03, "t": 196.1}
{"event": "train", "step": 3500, "phase": 2, "objective": "parity_a", "lm_loss": 1.4507, "bpb": 2.09292, "chunk_loss": 0.030225, "F": 0.15329, "G": 0.21209, "chunks": 156.97, "bytes_per_chunk": 6.524, "boundary_entropy": 0.29341, "router_noise": 0.0, "grad_norm": 0.7557, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.7543, "alpha": 0.03, "t": 198.1}
{"event": "train", "step": 3550, "phase": 2, "objective": "parity_a", "lm_loss": 1.38387, "bpb": 1.99651, "chunk_loss": 0.030215, "F": 0.13702, "G": 0.20578, "chunks": 140.31, "bytes_per_chunk": 7.298, "boundary_entropy": 0.30952, "router_noise": 0.0, "grad_norm": 0.6663, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.6795, "alpha": 0.03, "t": 200.1}
{"event": "probe", "step": 3600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3414.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196173, "cps": 26.671875, "mean_p": 0.226192, "boundary_entropy": 0.243419, "bpb": 2.113814}, "de": {"chunks": 3794.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.184776, "cps": 29.640625, "mean_p": 0.212233, "boundary_entropy": 0.229816, "bpb": 2.32869}, "vi": {"chunks": 3099.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.168104, "cps": 24.210938, "mean_p": 0.200154, "boundary_entropy": 0.304819, "bpb": 1.99715}, "tr": {"chunks": 3515.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194167, "cps": 27.460938, "mean_p": 0.212722, "boundary_entropy": 0.249777, "bpb": 2.321665}, "fi": {"chunks": 3487.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18537, "cps": 27.242188, "mean_p": 0.218327, "boundary_entropy": 0.253909, "bpb": 2.496229}, "ru": {"chunks": 3288.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.173509, "cps": 25.6875, "mean_p": 0.152677, "boundary_entropy": 0.262466, "bpb": 1.487168}, "el": {"chunks": 2958.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.144131, "cps": 23.109375, "mean_p": 0.14916, "boundary_entropy": 0.301912, "bpb": 1.586153}, "hi": {"chunks": 2512.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.149302, "cps": 19.625, "mean_p": 0.191533, "boundary_entropy": 0.40704, "bpb": 1.276255}, "ta": {"chunks": 2817.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.150682, "cps": 22.007812, "mean_p": 0.179486, "boundary_entropy": 0.396631, "bpb": 1.039974}, "th": {"chunks": 2886.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.179422, "cps": 22.546875, "mean_p": 0.198242, "boundary_entropy": 0.426464, "bpb": 1.201671}, "zh": {"chunks": 2909.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.49255, "cps": 22.726562, "mean_p": 0.28762, "boundary_entropy": 0.463902, "bpb": 3.25658}, "my": {"chunks": 2572.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.13574, "cps": 20.09375, "mean_p": 0.222172, "boundary_entropy": 0.479876, "bpb": 1.149726}}, "bpb": {"en": 2.11381, "de": 2.32869, "vi": 1.99715, "tr": 2.32166, "fi": 2.49623, "ru": 1.48717, "el": 1.58615, "hi": 1.27626, "ta": 1.03997, "th": 1.20167, "zh": 3.25658, "my": 1.14973}, "gini_cpc": 0.174769, "gini_cps": 0.06979, "bpic": 10.302569, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9419015743613713, "vi": 0.8569175505254533, "tr": 0.9897724959315505, "fi": 0.9449322410633796, "ru": 0.8844701856239943, "el": 0.7347133444606321, "hi": 0.7610709855668165, "ta": 0.7681074561537511, "th": 0.9146098303553234, "zh": 2.5107928429996127, "my": 0.6919396086523733}, "t": 203.1}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 2.12335, "de": 2.33467, "vi": 1.99932, "tr": 2.32861, "fi": 2.50596, "ru": 1.46416, "el": 1.57828, "hi": 1.27284, "ta": 1.04543, "th": 1.21835, "zh": 3.25471, "my": 1.15898}, "bytes_per_chunk": {"en": 5.172, "de": 5.4, "vi": 7.616, "tr": 5.343, "fi": 5.156, "ru": 10.047, "el": 11.946, "hi": 16.549, "ta": 17.705, "th": 16.147, "zh": 5.872, "my": 19.461}, "bpb_mean": 1.85706, "t": 204.3}
{"event": "train", "step": 3600, "phase": 2, "objective": "parity_a", "lm_loss": 1.63832, "bpb": 2.36359, "chunk_loss": 0.030518, "F": 0.16574, "G": 0.23207, "chunks": 169.72, "bytes_per_chunk": 6.034, "boundary_entropy": 0.32883, "router_noise": 0.0, "grad_norm": 0.7457, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.085, "n_target_mean": 7.8431, "alpha": 0.03, "t": 204.4}
{"event": "train", "step": 3650, "phase": 2, "objective": "parity_a", "lm_loss": 1.44693, "bpb": 2.08748, "chunk_loss": 0.030435, "F": 0.15033, "G": 0.20906, "chunks": 153.94, "bytes_per_chunk": 6.652, "boundary_entropy": 0.2983, "router_noise": 0.0, "grad_norm": 0.7722, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.856, "alpha": 0.03, "t": 206.4}
{"event": "train", "step": 3700, "phase": 2, "objective": "parity_a", "lm_loss": 1.35838, "bpb": 1.95973, "chunk_loss": 0.030615, "F": 0.13559, "G": 0.2, "chunks": 138.84, "bytes_per_chunk": 7.375, "boundary_entropy": 0.30406, "router_noise": 0.0, "grad_norm": 0.7172, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 10.295, "alpha": 0.03, "t": 208.4}
{"event": "train", "step": 3750, "phase": 2, "objective": "parity_a", "lm_loss": 1.40258, "bpb": 2.0235, "chunk_loss": 0.030416, "F": 0.14926, "G": 0.20889, "chunks": 152.84, "bytes_per_chunk": 6.7, "boundary_entropy": 0.30299, "router_noise": 0.0, "grad_norm": 0.7213, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 9.0563, "alpha": 0.03, "t": 210.5}
{"event": "probe", "step": 3800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3222.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.18514, "cps": 25.171875, "mean_p": 0.21187, "boundary_entropy": 0.229078, "bpb": 2.106782}, "de": {"chunks": 3657.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178104, "cps": 28.570312, "mean_p": 0.200101, "boundary_entropy": 0.22101, "bpb": 2.310819}, "vi": {"chunks": 2902.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.157418, "cps": 22.671875, "mean_p": 0.187282, "boundary_entropy": 0.292546, "bpb": 1.978845}, "tr": {"chunks": 3465.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191405, "cps": 27.070312, "mean_p": 0.208095, "boundary_entropy": 0.24464, "bpb": 2.299957}, "fi": {"chunks": 3470.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184467, "cps": 27.109375, "mean_p": 0.214988, "boundary_entropy": 0.250632, "bpb": 2.48034}, "ru": {"chunks": 3178.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.167704, "cps": 24.828125, "mean_p": 0.141016, "boundary_entropy": 0.238351, "bpb": 1.472302}, "el": {"chunks": 2873.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.139989, "cps": 22.445312, "mean_p": 0.146895, "boundary_entropy": 0.301079, "bpb": 1.562808}, "hi": {"chunks": 2799.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.16636, "cps": 21.867188, "mean_p": 0.18805, "boundary_entropy": 0.39875, "bpb": 1.261897}, "ta": {"chunks": 2394.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.128056, "cps": 18.703125, "mean_p": 0.167261, "boundary_entropy": 0.378751, "bpb": 1.035715}, "th": {"chunks": 2509.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.155984, "cps": 19.601562, "mean_p": 0.179389, "boundary_entropy": 0.399192, "bpb": 1.190408}, "zh": {"chunks": 3123.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.528784, "cps": 24.398438, "mean_p": 0.288151, "boundary_entropy": 0.452915, "bpb": 3.221427}, "my": {"chunks": 2093.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.11046, "cps": 16.351562, "mean_p": 0.215564, "boundary_entropy": 0.472525, "bpb": 1.150237}}, "bpb": {"en": 2.10678, "de": 2.31082, "vi": 1.97884, "tr": 2.29996, "fi": 2.48034, "ru": 1.4723, "el": 1.56281, "hi": 1.2619, "ta": 1.03571, "th": 1.19041, "zh": 3.22143, "my": 1.15024}, "gini_cpc": 0.20684, "gini_cps": 0.08601, "bpic": 10.754687, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9619912842305628, "vi": 0.8502621571892587, "tr": 1.0338350969329138, "fi": 0.9963597510747654, "ru": 0.9058228308348442, "el": 0.7561245950970574, "hi": 0.8985586094117061, "ta": 0.6916673265788211, "th": 0.8425160445931733, "zh": 2.8561244411988946, "my": 0.5966291062592555}, "t": 213.5}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 2.11597, "de": 2.317, "vi": 1.97961, "tr": 2.30556, "fi": 2.48865, "ru": 1.45138, "el": 1.55416, "hi": 1.25802, "ta": 1.04163, "th": 1.20657, "zh": 3.21932, "my": 1.15903}, "bytes_per_chunk": {"en": 5.442, "de": 5.638, "vi": 8.039, "tr": 5.441, "fi": 5.239, "ru": 10.417, "el": 12.642, "hi": 15.693, "ta": 21.456, "th": 18.283, "zh": 5.486, "my": 24.239}, "bpb_mean": 1.84141, "t": 214.7}
{"event": "train", "step": 3800, "phase": 2, "objective": "parity_a", "lm_loss": 1.47615, "bpb": 2.12964, "chunk_loss": 0.030246, "F": 0.14191, "G": 0.20725, "chunks": 145.31, "bytes_per_chunk": 7.047, "boundary_entropy": 0.31313, "router_noise": 0.0, "grad_norm": 0.7026, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.086, "n_target_mean": 9.1031, "alpha": 0.03, "t": 214.8}
{"event": "train", "step": 3850, "phase": 2, "objective": "parity_a", "lm_loss": 1.47458, "bpb": 2.12737, "chunk_loss": 0.030731, "F": 0.15421, "G": 0.21944, "chunks": 157.91, "bytes_per_chunk": 6.485, "boundary_entropy": 0.31919, "router_noise": 0.0, "grad_norm": 0.7046, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 9.1437, "alpha": 0.03, "t": 216.8}
{"event": "train", "step": 3900, "phase": 2, "objective": "parity_a", "lm_loss": 1.47699, "bpb": 2.13085, "chunk_loss": 0.030419, "F": 0.16022, "G": 0.21072, "chunks": 164.06, "bytes_per_chunk": 6.242, "boundary_entropy": 0.2756, "router_noise": 0.0, "grad_norm": 0.7433, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 8.4141, "alpha": 0.03, "t": 218.9}
{"event": "train", "step": 3950, "phase": 2, "objective": "parity_a", "lm_loss": 1.55241, "bpb": 2.23965, "chunk_loss": 0.030824, "F": 0.15656, "G": 0.22392, "chunks": 160.31, "bytes_per_chunk": 6.388, "boundary_entropy": 0.32556, "router_noise": 0.0, "grad_norm": 0.7198, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 9.0542, "alpha": 0.03, "t": 221.0}
{"event": "probe", "step": 4000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3281.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188531, "cps": 25.632812, "mean_p": 0.216686, "boundary_entropy": 0.230042, "bpb": 2.09002}, "de": {"chunks": 3670.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178737, "cps": 28.671875, "mean_p": 0.20202, "boundary_entropy": 0.217694, "bpb": 2.304034}, "vi": {"chunks": 2784.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.151017, "cps": 21.75, "mean_p": 0.180203, "boundary_entropy": 0.287161, "bpb": 1.97323}, "tr": {"chunks": 3322.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183505, "cps": 25.953125, "mean_p": 0.201228, "boundary_entropy": 0.234336, "bpb": 2.288922}, "fi": {"chunks": 3445.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.183138, "cps": 26.914062, "mean_p": 0.21132, "boundary_entropy": 0.240597, "bpb": 2.458505}, "ru": {"chunks": 3205.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.169129, "cps": 25.039062, "mean_p": 0.144039, "boundary_entropy": 0.241046, "bpb": 1.461644}, "el": {"chunks": 2586.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.126005, "cps": 20.203125, "mean_p": 0.141572, "boundary_entropy": 0.295086, "bpb": 1.554071}, "hi": {"chunks": 2804.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.166657, "cps": 21.90625, "mean_p": 0.187515, "boundary_entropy": 0.396526, "bpb": 1.249934}, "ta": {"chunks": 2436.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.130302, "cps": 19.03125, "mean_p": 0.169724, "boundary_entropy": 0.381062, "bpb": 1.029348}, "th": {"chunks": 3198.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.198819, "cps": 24.984375, "mean_p": 0.198548, "boundary_entropy": 0.421816, "bpb": 1.175374}, "zh": {"chunks": 3228.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.546563, "cps": 25.21875, "mean_p": 0.288643, "boundary_entropy": 0.445217, "bpb": 3.196183}, "my": {"chunks": 2123.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.112043, "cps": 16.585938, "mean_p": 0.245791, "boundary_entropy": 0.507625, "bpb": 1.142408}}, "bpb": {"en": 2.09002, "de": 2.30403, "vi": 1.97323, "tr": 2.28892, "fi": 2.4585, "ru": 1.46164, "el": 1.55407, "hi": 1.24993, "ta": 1.02935, "th": 1.17537, "zh": 3.19618, "my": 1.14241}, "gini_cpc": 0.216473, "gini_cps": 0.080867, "bpic": 10.636356, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9480506643634433, "vi": 0.8010211417712109, "tr": 0.973345368049801, "fi": 0.9713935883909516, "ru": 0.8970914331253985, "el": 0.668352482016185, "hi": 0.8839767167712884, "ta": 0.6911458494240833, "th": 1.0545696878354165, "zh": 2.8990651363900537, "my": 0.5942983273106582}, "t": 223.9}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.09855, "de": 2.30822, "vi": 1.97538, "tr": 2.29545, "fi": 2.46782, "ru": 1.43929, "el": 1.54652, "hi": 1.24814, "ta": 1.03585, "th": 1.19159, "zh": 3.19405, "my": 1.15132}, "bytes_per_chunk": {"en": 5.329, "de": 5.613, "vi": 8.428, "tr": 5.674, "fi": 5.247, "ru": 10.33, "el": 14.003, "hi": 15.34, "ta": 21.51, "th": 14.465, "zh": 5.372, "my": 24.295}, "bpb_mean": 1.82935, "t": 225.2}
{"event": "train", "step": 4000, "phase": 2, "objective": "parity_a", "lm_loss": 1.34455, "bpb": 1.93977, "chunk_loss": 0.030182, "F": 0.14914, "G": 0.20265, "chunks": 152.72, "bytes_per_chunk": 6.705, "boundary_entropy": 0.28201, "router_noise": 0.0, "grad_norm": 0.7144, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.086, "n_target_mean": 8.7169, "alpha": 0.03, "t": 225.3}
{"event": "train", "step": 4050, "phase": 2, "objective": "parity_a", "lm_loss": 1.5414, "bpb": 2.22377, "chunk_loss": 0.030247, "F": 0.15536, "G": 0.21515, "chunks": 159.09, "bytes_per_chunk": 6.436, "boundary_entropy": 0.30438, "router_noise": 0.0, "grad_norm": 0.6973, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 8.3032, "alpha": 0.03, "t": 227.3}
{"event": "train", "step": 4100, "phase": 2, "objective": "parity_a", "lm_loss": 1.40818, "bpb": 2.03158, "chunk_loss": 0.030379, "F": 0.15268, "G": 0.20537, "chunks": 156.34, "bytes_per_chunk": 6.55, "boundary_entropy": 0.28192, "router_noise": 0.0, "grad_norm": 0.7056, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 8.5769, "alpha": 0.03, "t": 229.4}
{"event": "train", "step": 4150, "phase": 2, "objective": "parity_a", "lm_loss": 1.39392, "bpb": 2.011, "chunk_loss": 0.030367, "F": 0.15826, "G": 0.2057, "chunks": 162.06, "bytes_per_chunk": 6.319, "boundary_entropy": 0.26127, "router_noise": 0.0, "grad_norm": 0.7598, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.041, "n_target_mean": 8.3506, "alpha": 0.03, "t": 231.4}
{"event": "probe", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3273.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188071, "cps": 25.570312, "mean_p": 0.213148, "boundary_entropy": 0.221261, "bpb": 2.078355}, "de": {"chunks": 3680.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.179224, "cps": 28.75, "mean_p": 0.201266, "boundary_entropy": 0.212343, "bpb": 2.277945}, "vi": {"chunks": 2877.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.156062, "cps": 22.476562, "mean_p": 0.18361, "boundary_entropy": 0.287224, "bpb": 1.960366}, "tr": {"chunks": 3289.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181683, "cps": 25.695312, "mean_p": 0.19699, "boundary_entropy": 0.227683, "bpb": 2.268921}, "fi": {"chunks": 3534.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187869, "cps": 27.609375, "mean_p": 0.215314, "boundary_entropy": 0.241617, "bpb": 2.441142}, "ru": {"chunks": 3244.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.171187, "cps": 25.34375, "mean_p": 0.142659, "boundary_entropy": 0.232147, "bpb": 1.445345}, "el": {"chunks": 3005.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.146421, "cps": 23.476562, "mean_p": 0.147429, "boundary_entropy": 0.291147, "bpb": 1.534554}, "hi": {"chunks": 3055.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.181575, "cps": 23.867188, "mean_p": 0.187076, "boundary_entropy": 0.393727, "bpb": 1.246343}, "ta": {"chunks": 2760.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.147633, "cps": 21.5625, "mean_p": 0.183435, "boundary_entropy": 0.396539, "bpb": 1.012148}, "th": {"chunks": 2905.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.180603, "cps": 22.695312, "mean_p": 0.189153, "boundary_entropy": 0.410634, "bpb": 1.165999}, "zh": {"chunks": 3123.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.528784, "cps": 24.398438, "mean_p": 0.282084, "boundary_entropy": 0.441325, "bpb": 3.193605}, "my": {"chunks": 2161.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.114049, "cps": 16.882812, "mean_p": 0.227713, "boundary_entropy": 0.487923, "bpb": 1.136411}}, "bpb": {"en": 2.07835, "de": 2.27794, "vi": 1.96037, "tr": 2.26892, "fi": 2.44114, "ru": 1.44534, "el": 1.53455, "hi": 1.24634, "ta": 1.01215, "th": 1.166, "zh": 3.1936, "my": 1.13641}, "gini_cpc": 0.189691, "gini_cps": 0.065897, "bpic": 10.398878, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9529574868257725, "vi": 0.8298026832453412, "tr": 0.9660318313988299, "fi": 0.9989247569081824, "ru": 0.9102270677091774, "el": 0.7785414615861224, "hi": 0.9654599362881227, "ta": 0.7849856677050114, "th": 0.9602917247311559, "zh": 2.811620210676089, "my": 0.6064144002003071}, "t": 234.4}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 2.08845, "de": 2.28341, "vi": 1.96228, "tr": 2.27511, "fi": 2.44791, "ru": 1.42426, "el": 1.52591, "hi": 1.24215, "ta": 1.01969, "th": 1.18191, "zh": 3.19042, "my": 1.14475}, "bytes_per_chunk": {"en": 5.374, "de": 5.573, "vi": 8.144, "tr": 5.804, "fi": 5.157, "ru": 10.18, "el": 12.114, "hi": 14.325, "ta": 18.921, "th": 16.186, "zh": 5.676, "my": 22.197}, "bpb_mean": 1.81552, "t": 235.7}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584, "t": 235.7}
{"event": "train", "step": 4200, "phase": 3, "objective": "none", "lm_loss": 1.2023, "bpb": 1.73455, "chunk_loss": 0.0, "F": 0.12552, "G": 0.20139, "chunks": 128.53, "bytes_per_chunk": 7.967, "boundary_entropy": 0.31721, "router_noise": 0.0, "grad_norm": 0.6503, "lr": 9.12e-05, "router_trainable": false, "sec_per_step": 0.086, "alpha": 0.0, "t": 235.8}
{"event": "train", "step": 4250, "phase": 3, "objective": "none", "lm_loss": 1.25538, "bpb": 1.81114, "chunk_loss": 0.0, "F": 0.1431, "G": 0.204, "chunks": 146.53, "bytes_per_chunk": 6.988, "boundary_entropy": 0.29223, "router_noise": 0.0, "grad_norm": 0.6945, "lr": 8.81e-05, "router_trainable": false, "sec_per_step": 0.041, "alpha": 0.0, "t": 237.8}
{"event": "train", "step": 4300, "phase": 3, "objective": "none", "lm_loss": 1.48784, "bpb": 2.14649, "chunk_loss": 0.0, "F": 0.16211, "G": 0.21923, "chunks": 166.0, "bytes_per_chunk": 6.169, "boundary_entropy": 0.28768, "router_noise": 0.0, "grad_norm": 0.7394, "lr": 8.5e-05, "router_trainable": false, "sec_per_step": 0.041, "alpha": 0.0, "t": 239.8}
{"event": "train", "step": 4350, "phase": 3, "objective": "none", "lm_loss": 1.47316, "bpb": 2.12532, "chunk_loss": 0.0, "F": 0.16232, "G": 0.21658, "chunks": 166.22, "bytes_per_chunk": 6.161, "boundary_entropy": 0.28024, "router_noise": 0.0, "grad_norm": 0.7805, "lr": 8.21e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 241.8}
{"event": "probe", "step": 4400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3470.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199391, "cps": 27.109375, "mean_p": 0.230147, "boundary_entropy": 0.239502, "bpb": 2.064361}, "de": {"chunks": 3877.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188818, "cps": 30.289062, "mean_p": 0.217665, "boundary_entropy": 0.227418, "bpb": 2.262963}, "vi": {"chunks": 3088.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.167507, "cps": 24.125, "mean_p": 0.199953, "boundary_entropy": 0.300845, "bpb": 1.950059}, "tr": {"chunks": 3573.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197371, "cps": 27.914062, "mean_p": 0.216947, "boundary_entropy": 0.244364, "bpb": 2.246195}, "fi": {"chunks": 3719.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.197703, "cps": 29.054688, "mean_p": 0.22945, "boundary_entropy": 0.253182, "bpb": 2.421091}, "ru": {"chunks": 3455.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.182322, "cps": 26.992188, "mean_p": 0.157859, "boundary_entropy": 0.255562, "bpb": 1.429212}, "el": {"chunks": 3754.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.182917, "cps": 29.328125, "mean_p": 0.166968, "boundary_entropy": 0.315481, "bpb": 1.51441}, "hi": {"chunks": 3262.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.193878, "cps": 25.484375, "mean_p": 0.201674, "boundary_entropy": 0.412681, "bpb": 1.230259}, "ta": {"chunks": 3759.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.20107, "cps": 29.367188, "mean_p": 0.198261, "boundary_entropy": 0.409066, "bpb": 1.005944}, "th": {"chunks": 3708.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.230525, "cps": 28.96875, "mean_p": 0.216472, "boundary_entropy": 0.44259, "bpb": 1.154805}, "zh": {"chunks": 3309.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.560278, "cps": 25.851562, "mean_p": 0.29231, "boundary_entropy": 0.448101, "bpb": 3.171309}, "my": {"chunks": 3477.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.183502, "cps": 27.164062, "mean_p": 0.262789, "boundary_entropy": 0.518323, "bpb": 1.120675}}, "bpb": {"en": 2.06436, "de": 2.26296, "vi": 1.95006, "tr": 2.24619, "fi": 2.42109, "ru": 1.42921, "el": 1.51441, "hi": 1.23026, "ta": 1.00594, "th": 1.15481, "zh": 3.17131, "my": 1.12068}, "gini_cpc": 0.154886, "gini_cps": 0.036114, "bpic": 9.040564, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9469739651542867, "vi": 0.8400957644625676, "tr": 0.9898676019912314, "fi": 0.9915370352468672, "ru": 0.914394242394288, "el": 0.9173774668499431, "hi": 0.9723520389156043, "ta": 1.0084201188038227, "th": 1.1561476629469345, "zh": 2.8099459739570274, "my": 0.9203138545718025}, "t": 244.8}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 2.07367, "de": 2.26813, "vi": 1.95217, "tr": 2.25161, "fi": 2.42842, "ru": 1.40747, "el": 1.50561, "hi": 1.22688, "ta": 1.01204, "th": 1.17197, "zh": 3.16899, "my": 1.12953}, "bytes_per_chunk": {"en": 5.015, "de": 5.244, "vi": 7.571, "tr": 5.272, "fi": 4.867, "ru": 9.642, "el": 9.674, "hi": 12.709, "ta": 13.781, "th": 12.575, "zh": 5.258, "my": 14.114}, "bpb_mean": 1.79971, "t": 246.1}
{"event": "train", "step": 4400, "phase": 3, "objective": "none", "lm_loss": 1.48914, "bpb": 2.14838, "chunk_loss": 0.0, "F": 0.17191, "G": 0.23112, "chunks": 176.03, "bytes_per_chunk": 5.817, "boundary_entropy": 0.30863, "router_noise": 0.0, "grad_norm": 0.8476, "lr": 7.92e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 246.1}
{"event": "train", "step": 4450, "phase": 3, "objective": "none", "lm_loss": 1.49896, "bpb": 2.16254, "chunk_loss": 0.0, "F": 0.15762, "G": 0.22623, "chunks": 161.41, "bytes_per_chunk": 6.344, "boundary_entropy": 0.32525, "router_noise": 0.0, "grad_norm": 0.7163, "lr": 7.63e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 248.1}
{"event": "train", "step": 4500, "phase": 3, "objective": "none", "lm_loss": 1.42333, "bpb": 2.05343, "chunk_loss": 0.0, "F": 0.16486, "G": 0.2208, "chunks": 168.81, "bytes_per_chunk": 6.066, "boundary_entropy": 0.28374, "router_noise": 0.0, "grad_norm": 0.786, "lr": 7.36e-05, "router_trainable": false, "sec_per_step": 0.041, "alpha": 0.0, "t": 250.2}
{"event": "train", "step": 4550, "phase": 3, "objective": "none", "lm_loss": 1.23124, "bpb": 1.77631, "chunk_loss": 0.0, "F": 0.14554, "G": 0.20416, "chunks": 149.03, "bytes_per_chunk": 6.871, "boundary_entropy": 0.28356, "router_noise": 0.0, "grad_norm": 0.7103, "lr": 7.09e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 252.2}
{"event": "probe", "step": 4600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3447.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198069, "cps": 26.929688, "mean_p": 0.228233, "boundary_entropy": 0.235689, "bpb": 2.055622}, "de": {"chunks": 3885.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189208, "cps": 30.351562, "mean_p": 0.217209, "boundary_entropy": 0.225439, "bpb": 2.250983}, "vi": {"chunks": 3162.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.171522, "cps": 24.703125, "mean_p": 0.202865, "boundary_entropy": 0.303518, "bpb": 1.945355}, "tr": {"chunks": 3477.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.192068, "cps": 27.164062, "mean_p": 0.211214, "boundary_entropy": 0.241723, "bpb": 2.235328}, "fi": {"chunks": 3606.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.191696, "cps": 28.171875, "mean_p": 0.224241, "boundary_entropy": 0.250937, "bpb": 2.400892}, "ru": {"chunks": 3523.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.18591, "cps": 27.523438, "mean_p": 0.158549, "boundary_entropy": 0.253412, "bpb": 1.421734}, "el": {"chunks": 3778.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.184086, "cps": 29.515625, "mean_p": 0.166528, "boundary_entropy": 0.314598, "bpb": 1.50546}, "hi": {"chunks": 2972.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.176642, "cps": 23.21875, "mean_p": 0.200892, "boundary_entropy": 0.413141, "bpb": 1.22584}, "ta": {"chunks": 2703.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.144584, "cps": 21.117188, "mean_p": 0.188449, "boundary_entropy": 0.410099, "bpb": 1.005854}, "th": {"chunks": 3857.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.239789, "cps": 30.132812, "mean_p": 0.216983, "boundary_entropy": 0.443944, "bpb": 1.148061}, "zh": {"chunks": 3222.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.545547, "cps": 25.171875, "mean_p": 0.29197, "boundary_entropy": 0.452674, "bpb": 3.150825}, "my": {"chunks": 2711.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.143076, "cps": 21.179688, "mean_p": 0.248525, "boundary_entropy": 0.50914, "bpb": 1.116087}}, "bpb": {"en": 2.05562, "de": 2.25098, "vi": 1.94535, "tr": 2.23533, "fi": 2.40089, "ru": 1.42173, "el": 1.50546, "hi": 1.22584, "ta": 1.00585, "th": 1.14806, "zh": 3.15082, "my": 1.11609}, "gini_cpc": 0.181261, "gini_cps": 0.066329, "bpic": 9.512951, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9552596921290429, "vi": 0.8659674348019751, "tr": 0.9696990631935131, "fi": 0.9678246423506718, "ru": 0.9386123530613979, "el": 0.9294027329799207, "hi": 0.8918186885165342, "ta": 0.7299673122970985, "th": 1.210629940538366, "zh": 2.7543233902063573, "my": 0.7223521642209644}, "t": 255.1}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 2.06424, "de": 2.25719, "vi": 1.94615, "tr": 2.24132, "fi": 2.41033, "ru": 1.4005, "el": 1.49596, "hi": 1.22275, "ta": 1.01212, "th": 1.16428, "zh": 3.14756, "my": 1.12448}, "bytes_per_chunk": {"en": 5.081, "de": 5.259, "vi": 7.438, "tr": 5.482, "fi": 5.039, "ru": 9.505, "el": 9.646, "hi": 13.702, "ta": 18.763, "th": 11.843, "zh": 5.347, "my": 18.347}, "bpb_mean": 1.79057, "t": 256.4}
{"event": "train", "step": 4600, "phase": 3, "objective": "none", "lm_loss": 1.30649, "bpb": 1.88486, "chunk_loss": 0.0, "F": 0.14142, "G": 0.21524, "chunks": 144.81, "bytes_per_chunk": 7.071, "boundary_entropy": 0.32195, "router_noise": 0.0, "grad_norm": 0.724, "lr": 6.82e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 256.4}
{"event": "train", "step": 4650, "phase": 3, "objective": "none", "lm_loss": 1.41439, "bpb": 2.04053, "chunk_loss": 0.0, "F": 0.15384, "G": 0.22512, "chunks": 157.53, "bytes_per_chunk": 6.5, "boundary_entropy": 0.32472, "router_noise": 0.0, "grad_norm": 0.7362, "lr": 6.57e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 258.4}
{"event": "train", "step": 4700, "phase": 3, "objective": "none", "lm_loss": 1.32555, "bpb": 1.91237, "chunk_loss": 0.0, "F": 0.1557, "G": 0.2159, "chunks": 159.44, "bytes_per_chunk": 6.423, "boundary_entropy": 0.29322, "router_noise": 0.0, "grad_norm": 0.7544, "lr": 6.32e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 260.4}
{"event": "train", "step": 4750, "phase": 3, "objective": "none", "lm_loss": 1.33706, "bpb": 1.92897, "chunk_loss": 0.0, "F": 0.15939, "G": 0.22665, "chunks": 163.22, "bytes_per_chunk": 6.274, "boundary_entropy": 0.30339, "router_noise": 0.0, "grad_norm": 0.7226, "lr": 6.08e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 262.4}
{"event": "probe", "step": 4800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3406.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195713, "cps": 26.609375, "mean_p": 0.226614, "boundary_entropy": 0.234139, "bpb": 2.049012}, "de": {"chunks": 3800.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.185068, "cps": 29.6875, "mean_p": 0.21318, "boundary_entropy": 0.225533, "bpb": 2.241199}, "vi": {"chunks": 3095.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.167887, "cps": 24.179688, "mean_p": 0.20071, "boundary_entropy": 0.302314, "bpb": 1.939665}, "tr": {"chunks": 3453.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.190742, "cps": 26.976562, "mean_p": 0.212936, "boundary_entropy": 0.243181, "bpb": 2.222297}, "fi": {"chunks": 3523.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187284, "cps": 27.523438, "mean_p": 0.218631, "boundary_entropy": 0.246307, "bpb": 2.389954}, "ru": {"chunks": 3490.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.184169, "cps": 27.265625, "mean_p": 0.158878, "boundary_entropy": 0.255375, "bpb": 1.40815}, "el": {"chunks": 3849.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.187546, "cps": 30.070312, "mean_p": 0.169449, "boundary_entropy": 0.31863, "bpb": 1.490158}, "hi": {"chunks": 3203.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.190371, "cps": 25.023438, "mean_p": 0.201966, "boundary_entropy": 0.413987, "bpb": 1.219617}, "ta": {"chunks": 3005.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.160738, "cps": 23.476562, "mean_p": 0.190709, "boundary_entropy": 0.40564, "bpb": 0.999854}, "th": {"chunks": 3344.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.207896, "cps": 26.125, "mean_p": 0.211694, "boundary_entropy": 0.440344, "bpb": 1.147755}, "zh": {"chunks": 3397.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.575178, "cps": 26.539062, "mean_p": 0.294812, "boundary_entropy": 0.448739, "bpb": 3.146116}, "my": {"chunks": 2781.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.14677, "cps": 21.726562, "mean_p": 0.244459, "boundary_entropy": 0.504893, "bpb": 1.111901}}, "bpb": {"en": 2.04901, "de": 2.2412, "vi": 1.93967, "tr": 2.2223, "fi": 2.38995, "ru": 1.40815, "el": 1.49016, "hi": 1.21962, "ta": 0.99985, "th": 1.14776, "zh": 3.14612, "my": 1.1119}, "gini_cpc": 0.173839, "gini_cps": 0.04889, "bpic": 9.512244, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9456069728808865, "vi": 0.8578216204878483, "tr": 0.9745979723451721, "fi": 0.9569301460368528, "ru": 0.9410131430333247, "el": 0.9582670220132267, "hi": 0.9727054346014681, "ta": 0.8212936848068092, "th": 1.0622449622171994, "zh": 2.938878155498981, "my": 0.7499237180447871}, "t": 265.3}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 2.05784, "de": 2.24781, "vi": 1.94068, "tr": 2.22722, "fi": 2.39825, "ru": 1.38702, "el": 1.48112, "hi": 1.21529, "ta": 1.00604, "th": 1.16292, "zh": 3.1411, "my": 1.12016}, "bytes_per_chunk": {"en": 5.15, "de": 5.372, "vi": 7.584, "tr": 5.435, "fi": 5.141, "ru": 9.575, "el": 9.537, "hi": 13.007, "ta": 17.136, "th": 13.641, "zh": 5.13, "my": 18.085}, "bpb_mean": 1.78212, "t": 266.6}
{"event": "train", "step": 4800, "phase": 3, "objective": "none", "lm_loss": 1.26385, "bpb": 1.82335, "chunk_loss": 0.0, "F": 0.14639, "G": 0.21305, "chunks": 149.91, "bytes_per_chunk": 6.831, "boundary_entropy": 0.30376, "router_noise": 0.0, "grad_norm": 0.7059, "lr": 5.85e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 266.6}
{"event": "train", "step": 4850, "phase": 3, "objective": "none", "lm_loss": 1.49256, "bpb": 2.15331, "chunk_loss": 0.0, "F": 0.17004, "G": 0.22868, "chunks": 174.12, "bytes_per_chunk": 5.881, "boundary_entropy": 0.29873, "router_noise": 0.0, "grad_norm": 0.781, "lr": 5.62e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 268.6}
{"event": "train", "step": 4900, "phase": 3, "objective": "none", "lm_loss": 1.40828, "bpb": 2.03173, "chunk_loss": 0.0, "F": 0.15509, "G": 0.22769, "chunks": 158.81, "bytes_per_chunk": 6.448, "boundary_entropy": 0.33496, "router_noise": 0.0, "grad_norm": 0.7213, "lr": 5.41e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 270.6}
{"event": "train", "step": 4950, "phase": 3, "objective": "none", "lm_loss": 1.51752, "bpb": 2.18931, "chunk_loss": 0.0, "F": 0.17856, "G": 0.2337, "chunks": 182.84, "bytes_per_chunk": 5.6, "boundary_entropy": 0.29944, "router_noise": 0.0, "grad_norm": 0.7746, "lr": 5.2e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 272.6}
{"event": "probe", "step": 5000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3514.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.201919, "cps": 27.453125, "mean_p": 0.231678, "boundary_entropy": 0.239927, "bpb": 2.037868}, "de": {"chunks": 3899.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189889, "cps": 30.460938, "mean_p": 0.219049, "boundary_entropy": 0.229466, "bpb": 2.226204}, "vi": {"chunks": 3253.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.176458, "cps": 25.414062, "mean_p": 0.211684, "boundary_entropy": 0.311983, "bpb": 1.935036}, "tr": {"chunks": 3626.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200298, "cps": 28.328125, "mean_p": 0.22113, "boundary_entropy": 0.249202, "bpb": 2.212631}, "fi": {"chunks": 3743.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.198979, "cps": 29.242188, "mean_p": 0.232853, "boundary_entropy": 0.256801, "bpb": 2.372594}, "ru": {"chunks": 3542.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.186913, "cps": 27.671875, "mean_p": 0.162214, "boundary_entropy": 0.260698, "bpb": 1.399952}, "el": {"chunks": 3944.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.192175, "cps": 30.8125, "mean_p": 0.17081, "boundary_entropy": 0.318197, "bpb": 1.478978}, "hi": {"chunks": 3585.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.213076, "cps": 28.007812, "mean_p": 0.210971, "boundary_entropy": 0.426483, "bpb": 1.215791}, "ta": {"chunks": 3157.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.168869, "cps": 24.664062, "mean_p": 0.197852, "boundary_entropy": 0.41647, "bpb": 0.993418}, "th": {"chunks": 3554.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.220951, "cps": 27.765625, "mean_p": 0.214917, "boundary_entropy": 0.443545, "bpb": 1.136811}, "zh": {"chunks": 3373.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.571114, "cps": 26.351562, "mean_p": 0.296066, "boundary_entropy": 0.452308, "bpb": 3.138518}, "my": {"chunks": 2974.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.156956, "cps": 23.234375, "mean_p": 0.25629, "boundary_entropy": 0.519367, "bpb": 1.107533}}, "bpb": {"en": 2.03787, "de": 2.2262, "vi": 1.93504, "tr": 2.21263, "fi": 2.37259, "ru": 1.39995, "el": 1.47898, "hi": 1.21579, "ta": 0.99342, "th": 1.13681, "zh": 3.13852, "my": 1.10753}, "gini_cpc": 0.16787, "gini_cps": 0.044011, "bpic": 9.102101, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9404228893610771, "vi": 0.8739031089405022, "tr": 0.9919724515735874, "fi": 0.9854402722198987, "ru": 0.9256817584543094, "el": 0.9517402231032435, "hi": 1.0552526468103387, "ta": 0.8363180602010253, "th": 1.0942554574101127, "zh": 2.828428870748924, "my": 0.7773201953788033}, "t": 275.5}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.04595, "de": 2.23216, "vi": 1.93552, "tr": 2.21957, "fi": 2.38214, "ru": 1.37859, "el": 1.46956, "hi": 1.2125, "ta": 1.00001, "th": 1.15352, "zh": 3.13415, "my": 1.11591}, "bytes_per_chunk": {"en": 5.029, "de": 5.231, "vi": 7.241, "tr": 5.216, "fi": 4.857, "ru": 9.402, "el": 9.149, "hi": 11.417, "ta": 16.115, "th": 13.062, "zh": 5.051, "my": 16.972}, "bpb_mean": 1.7733, "t": 276.8}
{"event": "train", "step": 5000, "phase": 3, "objective": "none", "lm_loss": 1.29162, "bpb": 1.86342, "chunk_loss": 0.0, "F": 0.15582, "G": 0.22042, "chunks": 159.56, "bytes_per_chunk": 6.418, "boundary_entropy": 0.31391, "router_noise": 0.0, "grad_norm": 0.7478, "lr": 5e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 276.8}
{"event": "train", "step": 5050, "phase": 3, "objective": "none", "lm_loss": 1.2695, "bpb": 1.8315, "chunk_loss": 0.0, "F": 0.1539, "G": 0.21719, "chunks": 157.59, "bytes_per_chunk": 6.498, "boundary_entropy": 0.29753, "router_noise": 0.0, "grad_norm": 0.7118, "lr": 4.81e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 278.8}
{"event": "train", "step": 5100, "phase": 3, "objective": "none", "lm_loss": 1.31227, "bpb": 1.89321, "chunk_loss": 0.0, "F": 0.15335, "G": 0.21942, "chunks": 157.03, "bytes_per_chunk": 6.521, "boundary_entropy": 0.30466, "router_noise": 0.0, "grad_norm": 0.735, "lr": 4.63e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 280.8}
{"event": "train", "step": 5150, "phase": 3, "objective": "none", "lm_loss": 1.33964, "bpb": 1.93269, "chunk_loss": 0.0, "F": 0.15897, "G": 0.21387, "chunks": 162.78, "bytes_per_chunk": 6.291, "boundary_entropy": 0.28489, "router_noise": 0.0, "grad_norm": 0.7289, "lr": 4.45e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 282.8}
{"event": "probe", "step": 5200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3529.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.202781, "cps": 27.570312, "mean_p": 0.230858, "boundary_entropy": 0.237163, "bpb": 2.033134}, "de": {"chunks": 3892.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189549, "cps": 30.40625, "mean_p": 0.218573, "boundary_entropy": 0.226994, "bpb": 2.220393}, "vi": {"chunks": 3197.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.17342, "cps": 24.976562, "mean_p": 0.207832, "boundary_entropy": 0.307759, "bpb": 1.930255}, "tr": {"chunks": 3562.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196763, "cps": 27.828125, "mean_p": 0.218302, "boundary_entropy": 0.247206, "bpb": 2.20229}, "fi": {"chunks": 3680.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19563, "cps": 28.75, "mean_p": 0.22902, "boundary_entropy": 0.253155, "bpb": 2.363459}, "ru": {"chunks": 3555.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.187599, "cps": 27.773438, "mean_p": 0.163715, "boundary_entropy": 0.261057, "bpb": 1.392233}, "el": {"chunks": 4194.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.204356, "cps": 32.765625, "mean_p": 0.17712, "boundary_entropy": 0.320498, "bpb": 1.474816}, "hi": {"chunks": 3388.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.201367, "cps": 26.46875, "mean_p": 0.207206, "boundary_entropy": 0.421174, "bpb": 1.203828}, "ta": {"chunks": 3258.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.174271, "cps": 25.453125, "mean_p": 0.203605, "boundary_entropy": 0.425588, "bpb": 0.99121}, "th": {"chunks": 3752.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.233261, "cps": 29.3125, "mean_p": 0.220204, "boundary_entropy": 0.448971, "bpb": 1.127924}, "zh": {"chunks": 3438.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.58212, "cps": 26.859375, "mean_p": 0.296334, "boundary_entropy": 0.448219, "bpb": 3.121371}, "my": {"chunks": 2568.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.135529, "cps": 20.0625, "mean_p": 0.248119, "boundary_entropy": 0.508608, "bpb": 1.109081}}, "bpb": {"en": 2.03313, "de": 2.22039, "vi": 1.93025, "tr": 2.20229, "fi": 2.36346, "ru": 1.39223, "el": 1.47482, "hi": 1.20383, "ta": 0.99121, "th": 1.12792, "zh": 3.12137, "my": 1.10908}, "gini_cpc": 0.17977, "gini_cps": 0.058428, "bpic": 9.134815, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9347444307145986, "vi": 0.855208396498984, "tr": 0.9703218823392549, "fi": 0.9647358048629144, "ru": 0.9251301877919179, "el": 1.0077667907428454, "hi": 0.9930263909689909, "ta": 0.8594053883267748, "th": 1.1503082281162, "zh": 2.8706807136303842, "my": 0.6683502351187331}, "t": 285.7}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 2.04163, "de": 2.22606, "vi": 1.93161, "tr": 2.20821, "fi": 2.37281, "ru": 1.37031, "el": 1.46545, "hi": 1.20081, "ta": 0.99817, "th": 1.14519, "zh": 3.11713, "my": 1.11698}, "bytes_per_chunk": {"en": 5.017, "de": 5.255, "vi": 7.319, "tr": 5.299, "fi": 4.944, "ru": 9.381, "el": 8.7, "hi": 12.06, "ta": 15.546, "th": 12.402, "zh": 4.975, "my": 19.798}, "bpb_mean": 1.7662, "t": 287.0}
{"event": "train", "step": 5200, "phase": 3, "objective": "none", "lm_loss": 1.29144, "bpb": 1.86316, "chunk_loss": 0.0, "F": 0.15512, "G": 0.21821, "chunks": 158.84, "bytes_per_chunk": 6.447, "boundary_entropy": 0.30293, "router_noise": 0.0, "grad_norm": 0.7864, "lr": 4.29e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 287.1}
{"event": "train", "step": 5250, "phase": 3, "objective": "none", "lm_loss": 1.38366, "bpb": 1.9962, "chunk_loss": 0.0, "F": 0.16229, "G": 0.22768, "chunks": 166.19, "bytes_per_chunk": 6.162, "boundary_entropy": 0.3116, "router_noise": 0.0, "grad_norm": 0.7559, "lr": 4.14e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 289.0}
{"event": "train", "step": 5300, "phase": 3, "objective": "none", "lm_loss": 1.37776, "bpb": 1.98768, "chunk_loss": 0.0, "F": 0.16666, "G": 0.22949, "chunks": 170.66, "bytes_per_chunk": 6.0, "boundary_entropy": 0.30989, "router_noise": 0.0, "grad_norm": 0.7484, "lr": 3.99e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 291.0}
{"event": "train", "step": 5350, "phase": 3, "objective": "none", "lm_loss": 1.42474, "bpb": 2.05547, "chunk_loss": 0.0, "F": 0.16757, "G": 0.22945, "chunks": 171.59, "bytes_per_chunk": 5.968, "boundary_entropy": 0.30591, "router_noise": 0.0, "grad_norm": 0.7716, "lr": 3.86e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 293.0}
{"event": "probe", "step": 5400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3544.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203643, "cps": 27.6875, "mean_p": 0.232378, "boundary_entropy": 0.239988, "bpb": 2.025682}, "de": {"chunks": 3925.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191156, "cps": 30.664062, "mean_p": 0.221348, "boundary_entropy": 0.231772, "bpb": 2.213438}, "vi": {"chunks": 3205.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.173854, "cps": 25.039062, "mean_p": 0.209952, "boundary_entropy": 0.310638, "bpb": 1.920612}, "tr": {"chunks": 3553.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196266, "cps": 27.757812, "mean_p": 0.21846, "boundary_entropy": 0.24765, "bpb": 2.195604}, "fi": {"chunks": 3704.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.196906, "cps": 28.9375, "mean_p": 0.231004, "boundary_entropy": 0.256596, "bpb": 2.352469}, "ru": {"chunks": 3523.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.18591, "cps": 27.523438, "mean_p": 0.162845, "boundary_entropy": 0.26096, "bpb": 1.387497}, "el": {"chunks": 4324.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.21069, "cps": 33.78125, "mean_p": 0.179215, "boundary_entropy": 0.323643, "bpb": 1.46289}, "hi": {"chunks": 3273.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.194532, "cps": 25.570312, "mean_p": 0.20781, "boundary_entropy": 0.423688, "bpb": 1.20234}, "ta": {"chunks": 3356.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.179513, "cps": 26.21875, "mean_p": 0.197962, "boundary_entropy": 0.411149, "bpb": 0.983512}, "th": {"chunks": 3729.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.231831, "cps": 29.132812, "mean_p": 0.219726, "boundary_entropy": 0.449791, "bpb": 1.122832}, "zh": {"chunks": 3294.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.557738, "cps": 25.734375, "mean_p": 0.292125, "boundary_entropy": 0.447777, "bpb": 3.120097}, "my": {"chunks": 3038.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.160334, "cps": 23.734375, "mean_p": 0.259264, "boundary_entropy": 0.515708, "bpb": 1.102828}}, "bpb": {"en": 2.02568, "de": 2.21344, "vi": 1.92061, "tr": 2.1956, "fi": 2.35247, "ru": 1.3875, "el": 1.46289, "hi": 1.20234, "ta": 0.98351, "th": 1.12283, "zh": 3.1201, "my": 1.10283}, "gini_cpc": 0.162143, "gini_cps": 0.051961, "bpic": 9.036945, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9386802080095918, "vi": 0.8537196917238961, "tr": 0.9637736826728519, "fi": 0.9669176804840067, "ru": 0.9129223422693676, "el": 1.0346066060992627, "hi": 0.9552594427431501, "ta": 0.881509282005244, "th": 1.138417924387302, "zh": 2.738801513401454, "my": 0.7873263752382069}, "t": 295.9}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 2.03448, "de": 2.21795, "vi": 1.92165, "tr": 2.20186, "fi": 2.36199, "ru": 1.36641, "el": 1.45354, "hi": 1.19828, "ta": 0.99028, "th": 1.13987, "zh": 3.11492, "my": 1.11128}, "bytes_per_chunk": {"en": 4.996, "de": 5.21, "vi": 7.269, "tr": 5.315, "fi": 4.925, "ru": 9.503, "el": 8.373, "hi": 12.566, "ta": 15.147, "th": 12.415, "zh": 5.229, "my": 16.407}, "bpb_mean": 1.75938, "t": 297.2}
{"event": "train", "step": 5400, "phase": 3, "objective": "none", "lm_loss": 1.14032, "bpb": 1.64514, "chunk_loss": 0.0, "F": 0.13583, "G": 0.21368, "chunks": 139.09, "bytes_per_chunk": 7.362, "boundary_entropy": 0.32322, "router_noise": 0.0, "grad_norm": 0.7181, "lr": 3.73e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 297.3}
{"event": "train", "step": 5450, "phase": 3, "objective": "none", "lm_loss": 1.20052, "bpb": 1.73198, "chunk_loss": 0.0, "F": 0.14374, "G": 0.21326, "chunks": 147.19, "bytes_per_chunk": 6.957, "boundary_entropy": 0.30828, "router_noise": 0.0, "grad_norm": 0.7665, "lr": 3.62e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 299.3}
{"event": "train", "step": 5500, "phase": 3, "objective": "none", "lm_loss": 1.39893, "bpb": 2.01823, "chunk_loss": 0.0, "F": 0.17142, "G": 0.23388, "chunks": 175.53, "bytes_per_chunk": 5.834, "boundary_entropy": 0.30132, "router_noise": 0.0, "grad_norm": 0.7732, "lr": 3.51e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 301.3}
{"event": "train", "step": 5550, "phase": 3, "objective": "none", "lm_loss": 1.41018, "bpb": 2.03446, "chunk_loss": 0.0, "F": 0.16742, "G": 0.22802, "chunks": 171.44, "bytes_per_chunk": 5.973, "boundary_entropy": 0.30519, "router_noise": 0.0, "grad_norm": 0.7809, "lr": 3.41e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 303.3}
{"event": "probe", "step": 5600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3606.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207206, "cps": 28.171875, "mean_p": 0.235971, "boundary_entropy": 0.24107, "bpb": 2.018792}, "de": {"chunks": 3935.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191643, "cps": 30.742188, "mean_p": 0.222061, "boundary_entropy": 0.231157, "bpb": 2.206526}, "vi": {"chunks": 3240.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.175753, "cps": 25.3125, "mean_p": 0.209886, "boundary_entropy": 0.308925, "bpb": 1.921018}, "tr": {"chunks": 3594.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198531, "cps": 28.078125, "mean_p": 0.220803, "boundary_entropy": 0.248011, "bpb": 2.184811}, "fi": {"chunks": 3751.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.199405, "cps": 29.304688, "mean_p": 0.234179, "boundary_entropy": 0.257768, "bpb": 2.34545}, "ru": {"chunks": 3562.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.187968, "cps": 27.828125, "mean_p": 0.163288, "boundary_entropy": 0.260807, "bpb": 1.384863}, "el": {"chunks": 4335.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.211226, "cps": 33.867188, "mean_p": 0.181735, "boundary_entropy": 0.327282, "bpb": 1.453129}, "hi": {"chunks": 3242.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.192689, "cps": 25.328125, "mean_p": 0.206163, "boundary_entropy": 0.422531, "bpb": 1.195262}, "ta": {"chunks": 2815.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.150575, "cps": 21.992188, "mean_p": 0.184525, "boundary_entropy": 0.397748, "bpb": 0.982803}, "th": {"chunks": 3351.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.208331, "cps": 26.179688, "mean_p": 0.212398, "boundary_entropy": 0.44096, "bpb": 1.122021}, "zh": {"chunks": 3312.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.560786, "cps": 25.875, "mean_p": 0.293084, "boundary_entropy": 0.44816, "bpb": 3.108787}, "my": {"chunks": 2855.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.150676, "cps": 22.304688, "mean_p": 0.254275, "boundary_entropy": 0.511679, "bpb": 1.101538}}, "bpb": {"en": 2.01879, "de": 2.20653, "vi": 1.92102, "tr": 2.18481, "fi": 2.34545, "ru": 1.38486, "el": 1.45313, "hi": 1.19526, "ta": 0.9828, "th": 1.12202, "zh": 3.10879, "my": 1.10154}, "gini_cpc": 0.171128, "gini_cps": 0.065897, "bpic": 9.225948, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9248913704354655, "vi": 0.8482039020942854, "tr": 0.958133266445482, "fi": 0.9623511697186418, "ru": 0.9071583420771888, "el": 1.0194047537104387, "hi": 0.9299430122653427, "ta": 0.7266935868230671, "th": 1.0054298167441373, "zh": 2.706420555423769, "my": 0.727178674835446}, "t": 306.2}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 2.02869, "de": 2.21159, "vi": 1.92137, "tr": 2.19068, "fi": 2.35451, "ru": 1.36305, "el": 1.44379, "hi": 1.19249, "ta": 0.98889, "th": 1.13884, "zh": 3.104, "my": 1.10939}, "bytes_per_chunk": {"en": 4.948, "de": 5.182, "vi": 7.285, "tr": 5.252, "fi": 4.857, "ru": 9.372, "el": 8.332, "hi": 12.66, "ta": 18.099, "th": 13.718, "zh": 5.172, "my": 17.576}, "bpb_mean": 1.75394, "t": 307.5}
{"event": "train", "step": 5600, "phase": 3, "objective": "none", "lm_loss": 1.46353, "bpb": 2.11142, "chunk_loss": 0.0, "F": 0.17285, "G": 0.22923, "chunks": 177.0, "bytes_per_chunk": 5.785, "boundary_entropy": 0.29861, "router_noise": 0.0, "grad_norm": 0.7692, "lr": 3.33e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 307.5}
{"event": "train", "step": 5650, "phase": 3, "objective": "none", "lm_loss": 1.27468, "bpb": 1.83897, "chunk_loss": 0.0, "F": 0.155, "G": 0.22126, "chunks": 158.72, "bytes_per_chunk": 6.452, "boundary_entropy": 0.31037, "router_noise": 0.0, "grad_norm": 0.7199, "lr": 3.25e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 309.5}
{"event": "train", "step": 5700, "phase": 3, "objective": "none", "lm_loss": 1.32394, "bpb": 1.91004, "chunk_loss": 0.0, "F": 0.16653, "G": 0.22437, "chunks": 170.53, "bytes_per_chunk": 6.005, "boundary_entropy": 0.29065, "router_noise": 0.0, "grad_norm": 0.7457, "lr": 3.18e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 311.5}
{"event": "train", "step": 5750, "phase": 3, "objective": "none", "lm_loss": 1.20927, "bpb": 1.7446, "chunk_loss": 0.0, "F": 0.15244, "G": 0.2154, "chunks": 156.09, "bytes_per_chunk": 6.56, "boundary_entropy": 0.29624, "router_noise": 0.0, "grad_norm": 0.7291, "lr": 3.13e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 313.5}
{"event": "probe", "step": 5800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3573.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.205309, "cps": 27.914062, "mean_p": 0.236962, "boundary_entropy": 0.241986, "bpb": 2.01612}, "de": {"chunks": 3956.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.192665, "cps": 30.90625, "mean_p": 0.223836, "boundary_entropy": 0.232428, "bpb": 2.203259}, "vi": {"chunks": 3221.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.174722, "cps": 25.164062, "mean_p": 0.209514, "boundary_entropy": 0.309933, "bpb": 1.916569}, "tr": {"chunks": 3593.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198475, "cps": 28.070312, "mean_p": 0.221169, "boundary_entropy": 0.247764, "bpb": 2.177811}, "fi": {"chunks": 3671.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195152, "cps": 28.679688, "mean_p": 0.229669, "boundary_entropy": 0.254548, "bpb": 2.337794}, "ru": {"chunks": 3634.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.191768, "cps": 28.390625, "mean_p": 0.166403, "boundary_entropy": 0.262075, "bpb": 1.379692}, "el": {"chunks": 4218.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.205526, "cps": 32.953125, "mean_p": 0.178145, "boundary_entropy": 0.32275, "bpb": 1.450067}, "hi": {"chunks": 3528.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.209688, "cps": 27.5625, "mean_p": 0.209729, "boundary_entropy": 0.425839, "bpb": 1.189773}, "ta": {"chunks": 3167.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.169404, "cps": 24.742188, "mean_p": 0.193667, "boundary_entropy": 0.405384, "bpb": 0.978917}, "th": {"chunks": 3533.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.219646, "cps": 27.601562, "mean_p": 0.216048, "boundary_entropy": 0.444468, "bpb": 1.119016}, "zh": {"chunks": 3485.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.590078, "cps": 27.226562, "mean_p": 0.297976, "boundary_entropy": 0.448975, "bpb": 3.104372}, "my": {"chunks": 3034.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.160122, "cps": 23.703125, "mean_p": 0.255709, "boundary_entropy": 0.515693, "bpb": 1.097842}}, "bpb": {"en": 2.01612, "de": 2.20326, "vi": 1.91657, "tr": 2.17781, "fi": 2.33779, "ru": 1.37969, "el": 1.45007, "hi": 1.18977, "ta": 0.97892, "th": 1.11902, "zh": 3.10437, "my": 1.09784}, "gini_cpc": 0.170089, "gini_cps": 0.047433, "bpic": 9.006195, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9384150835318527, "vi": 0.8510178841082097, "tr": 0.9667134694728685, "fi": 0.9505251339934059, "ru": 0.934042876543292, "el": 1.001052453160925, "hi": 1.0213265164813365, "ta": 0.8251135095636818, "th": 1.0698273002340892, "zh": 2.8740905305424502, "my": 0.7799078728348761}, "t": 316.5}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 2.02514, "de": 2.20919, "vi": 1.91677, "tr": 2.18421, "fi": 2.34771, "ru": 1.3581, "el": 1.44199, "hi": 1.18679, "ta": 0.98568, "th": 1.13551, "zh": 3.10048, "my": 1.1064}, "bytes_per_chunk": {"en": 4.894, "de": 5.141, "vi": 7.207, "tr": 5.235, "fi": 4.942, "ru": 9.171, "el": 8.605, "hi": 11.59, "ta": 15.911, "th": 13.099, "zh": 4.906, "my": 16.819}, "bpb_mean": 1.74983, "t": 317.7}
{"event": "train", "step": 5800, "phase": 3, "objective": "none", "lm_loss": 1.26286, "bpb": 1.82192, "chunk_loss": 0.0, "F": 0.15906, "G": 0.22433, "chunks": 162.88, "bytes_per_chunk": 6.287, "boundary_entropy": 0.3067, "router_noise": 0.0, "grad_norm": 0.7307, "lr": 3.08e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 317.8}
{"event": "train", "step": 5850, "phase": 3, "objective": "none", "lm_loss": 1.37226, "bpb": 1.97975, "chunk_loss": 0.0, "F": 0.16418, "G": 0.22991, "chunks": 168.12, "bytes_per_chunk": 6.091, "boundary_entropy": 0.31396, "router_noise": 0.0, "grad_norm": 0.7256, "lr": 3.05e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 319.8}
{"event": "train", "step": 5900, "phase": 3, "objective": "none", "lm_loss": 1.32588, "bpb": 1.91284, "chunk_loss": 0.0, "F": 0.16217, "G": 0.21799, "chunks": 166.06, "bytes_per_chunk": 6.166, "boundary_entropy": 0.28748, "router_noise": 0.0, "grad_norm": 0.7654, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 321.8}
{"event": "train", "step": 5950, "phase": 3, "objective": "none", "lm_loss": 1.2877, "bpb": 1.85776, "chunk_loss": 0.0, "F": 0.15637, "G": 0.2259, "chunks": 160.12, "bytes_per_chunk": 6.395, "boundary_entropy": 0.31686, "router_noise": 0.0, "grad_norm": 0.7363, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.041, "alpha": 0.0, "t": 323.8}
{"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3578.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.205597, "cps": 27.953125, "mean_p": 0.237197, "boundary_entropy": 0.240191, "bpb": 2.011496}, "de": {"chunks": 3958.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.192763, "cps": 30.921875, "mean_p": 0.22278, "boundary_entropy": 0.230224, "bpb": 2.198401}, "vi": {"chunks": 3221.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.174722, "cps": 25.164062, "mean_p": 0.21067, "boundary_entropy": 0.310628, "bpb": 1.914011}, "tr": {"chunks": 3678.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203171, "cps": 28.734375, "mean_p": 0.223989, "boundary_entropy": 0.249267, "bpb": 2.177013}, "fi": {"chunks": 3765.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.200149, "cps": 29.414062, "mean_p": 0.233132, "boundary_entropy": 0.254217, "bpb": 2.335639}, "ru": {"chunks": 3583.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.189077, "cps": 27.992188, "mean_p": 0.164845, "boundary_entropy": 0.262281, "bpb": 1.376911}, "el": {"chunks": 4160.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.202699, "cps": 32.5, "mean_p": 0.177665, "boundary_entropy": 0.323576, "bpb": 1.445661}, "hi": {"chunks": 3358.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.199584, "cps": 26.234375, "mean_p": 0.208767, "boundary_entropy": 0.426909, "bpb": 1.192503}, "ta": {"chunks": 2971.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.158919, "cps": 23.210938, "mean_p": 0.187377, "boundary_entropy": 0.399676, "bpb": 0.978792}, "th": {"chunks": 3789.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.235561, "cps": 29.601562, "mean_p": 0.220679, "boundary_entropy": 0.448367, "bpb": 1.113737}, "zh": {"chunks": 3506.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.593634, "cps": 27.390625, "mean_p": 0.298378, "boundary_entropy": 0.449662, "bpb": 3.097201}, "my": {"chunks": 2815.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.148564, "cps": 21.992188, "mean_p": 0.253849, "boundary_entropy": 0.513698, "bpb": 1.097345}}, "bpb": {"en": 2.0115, "de": 2.1984, "vi": 1.91401, "tr": 2.17701, "fi": 2.33564, "ru": 1.37691, "el": 1.44566, "hi": 1.1925, "ta": 0.97879, "th": 1.11374, "zh": 3.0972, "my": 1.09734}, "gini_cpc": 0.181877, "gini_cps": 0.0594, "bpic": 9.055283, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9375774785858493, "vi": 0.8498286472662473, "tr": 0.988200245492943, "fi": 0.9735020739236011, "ru": 0.9196474644964611, "el": 0.9859077248454862, "hi": 0.9707544786241161, "ta": 0.7729670233729019, "th": 1.145743293949329, "zh": 2.8873687585854557, "my": 0.7226014140314942}, "t": 326.7}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.02227, "de": 2.20454, "vi": 1.91382, "tr": 2.18179, "fi": 2.34513, "ru": 1.3553, "el": 1.4374, "hi": 1.1898, "ta": 0.9857, "th": 1.13089, "zh": 3.0932, "my": 1.10586}, "bytes_per_chunk": {"en": 4.882, "de": 5.161, "vi": 7.19, "tr": 5.129, "fi": 4.845, "ru": 9.311, "el": 8.702, "hi": 12.519, "ta": 17.027, "th": 12.402, "zh": 4.93, "my": 18.125}, "bpb_mean": 1.74714, "t": 328.0}
{"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3578.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.205597, "cps": 27.953125, "mean_p": 0.237197, "boundary_entropy": 0.240191, "bpb": 2.011496}, "de": {"chunks": 3958.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.192763, "cps": 30.921875, "mean_p": 0.22278, "boundary_entropy": 0.230224, "bpb": 2.198401}, "vi": {"chunks": 3221.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.174722, "cps": 25.164062, "mean_p": 0.21067, "boundary_entropy": 0.310628, "bpb": 1.914011}, "tr": {"chunks": 3678.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203171, "cps": 28.734375, "mean_p": 0.223989, "boundary_entropy": 0.249267, "bpb": 2.177013}, "fi": {"chunks": 3765.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.200149, "cps": 29.414062, "mean_p": 0.233132, "boundary_entropy": 0.254217, "bpb": 2.335639}, "ru": {"chunks": 3583.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.189077, "cps": 27.992188, "mean_p": 0.164845, "boundary_entropy": 0.262281, "bpb": 1.376911}, "el": {"chunks": 4160.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.202699, "cps": 32.5, "mean_p": 0.177665, "boundary_entropy": 0.323576, "bpb": 1.445661}, "hi": {"chunks": 3358.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.199584, "cps": 26.234375, "mean_p": 0.208767, "boundary_entropy": 0.426909, "bpb": 1.192503}, "ta": {"chunks": 2971.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.158919, "cps": 23.210938, "mean_p": 0.187377, "boundary_entropy": 0.399676, "bpb": 0.978792}, "th": {"chunks": 3789.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.235561, "cps": 29.601562, "mean_p": 0.220679, "boundary_entropy": 0.448367, "bpb": 1.113737}, "zh": {"chunks": 3506.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.593634, "cps": 27.390625, "mean_p": 0.298378, "boundary_entropy": 0.449662, "bpb": 3.097201}, "my": {"chunks": 2815.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.148564, "cps": 21.992188, "mean_p": 0.253849, "boundary_entropy": 0.513698, "bpb": 1.097345}}, "bpb": {"en": 2.0115, "de": 2.1984, "vi": 1.91401, "tr": 2.17701, "fi": 2.33564, "ru": 1.37691, "el": 1.44566, "hi": 1.1925, "ta": 0.97879, "th": 1.11374, "zh": 3.0972, "my": 1.09734}, "gini_cpc": 0.181877, "gini_cps": 0.0594, "bpic": 9.055283, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9375774785858493, "vi": 0.8498286472662473, "tr": 0.988200245492943, "fi": 0.9735020739236011, "ru": 0.9196474644964611, "el": 0.9859077248454862, "hi": 0.9707544786241161, "ta": 0.7729670233729019, "th": 1.145743293949329, "zh": 2.8873687585854557, "my": 0.7226014140314942}, "t": 329.0}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_B_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_B_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_B_s0/th0.60"], "t": 330.2}
{"event": "final", "run": "pilot_B_s0", "condition": "B", "step": 6000, "params": 22516992, "per_lang": {"en": {"cpc": 0.20559673619490892, "cps": 27.953125, "bpb": 2.030971458180637, "bytes_per_chunk": 4.841428729730728}, "de": {"cpc": 0.19276286952710273, "cps": 30.921875, "bpb": 2.1954336946821793, "bytes_per_chunk": 5.174984207201516}, "vi": {"cpc": 0.17472199620287496, "cps": 25.1640625, "bpb": 1.9176065463714007, "bytes_per_chunk": 7.180060257463708}, "tr": {"cpc": 0.20317074518035685, "cps": 28.734375, "bpb": 2.1736990002184275, "bytes_per_chunk": 5.15579856228304}, "fi": {"cpc": 0.20014884907766733, "cps": 29.4140625, "bpb": 2.363505277332801, "bytes_per_chunk": 4.845366160215889}, "ru": {"cpc": 0.18907651715039578, "cps": 27.9921875, "bpb": 1.3564596129569486, "bytes_per_chunk": 9.31041341099588}, "el": {"cpc": 0.20269941041758027, "cps": 32.5, "bpb": 1.4347470400364475, "bytes_per_chunk": 8.671650678134304}, "hi": {"cpc": 0.19958395245170876, "cps": 26.234375, "bpb": 1.1980217889792264, "bytes_per_chunk": 12.50180031952691}, "ta": {"cpc": 0.15891949719176252, "cps": 23.2109375, "bpb": 0.9865577903862777, "bytes_per_chunk": 17.10006523157208}, "th": {"cpc": 0.23556108175318619, "cps": 29.6015625, "bpb": 1.1292825162081603, "bytes_per_chunk": 12.512242852369816}, "zh": {"cpc": 0.5936335929563156, "cps": 27.390625, "bpb": 3.1025653662320534, "bytes_per_chunk": 4.933360934941755}, "my": {"cpc": 0.14856449229470128, "cps": 21.9921875, "bpb": 1.1157591871544317, "bytes_per_chunk": 17.892567060268924}}, "gini_cpc": 0.18187695586727237, "gini_cps": 0.05940021707328591, "bpic": 9.055282903119249, "seed": 0, "mean_boundary_entropy": 0.29383590817451477, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 2400, "phase": 2, "objective": "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}], "t": 332.6}