Upload folder using huggingface_hub

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 checkpoints-v5/checkpoint-12288/eval_state.json filter=lfs diff=lfs merge=lfs -text
+checkpoints-v5.1/checkpoint-12288/eval_state.json filter=lfs diff=lfs merge=lfs -text

checkpoints-v5.1/checkpoint-12288/ema.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8afeba4d1431f7babd620f2474439028d75a2677561a50be67829184c14e91a
+size 54599376

checkpoints-v5.1/checkpoint-12288/eval_state.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ecf79c792deb8b161f482a6c98bb0973c7f46f4e5199124cacfcc43a675fea2
+size 41690129

checkpoints-v5.1/checkpoint-12288/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ac9f855cf379a513b27b1dd6f8f94319531a20ec661d289bb347448c1c910d3
+size 54599408

checkpoints-v5.1/checkpoint-12288/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6c0d7073b75def89079d08161458023a30dba9fa688d565f2112342155cabea
+size 76550347

checkpoints-v5.1/checkpoint-12288/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5f8e5d1fe4d3a12883c398185ffe51a550062eb0ca27cef82c662a9af043b7b
+size 14645

checkpoints-v5.1/checkpoint-12288/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b876371b69448f3749ca492ba9e273a97b4582bd9dcba73d308b5d464395d7b1
+size 1383

checkpoints-v5.1/checkpoint-12288/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b5e38a92cb7e4e0d3a7296066cd2cf1a7426f9f51c35b7d8712960c4e715ec8
+size 1465

checkpoints-v5.1/checkpoint-12288/trainer_state.json

@@ -0,0 +1,490 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5675488430095608,
+  "eval_steps": 1024,
+  "global_step": 12288,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.047295736917463395,
+      "grad_norm": 2.106490135192871,
+      "learning_rate": 0.00029970703124999995,
+      "loss": 7.3224077224731445,
+      "step": 1024
+    },
+    {
+      "epoch": 0.047295736917463395,
+      "eval_bleu": 0.16704781801498278,
+      "eval_ce_clean_loss": 0.8006052018300583,
+      "eval_ce_pred_loss": 5.09323315206728,
+      "eval_flow_cos_loss": 0.25552307487758874,
+      "eval_flow_mse_loss": 0.4737413005741764,
+      "eval_loss": 5.201592297314509,
+      "flow/cos_sim": 0.7444769605381848,
+      "flow/improvement_ratio": 0.9690007240804908,
+      "flow/mag_ratio_mean": 0.5684286470962986,
+      "flow/mag_ratio_std": 0.1943025813187094,
+      "step": 1024
+    },
+    {
+      "epoch": 0.047295736917463395,
+      "eval_bleu": 0.16704781801498278,
+      "eval_ce_clean_loss": 0.8006052018300583,
+      "eval_ce_pred_loss": 5.09323315206728,
+      "eval_flow_cos_loss": 0.25552307487758874,
+      "eval_flow_mse_loss": 0.4737413005741764,
+      "eval_loss": 5.201592297314509,
+      "eval_runtime": 184.3437,
+      "eval_samples_per_second": 151.852,
+      "eval_steps_per_second": 2.376,
+      "flow/cos_sim": 0.7444769605381848,
+      "flow/improvement_ratio": 0.9690007240804908,
+      "flow/mag_ratio_mean": 0.5684286470962986,
+      "flow/mag_ratio_std": 0.1943025813187094,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "grad_norm": 1.5151523351669312,
+      "learning_rate": 0.00029818297300322,
+      "loss": 5.1176676750183105,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "eval_bleu": 0.15959126254389083,
+      "eval_ce_clean_loss": 0.2092574786854117,
+      "eval_ce_pred_loss": 5.118338076491334,
+      "eval_flow_cos_loss": 0.13469866032129554,
+      "eval_flow_mse_loss": 0.3451302863963663,
+      "eval_loss": 5.006234865754707,
+      "flow/cos_sim": 0.865301351841182,
+      "flow/improvement_ratio": 0.993777159822586,
+      "flow/mag_ratio_mean": 0.764228024575264,
+      "flow/mag_ratio_std": 0.1451562752938706,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "eval_bleu": 0.15959126254389083,
+      "eval_ce_clean_loss": 0.2092574786854117,
+      "eval_ce_pred_loss": 5.118338076491334,
+      "eval_flow_cos_loss": 0.13469866032129554,
+      "eval_flow_mse_loss": 0.3451302863963663,
+      "eval_loss": 5.006234865754707,
+      "eval_runtime": 179.8087,
+      "eval_samples_per_second": 155.682,
+      "eval_steps_per_second": 2.436,
+      "flow/cos_sim": 0.865301351841182,
+      "flow/improvement_ratio": 0.993777159822586,
+      "flow/mag_ratio_mean": 0.764228024575264,
+      "flow/mag_ratio_std": 0.1451562752938706,
+      "step": 2048
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "grad_norm": 2.0519113540649414,
+      "learning_rate": 0.0002927689070858589,
+      "loss": 4.94038724899292,
+      "step": 3072
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "eval_bleu": 0.19109057338236854,
+      "eval_ce_clean_loss": 0.09927985731291172,
+      "eval_ce_pred_loss": 4.847544043031458,
+      "eval_flow_cos_loss": 0.11591754169905022,
+      "eval_flow_mse_loss": 0.3725693541574696,
+      "eval_loss": 4.7742662386262795,
+      "flow/cos_sim": 0.8840824780671019,
+      "flow/improvement_ratio": 0.9947053511120957,
+      "flow/mag_ratio_mean": 0.8077854450162687,
+      "flow/mag_ratio_std": 0.12759848946033547,
+      "step": 3072
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "eval_bleu": 0.19109057338236854,
+      "eval_ce_clean_loss": 0.09927985731291172,
+      "eval_ce_pred_loss": 4.847544043031458,
+      "eval_flow_cos_loss": 0.11591754169905022,
+      "eval_flow_mse_loss": 0.3725693541574696,
+      "eval_loss": 4.7742662386262795,
+      "eval_runtime": 179.382,
+      "eval_samples_per_second": 156.052,
+      "eval_steps_per_second": 2.442,
+      "flow/cos_sim": 0.8840824780671019,
+      "flow/improvement_ratio": 0.9947053511120957,
+      "flow/mag_ratio_mean": 0.8077854450162687,
+      "flow/mag_ratio_std": 0.12759848946033547,
+      "step": 3072
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "grad_norm": 2.0934338569641113,
+      "learning_rate": 0.0002838892232677901,
+      "loss": 4.741153240203857,
+      "step": 4096
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "eval_bleu": 0.18968657464122718,
+      "eval_ce_clean_loss": 0.05336422391967278,
+      "eval_ce_pred_loss": 4.809050649268442,
+      "eval_flow_cos_loss": 0.07764796834422029,
+      "eval_flow_mse_loss": 0.3626167481620562,
+      "eval_loss": 4.7155106383371574,
+      "flow/cos_sim": 0.9223520577498222,
+      "flow/improvement_ratio": 0.9944308828817655,
+      "flow/mag_ratio_mean": 0.8769120071848778,
+      "flow/mag_ratio_std": 0.09842952985537651,
+      "step": 4096
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "eval_bleu": 0.18968657464122718,
+      "eval_ce_clean_loss": 0.05336422391967278,
+      "eval_ce_pred_loss": 4.809050649268442,
+      "eval_flow_cos_loss": 0.07764796834422029,
+      "eval_flow_mse_loss": 0.3626167481620562,
+      "eval_loss": 4.7155106383371574,
+      "eval_runtime": 179.0768,
+      "eval_samples_per_second": 156.318,
+      "eval_steps_per_second": 2.446,
+      "flow/cos_sim": 0.9223520577498222,
+      "flow/improvement_ratio": 0.9944308828817655,
+      "flow/mag_ratio_mean": 0.8769120071848778,
+      "flow/mag_ratio_std": 0.09842952985537651,
+      "step": 4096
+    },
+    {
+      "epoch": 0.236478684587317,
+      "grad_norm": 2.3496689796447754,
+      "learning_rate": 0.00027177281107320826,
+      "loss": 4.68638801574707,
+      "step": 5120
+    },
+    {
+      "epoch": 0.236478684587317,
+      "eval_bleu": 0.22356960405367263,
+      "eval_ce_clean_loss": 0.0327355340824858,
+      "eval_ce_pred_loss": 4.565925827853756,
+      "eval_flow_cos_loss": 0.08153306739617429,
+      "eval_flow_mse_loss": 0.3821449596860093,
+      "eval_loss": 4.515134906115597,
+      "flow/cos_sim": 0.9184669488913393,
+      "flow/improvement_ratio": 0.9950582473517553,
+      "flow/mag_ratio_mean": 0.8821974423922361,
+      "flow/mag_ratio_std": 0.09247283822937644,
+      "step": 5120
+    },
+    {
+      "epoch": 0.236478684587317,
+      "eval_bleu": 0.22356960405367263,
+      "eval_ce_clean_loss": 0.0327355340824858,
+      "eval_ce_pred_loss": 4.565925827853756,
+      "eval_flow_cos_loss": 0.08153306739617429,
+      "eval_flow_mse_loss": 0.3821449596860093,
+      "eval_loss": 4.515134906115597,
+      "eval_runtime": 181.4276,
+      "eval_samples_per_second": 154.293,
+      "eval_steps_per_second": 2.414,
+      "flow/cos_sim": 0.9184669488913393,
+      "flow/improvement_ratio": 0.9950582473517553,
+      "flow/mag_ratio_mean": 0.8821974423922361,
+      "flow/mag_ratio_std": 0.09247283822937644,
+      "step": 5120
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "grad_norm": 2.1534059047698975,
+      "learning_rate": 0.00025669014822961186,
+      "loss": 4.549870491027832,
+      "step": 6144
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "eval_bleu": 0.2167831701158986,
+      "eval_ce_clean_loss": 0.022042056864561285,
+      "eval_ce_pred_loss": 4.496095672045668,
+      "eval_flow_cos_loss": 0.07625704317366423,
+      "eval_flow_mse_loss": 0.4058228160964844,
+      "eval_loss": 4.473577274579435,
+      "flow/cos_sim": 0.9237429750020101,
+      "flow/improvement_ratio": 0.9947892213793106,
+      "flow/mag_ratio_mean": 0.8912639768972789,
+      "flow/mag_ratio_std": 0.08584192658872365,
+      "step": 6144
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "eval_bleu": 0.2167831701158986,
+      "eval_ce_clean_loss": 0.022042056864561285,
+      "eval_ce_pred_loss": 4.496095672045668,
+      "eval_flow_cos_loss": 0.07625704317366423,
+      "eval_flow_mse_loss": 0.4058228160964844,
+      "eval_loss": 4.473577274579435,
+      "eval_runtime": 180.321,
+      "eval_samples_per_second": 155.24,
+      "eval_steps_per_second": 2.429,
+      "flow/cos_sim": 0.9237429750020101,
+      "flow/improvement_ratio": 0.9947892213793106,
+      "flow/mag_ratio_mean": 0.8912639768972789,
+      "flow/mag_ratio_std": 0.08584192658872365,
+      "step": 6144
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "grad_norm": 2.1275017261505127,
+      "learning_rate": 0.0002390360415767374,
+      "loss": 4.411749362945557,
+      "step": 7168
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "eval_bleu": 0.26130694526226583,
+      "eval_ce_clean_loss": 0.015259190446582355,
+      "eval_ce_pred_loss": 4.084311145081368,
+      "eval_flow_cos_loss": 0.11306819562974586,
+      "eval_flow_mse_loss": 0.5684606986094828,
+      "eval_loss": 4.274133609310133,
+      "flow/cos_sim": 0.8869318397349963,
+      "flow/improvement_ratio": 0.9916486408068165,
+      "flow/mag_ratio_mean": 0.853514781281284,
+      "flow/mag_ratio_std": 0.09556960212449504,
+      "step": 7168
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "eval_bleu": 0.26130694526226583,
+      "eval_ce_clean_loss": 0.015259190446582355,
+      "eval_ce_pred_loss": 4.084311145081368,
+      "eval_flow_cos_loss": 0.11306819562974586,
+      "eval_flow_mse_loss": 0.5684606986094828,
+      "eval_loss": 4.274133609310133,
+      "eval_runtime": 183.0141,
+      "eval_samples_per_second": 152.955,
+      "eval_steps_per_second": 2.393,
+      "flow/cos_sim": 0.8869318397349963,
+      "flow/improvement_ratio": 0.9916486408068165,
+      "flow/mag_ratio_mean": 0.853514781281284,
+      "flow/mag_ratio_std": 0.09556960212449504,
+      "step": 7168
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "grad_norm": 2.679063081741333,
+      "learning_rate": 0.00021920458819314276,
+      "loss": 4.291924476623535,
+      "step": 8192
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "eval_bleu": 0.26455338282933316,
+      "eval_ce_clean_loss": 0.010562141490759984,
+      "eval_ce_pred_loss": 4.048519775203374,
+      "eval_flow_cos_loss": 0.08972637632598071,
+      "eval_flow_mse_loss": 0.5574726984245048,
+      "eval_loss": 4.224628202447064,
+      "flow/cos_sim": 0.9102736386262118,
+      "flow/improvement_ratio": 0.9913910556601607,
+      "flow/mag_ratio_mean": 0.8761251777002256,
+      "flow/mag_ratio_std": 0.08667415052176067,
+      "step": 8192
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "eval_bleu": 0.26455338282933316,
+      "eval_ce_clean_loss": 0.010562141490759984,
+      "eval_ce_pred_loss": 4.048519775203374,
+      "eval_flow_cos_loss": 0.08972637632598071,
+      "eval_flow_mse_loss": 0.5574726984245048,
+      "eval_loss": 4.224628202447064,
+      "eval_runtime": 182.7475,
+      "eval_samples_per_second": 153.179,
+      "eval_steps_per_second": 2.397,
+      "flow/cos_sim": 0.9102736386262118,
+      "flow/improvement_ratio": 0.9913910556601607,
+      "flow/mag_ratio_mean": 0.8761251777002256,
+      "flow/mag_ratio_std": 0.08667415052176067,
+      "step": 8192
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "grad_norm": 3.4350733757019043,
+      "learning_rate": 0.0001976932406046495,
+      "loss": 4.231810569763184,
+      "step": 9216
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "eval_bleu": 0.27369821400928834,
+      "eval_ce_clean_loss": 0.007689533357032188,
+      "eval_ce_pred_loss": 3.966488937809043,
+      "eval_flow_cos_loss": 0.08683863282203674,
+      "eval_flow_mse_loss": 0.569647965025684,
+      "eval_loss": 4.161966539409063,
+      "flow/cos_sim": 0.9131613878626802,
+      "flow/improvement_ratio": 0.9904770306800599,
+      "flow/mag_ratio_mean": 0.8949770351795301,
+      "flow/mag_ratio_std": 0.08686588956340807,
+      "step": 9216
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "eval_bleu": 0.27369821400928834,
+      "eval_ce_clean_loss": 0.007689533357032188,
+      "eval_ce_pred_loss": 3.966488937809043,
+      "eval_flow_cos_loss": 0.08683863282203674,
+      "eval_flow_mse_loss": 0.569647965025684,
+      "eval_loss": 4.161966539409063,
+      "eval_runtime": 191.5369,
+      "eval_samples_per_second": 146.149,
+      "eval_steps_per_second": 2.287,
+      "flow/cos_sim": 0.9131613878626802,
+      "flow/improvement_ratio": 0.9904770306800599,
+      "flow/mag_ratio_mean": 0.8949770351795301,
+      "flow/mag_ratio_std": 0.08686588956340807,
+      "step": 9216
+    },
+    {
+      "epoch": 0.472957369174634,
+      "grad_norm": 2.2814252376556396,
+      "learning_rate": 0.00017504669739254724,
+      "loss": 4.164244174957275,
+      "step": 10240
+    },
+    {
+      "epoch": 0.472957369174634,
+      "eval_bleu": 0.30777640553962915,
+      "eval_ce_clean_loss": 0.0060966779907408375,
+      "eval_ce_pred_loss": 3.6631897802222264,
+      "eval_flow_cos_loss": 0.09066317927891805,
+      "eval_flow_mse_loss": 0.6576739563245207,
+      "eval_loss": 3.9778201345983706,
+      "flow/cos_sim": 0.9093368449439742,
+      "flow/improvement_ratio": 0.9863403809669355,
+      "flow/mag_ratio_mean": 0.8948755321437365,
+      "flow/mag_ratio_std": 0.0903660870203961,
+      "step": 10240
+    },
+    {
+      "epoch": 0.472957369174634,
+      "eval_bleu": 0.30777640553962915,
+      "eval_ce_clean_loss": 0.0060966779907408375,
+      "eval_ce_pred_loss": 3.6631897802222264,
+      "eval_flow_cos_loss": 0.09066317927891805,
+      "eval_flow_mse_loss": 0.6576739563245207,
+      "eval_loss": 3.9778201345983706,
+      "eval_runtime": 189.8476,
+      "eval_samples_per_second": 147.45,
+      "eval_steps_per_second": 2.307,
+      "flow/cos_sim": 0.9093368449439742,
+      "flow/improvement_ratio": 0.9863403809669355,
+      "flow/mag_ratio_mean": 0.8948755321437365,
+      "flow/mag_ratio_std": 0.0903660870203961,
+      "step": 10240
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "grad_norm": 2.119964122772217,
+      "learning_rate": 0.00015177050305675404,
+      "loss": 4.065644264221191,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "eval_bleu": 0.2925834655122522,
+      "eval_ce_clean_loss": 0.004753688909139986,
+      "eval_ce_pred_loss": 3.7541082803517174,
+      "eval_flow_cos_loss": 0.08232224217220528,
+      "eval_flow_mse_loss": 0.621043747554631,
+      "eval_loss": 4.020797040908849,
+      "flow/cos_sim": 0.9176777691057284,
+      "flow/improvement_ratio": 0.9826103085524416,
+      "flow/mag_ratio_mean": 0.8994121777412554,
+      "flow/mag_ratio_std": 0.09006089084360698,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "eval_bleu": 0.2925834655122522,
+      "eval_ce_clean_loss": 0.004753688909139986,
+      "eval_ce_pred_loss": 3.7541082803517174,
+      "eval_flow_cos_loss": 0.08232224217220528,
+      "eval_flow_mse_loss": 0.621043747554631,
+      "eval_loss": 4.020797040908849,
+      "eval_runtime": 189.3251,
+      "eval_samples_per_second": 147.857,
+      "eval_steps_per_second": 2.313,
+      "flow/cos_sim": 0.9176777691057284,
+      "flow/improvement_ratio": 0.9826103085524416,
+      "flow/mag_ratio_mean": 0.8994121777412554,
+      "flow/mag_ratio_std": 0.09006089084360698,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5675488430095608,
+      "grad_norm": 2.5909643173217773,
+      "learning_rate": 0.00012845133096612622,
+      "loss": 4.054959774017334,
+      "step": 12288
+    },
+    {
+      "epoch": 0.5675488430095608,
+      "eval_bleu": 0.28965271334128717,
+      "eval_ce_clean_loss": 0.004024346635361164,
+      "eval_ce_pred_loss": 3.718931865474405,
+      "eval_flow_cos_loss": 0.08415858520696697,
+      "eval_flow_mse_loss": 0.6252854074517341,
+      "eval_loss": 3.9937660802989248,
+      "flow/cos_sim": 0.9158414309699786,
+      "flow/improvement_ratio": 0.9831551257605966,
+      "flow/mag_ratio_mean": 0.8911215496934168,
+      "flow/mag_ratio_std": 0.08892078035974611,
+      "step": 12288
+    },
+    {
+      "epoch": 0.5675488430095608,
+      "eval_bleu": 0.28965271334128717,
+      "eval_ce_clean_loss": 0.004024346635361164,
+      "eval_ce_pred_loss": 3.718931865474405,
+      "eval_flow_cos_loss": 0.08415858520696697,
+      "eval_flow_mse_loss": 0.6252854074517341,
+      "eval_loss": 3.9937660802989248,
+      "eval_runtime": 188.6951,
+      "eval_samples_per_second": 148.35,
+      "eval_steps_per_second": 2.321,
+      "flow/cos_sim": 0.9158414309699786,
+      "flow/improvement_ratio": 0.9831551257605966,
+      "flow/mag_ratio_mean": 0.8911215496934168,
+      "flow/mag_ratio_std": 0.08892078035974611,
+      "step": 12288
+    }
+  ],
+  "logging_steps": 1024,
+  "max_steps": 21651,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1024,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoints-v5.1/checkpoint-12288/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e76013f70108cb4ddda27ec8328cf4ef6edec61c75334ae3d6e75eb2e082915f
+size 5137