repa-layer4: train.jsonl

repa-layer4/train.jsonl

@@ -71,3 +71,16 @@
 {"step": 6300, "epoch": 2, "lr": 0.0007902839913576722, "loss": 0.5734886527061462, "flow": 0.5734886527061462, "duration": 0.0, "ctc": 0.009695056825876236, "contrastive": 0.011732643470168114, "contrastive_hinge_mean": 0.046930573880672455, "repa": 0.15733769536018372, "elapsed_seconds": 28.812291425885633, "valid_target_frames": 12500, "gradient_norm_before_clip": 0.14719010889530182, "peak_cuda_gb": 8.5024733543396, "peak_reserved_gb": 9.412109375, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[1.9774398803710938, 2.0], [5.4554948806762695, 9.0], [4.514232635498047, 10.0], [4.216144561767578, 8.0], [2.2374155521392822, 3.0], [0.0, 0.0], [0.0, 0.0], [18.400728225708008, 32.0], [0.0, 0.0]], "diagnostics_rank0": {}}
 {"step": 6400, "epoch": 2, "lr": 0.0007898288868811812, "loss": 0.5722604393959045, "flow": 0.5722604393959045, "duration": 0.0, "ctc": 0.0024904359597712755, "contrastive": 0.011091254651546478, "contrastive_hinge_mean": 0.04436501860618591, "repa": 0.15615937113761902, "elapsed_seconds": 28.837283070897684, "valid_target_frames": 10960, "gradient_norm_before_clip": 0.11994996666908264, "peak_cuda_gb": 8.5024733543396, "peak_reserved_gb": 9.412109375, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[3.666083574295044, 4.0], [9.082342147827148, 15.0], [8.120018005371094, 18.0], [7.601717948913574, 14.0], [3.884737730026245, 5.0], [0.0, 0.0], [25.41442108154297, 44.0], [6.940475940704346, 12.0], [0.0, 0.0]], "diagnostics_rank0": {}}
 {"step": 6500, "epoch": 2, "lr": 0.0007893635188366755, "loss": 0.5642619132995605, "flow": 0.5642619132995605, "duration": 0.0, "ctc": 0.002694732742384076, "contrastive": 0.011607221327722073, "contrastive_hinge_mean": 0.04642888531088829, "repa": 0.16891446709632874, "elapsed_seconds": 28.26414952496998, "valid_target_frames": 17380, "gradient_norm_before_clip": 0.129752516746521, "peak_cuda_gb": 8.5024733543396, "peak_reserved_gb": 9.416015625, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[6.623420715332031, 7.0], [13.358503341674805, 23.0], [12.372335433959961, 28.0], [12.074045181274414, 23.0], [5.300894737243652, 7.0], [1.9184125661849976, 4.0], [25.412708282470703, 44.0], [22.3980770111084, 40.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 6600, "epoch": 2, "lr": 0.0007888879008775061, "loss": 0.5958790183067322, "flow": 0.5958790183067322, "duration": 0.0, "ctc": 0.009246142581105232, "contrastive": 0.011168875731527805, "contrastive_hinge_mean": 0.04467550292611122, "repa": 0.17091815173625946, "elapsed_seconds": 23.454599494114518, "valid_target_frames": 18172, "gradient_norm_before_clip": 0.1328054964542389, "peak_cuda_gb": 8.536772727966309, "peak_reserved_gb": 9.45703125, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[11.46543025970459, 12.0], [23.855207443237305, 39.0], [21.198469161987305, 46.0], [22.04473876953125, 39.0], [9.466838836669922, 12.0], [23.345165252685547, 40.0], [64.68549346923828, 108.0], [0.0, 0.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 6700, "epoch": 2, "lr": 0.0007884020469577446, "loss": 0.5570169687271118, "flow": 0.5570169687271118, "duration": 0.0, "ctc": 0.020934266969561577, "contrastive": 0.009451597929000854, "contrastive_hinge_mean": 0.03780639171600342, "repa": 0.14469926059246063, "elapsed_seconds": 21.96152242203243, "valid_target_frames": 12956, "gradient_norm_before_clip": 0.14568179845809937, "peak_cuda_gb": 8.536772727966309, "peak_reserved_gb": 9.45703125, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[5.40907621383667, 6.0], [10.959345817565918, 19.0], [9.400321960449219, 22.0], [10.127894401550293, 19.0], [4.756018161773682, 6.0], [1.9003791809082031, 4.0], [32.62386703491211, 56.0], [6.12840461730957, 12.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 6800, "epoch": 2, "lr": 0.0007879059713317731, "loss": 0.5688737034797668, "flow": 0.5688737034797668, "duration": 0.0, "ctc": 0.014210036024451256, "contrastive": 0.007812291383743286, "contrastive_hinge_mean": 0.031249165534973145, "repa": 0.17611797153949738, "elapsed_seconds": 28.28727808292024, "valid_target_frames": 18556, "gradient_norm_before_clip": 0.11288778483867645, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[7.292585849761963, 8.0], [15.482266426086426, 27.0], [13.229453086853027, 31.0], [14.293219566345215, 26.0], [6.387465476989746, 8.0], [2.170980215072632, 4.0], [54.51400375366211, 96.0], [0.0, 0.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 6900, "epoch": 2, "lr": 0.0007873996885538673, "loss": 0.5691432356834412, "flow": 0.5691432356834412, "duration": 0.0, "ctc": 0.0029615452513098717, "contrastive": 0.011799178086221218, "contrastive_hinge_mean": 0.04719671234488487, "repa": 0.16370545327663422, "elapsed_seconds": 24.70719800097868, "valid_target_frames": 12704, "gradient_norm_before_clip": 0.125839963555336, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[2.684370517730713, 3.0], [5.24609375, 9.0], [5.400796890258789, 12.0], [4.913494110107422, 9.0], [2.3434300422668457, 3.0], [0.0, 0.0], [0.0, 0.0], [20.588180541992188, 36.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 7000, "epoch": 2, "lr": 0.000786883213477768, "loss": 0.578972339630127, "flow": 0.578972339630127, "duration": 0.0, "ctc": 0.004431066103279591, "contrastive": 0.007905382663011551, "contrastive_hinge_mean": 0.031621530652046204, "repa": 0.1611223667860031, "elapsed_seconds": 24.864908284973353, "valid_target_frames": 11052, "gradient_norm_before_clip": 0.1330261081457138, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[3.7960898876190186, 4.0], [7.310476303100586, 13.0], [6.930738925933838, 15.0], [6.700697898864746, 12.0], [3.1649179458618164, 4.0], [0.0, 0.0], [16.9699764251709, 28.0], [10.932943344116211, 20.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 7000, "validation_loss": 0.6469201445579529, "validation_flow": 0.6469201445579529, "validation_text_gain": 0.01998811960220337}
+{"step": 7100, "epoch": 2, "lr": 0.0007863565612562464, "loss": 0.5826029181480408, "flow": 0.5826029181480408, "duration": 0.0, "ctc": 0.0035572350025177, "contrastive": 0.011770983226597309, "contrastive_hinge_mean": 0.047083932906389236, "repa": 0.15920411050319672, "elapsed_seconds": 26.68072430882603, "valid_target_frames": 16700, "gradient_norm_before_clip": 0.11568813025951385, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[6.363649368286133, 7.0], [15.482521057128906, 25.0], [14.035238265991211, 31.0], [13.517003059387207, 25.0], [6.297796249389648, 8.0], [5.141489028930664, 8.0], [50.554718017578125, 88.0], [0.0, 0.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 7200, "epoch": 3, "lr": 0.0007858197473406596, "loss": 0.5835497975349426, "flow": 0.5835497975349426, "duration": 0.0, "ctc": 0.007349094841629267, "contrastive": 0.012544531375169754, "contrastive_hinge_mean": 0.050178125500679016, "repa": 0.16842250525951385, "elapsed_seconds": 22.049995217006654, "valid_target_frames": 18792, "gradient_norm_before_clip": 0.12435689568519592, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[10.433008193969727, 11.0], [21.36808967590332, 36.0], [19.25169563293457, 42.0], [19.625635147094727, 36.0], [8.878360748291016, 11.0], [12.373425483703613, 20.0], [67.18338012695312, 116.0], [0.0, 0.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 7300, "epoch": 3, "lr": 0.0007852727874804958, "loss": 0.5743223428726196, "flow": 0.5743223428726196, "duration": 0.0, "ctc": 0.0027729037683457136, "contrastive": 0.009613689966499805, "contrastive_hinge_mean": 0.03845475986599922, "repa": 0.144287109375, "elapsed_seconds": 28.846032629022375, "valid_target_frames": 13104, "gradient_norm_before_clip": 0.161556214094162, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[2.848857879638672, 3.0], [7.36423397064209, 12.0], [5.758152484893799, 13.0], [6.926050186157227, 13.0], [2.422144651412964, 3.0], [0.0, 0.0], [8.92607593536377, 16.0], [16.39336395263672, 28.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 7400, "epoch": 3, "lr": 0.000784715697722914, "loss": 0.5833836197853088, "flow": 0.5833836197853088, "duration": 0.0, "ctc": 0.01443590596318245, "contrastive": 0.010671050287783146, "contrastive_hinge_mean": 0.042684201151132584, "repa": 0.16774490475654602, "elapsed_seconds": 28.86312492401339, "valid_target_frames": 15760, "gradient_norm_before_clip": 0.12296238541603088, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[9.676209449768066, 11.0], [22.256458282470703, 37.0], [19.51956558227539, 44.0], [20.765277862548828, 37.0], [8.519662857055664, 11.0], [32.86639404296875, 60.0], [47.87078094482422, 80.0], [0.0, 0.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 7500, "epoch": 3, "lr": 0.0007841484944122727, "loss": 0.5693833231925964, "flow": 0.5693833231925964, "duration": 0.0, "ctc": 0.0018570576794445515, "contrastive": 0.011110804975032806, "contrastive_hinge_mean": 0.044443219900131226, "repa": 0.17288100719451904, "elapsed_seconds": 24.338384313043207, "valid_target_frames": 17508, "gradient_norm_before_clip": 0.11617454141378403, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[9.43210506439209, 11.0], [18.814918518066406, 32.0], [18.30573272705078, 41.0], [17.665325164794922, 33.0], [8.676819801330566, 11.0], [16.554691314697266, 28.0], [56.340213775634766, 100.0], [0.0, 0.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 7600, "epoch": 3, "lr": 0.0007835711941896495, "loss": 0.5902774333953857, "flow": 0.5902774333953857, "duration": 0.0, "ctc": 0.015066113322973251, "contrastive": 0.012733094394207, "contrastive_hinge_mean": 0.050932377576828, "repa": 0.16430732607841492, "elapsed_seconds": 21.263784650946036, "valid_target_frames": 17524, "gradient_norm_before_clip": 0.10869599878787994, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[14.100706100463867, 15.0], [29.058809280395508, 48.0], [27.198217391967773, 58.0], [26.37110710144043, 48.0], [11.984107971191406, 15.0], [59.264461517333984, 100.0], [49.448482513427734, 84.0], [0.0, 0.0], [0.0, 0.0]], "diagnostics_rank0": {}}
+{"step": 7700, "epoch": 3, "lr": 0.0007829838139923539, "loss": 0.5924803614616394, "flow": 0.5924803614616394, "duration": 0.0, "ctc": 0.0011695041321218014, "contrastive": 0.012103235349059105, "contrastive_hinge_mean": 0.04841294139623642, "repa": 0.15932825207710266, "elapsed_seconds": 24.58513294090517, "valid_target_frames": 15640, "gradient_norm_before_clip": 0.12395243346691132, "peak_cuda_gb": 8.895436763763428, "peak_reserved_gb": 9.810546875, "flow_reduction": "frame", "time_and_length_buckets_sum_count": [[8.30445671081543, 9.0], [16.34391975402832, 28.0], [15.274717330932617, 34.0], [15.907221794128418, 28.0], [7.177741527557373, 9.0], [12.65335464477539, 24.0], [50.35470962524414, 84.0], [0.0, 0.0], [0.0, 0.0]], "diagnostics_rank0": {}}