Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip 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/checkpoint-12288/ema.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2defc5208e124177d996dd755609c9e3757f2a1ee1c8e9f2af5bf89bd9edd29d
+size 54599376

checkpoints-v5/checkpoint-12288/eval_state.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22cfc665216306d46d674ac631e4e1d7c86ff4e652f0b919e5c0c5830ad0644e
+size 41362527

checkpoints-v5/checkpoint-12288/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c66219640cb42428556fbb67d330b92a0701429c1736b06112bdb8306325d46
+size 54599408

checkpoints-v5/checkpoint-12288/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4cbe31070a18449cf928de26997b37cbce3899a755132bd6a49f862bbb75a129
+size 76550347

checkpoints-v5/checkpoint-12288/rng_state.pth

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

checkpoints-v5/checkpoint-12288/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee9cf5cfca6c27607a4ceb97fc3ab7572a75442fb2b962c24370088b7a473bdc
+size 1383

checkpoints-v5/checkpoint-12288/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:846ae5b395c49f67b65d2e1673da9babbf27b6bd69b624764425f9c10b173d5e
+size 1465

checkpoints-v5/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": 4.013998985290527,
+      "learning_rate": 0.00029970703124999995,
+      "loss": 7.93059778213501,
+      "step": 1024
+    },
+    {
+      "epoch": 0.047295736917463395,
+      "eval_bleu": 0.2254043225811079,
+      "eval_ce_clean_loss": 0.2796171638927503,
+      "eval_ce_pred_loss": 4.793793409926706,
+      "eval_flow_cos_loss": 0.21001066447801242,
+      "eval_flow_mse_loss": 0.5274724085309189,
+      "eval_loss": 3.2564889513738624,
+      "flow/cos_sim": 0.7899893473272455,
+      "flow/improvement_ratio": 0.9936718939374026,
+      "flow/mag_ratio_mean": 0.691851259502646,
+      "flow/mag_ratio_std": 0.1381261397884589,
+      "step": 1024
+    },
+    {
+      "epoch": 0.047295736917463395,
+      "eval_bleu": 0.2254043225811079,
+      "eval_ce_clean_loss": 0.2796171638927503,
+      "eval_ce_pred_loss": 4.793793409926706,
+      "eval_flow_cos_loss": 0.21001066447801242,
+      "eval_flow_mse_loss": 0.5274724085309189,
+      "eval_loss": 3.2564889513738624,
+      "eval_runtime": 186.2837,
+      "eval_samples_per_second": 150.271,
+      "eval_steps_per_second": 2.351,
+      "flow/cos_sim": 0.7899893473272455,
+      "flow/improvement_ratio": 0.9936718939374026,
+      "flow/mag_ratio_mean": 0.691851259502646,
+      "flow/mag_ratio_std": 0.1381261397884589,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "grad_norm": 2.68375301361084,
+      "learning_rate": 0.00029818297300322,
+      "loss": 3.0190515518188477,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "eval_bleu": 0.2379161607125303,
+      "eval_ce_clean_loss": 0.048818679968720156,
+      "eval_ce_pred_loss": 4.610883573418882,
+      "eval_flow_cos_loss": 0.06802858637654347,
+      "eval_flow_mse_loss": 0.43109905474806487,
+      "eval_loss": 2.802366664420524,
+      "flow/cos_sim": 0.931971434588846,
+      "flow/improvement_ratio": 0.9953692990895275,
+      "flow/mag_ratio_mean": 0.909526935042856,
+      "flow/mag_ratio_std": 0.09037123139114141,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "eval_bleu": 0.2379161607125303,
+      "eval_ce_clean_loss": 0.048818679968720156,
+      "eval_ce_pred_loss": 4.610883573418882,
+      "eval_flow_cos_loss": 0.06802858637654347,
+      "eval_flow_mse_loss": 0.43109905474806487,
+      "eval_loss": 2.802366664420524,
+      "eval_runtime": 179.0383,
+      "eval_samples_per_second": 156.352,
+      "eval_steps_per_second": 2.446,
+      "flow/cos_sim": 0.931971434588846,
+      "flow/improvement_ratio": 0.9953692990895275,
+      "flow/mag_ratio_mean": 0.909526935042856,
+      "flow/mag_ratio_std": 0.09037123139114141,
+      "step": 2048
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "grad_norm": 1.4580802917480469,
+      "learning_rate": 0.0002927689070858589,
+      "loss": 2.7789387702941895,
+      "step": 3072
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "eval_bleu": 0.2490751213264231,
+      "eval_ce_clean_loss": 0.018171085729412565,
+      "eval_ce_pred_loss": 4.420184469114155,
+      "eval_flow_cos_loss": 0.06013645735233342,
+      "eval_flow_mse_loss": 0.4371357409104909,
+      "eval_loss": 2.6804331704361797,
+      "flow/cos_sim": 0.939863565986015,
+      "flow/improvement_ratio": 0.9952474803957221,
+      "flow/mag_ratio_mean": 0.9349337605308724,
+      "flow/mag_ratio_std": 0.0872031468302692,
+      "step": 3072
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "eval_bleu": 0.2490751213264231,
+      "eval_ce_clean_loss": 0.018171085729412565,
+      "eval_ce_pred_loss": 4.420184469114155,
+      "eval_flow_cos_loss": 0.06013645735233342,
+      "eval_flow_mse_loss": 0.4371357409104909,
+      "eval_loss": 2.6804331704361797,
+      "eval_runtime": 179.2795,
+      "eval_samples_per_second": 156.142,
+      "eval_steps_per_second": 2.443,
+      "flow/cos_sim": 0.939863565986015,
+      "flow/improvement_ratio": 0.9952474803957221,
+      "flow/mag_ratio_mean": 0.9349337605308724,
+      "flow/mag_ratio_std": 0.0872031468302692,
+      "step": 3072
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "grad_norm": 1.043502926826477,
+      "learning_rate": 0.00028389952193475995,
+      "loss": 2.6741526126861572,
+      "step": 4096
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "eval_bleu": 0.28737562974532166,
+      "eval_ce_clean_loss": 0.008730428965792026,
+      "eval_ce_pred_loss": 4.164722618447047,
+      "eval_flow_cos_loss": 0.0743819943301754,
+      "eval_flow_mse_loss": 0.48018580021923535,
+      "eval_loss": 2.5898730254064413,
+      "flow/cos_sim": 0.9256180300288004,
+      "flow/improvement_ratio": 0.9946817859122742,
+      "flow/mag_ratio_mean": 0.9113472489189339,
+      "flow/mag_ratio_std": 0.09068145650570796,
+      "step": 4096
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "eval_bleu": 0.28737562974532166,
+      "eval_ce_clean_loss": 0.008730428965792026,
+      "eval_ce_pred_loss": 4.164722618447047,
+      "eval_flow_cos_loss": 0.0743819943301754,
+      "eval_flow_mse_loss": 0.48018580021923535,
+      "eval_loss": 2.5898730254064413,
+      "eval_runtime": 180.7231,
+      "eval_samples_per_second": 154.894,
+      "eval_steps_per_second": 2.424,
+      "flow/cos_sim": 0.9256180300288004,
+      "flow/improvement_ratio": 0.9946817859122742,
+      "flow/mag_ratio_mean": 0.9113472489189339,
+      "flow/mag_ratio_std": 0.09068145650570796,
+      "step": 4096
+    },
+    {
+      "epoch": 0.236478684587317,
+      "grad_norm": 1.0432350635528564,
+      "learning_rate": 0.00027177281107320826,
+      "loss": 2.5878536701202393,
+      "step": 5120
+    },
+    {
+      "epoch": 0.236478684587317,
+      "eval_bleu": 0.3171730575714231,
+      "eval_ce_clean_loss": 0.0047634225328065894,
+      "eval_ce_pred_loss": 3.9123527883939,
+      "eval_flow_cos_loss": 0.11630631098736367,
+      "eval_flow_mse_loss": 0.5211573920157402,
+      "eval_loss": 2.5111737806503087,
+      "flow/cos_sim": 0.8836937101736461,
+      "flow/improvement_ratio": 0.995179520213985,
+      "flow/mag_ratio_mean": 0.8669234891188199,
+      "flow/mag_ratio_std": 0.10931324717252766,
+      "step": 5120
+    },
+    {
+      "epoch": 0.236478684587317,
+      "eval_bleu": 0.3171730575714231,
+      "eval_ce_clean_loss": 0.0047634225328065894,
+      "eval_ce_pred_loss": 3.9123527883939,
+      "eval_flow_cos_loss": 0.11630631098736367,
+      "eval_flow_mse_loss": 0.5211573920157402,
+      "eval_loss": 2.5111737806503087,
+      "eval_runtime": 183.1981,
+      "eval_samples_per_second": 152.802,
+      "eval_steps_per_second": 2.391,
+      "flow/cos_sim": 0.8836937101736461,
+      "flow/improvement_ratio": 0.995179520213985,
+      "flow/mag_ratio_mean": 0.8669234891188199,
+      "flow/mag_ratio_std": 0.10931324717252766,
+      "step": 5120
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "grad_norm": 1.4073700904846191,
+      "learning_rate": 0.00025669014822961186,
+      "loss": 2.5205326080322266,
+      "step": 6144
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "eval_bleu": 0.3286377886298107,
+      "eval_ce_clean_loss": 0.0029047091786412642,
+      "eval_ce_pred_loss": 3.744860372586882,
+      "eval_flow_cos_loss": 0.13526482834187273,
+      "eval_flow_mse_loss": 0.5445779024465988,
+      "eval_loss": 2.4537290106080984,
+      "flow/cos_sim": 0.8647351836505002,
+      "flow/improvement_ratio": 0.9945465897860593,
+      "flow/mag_ratio_mean": 0.8419829230058139,
+      "flow/mag_ratio_std": 0.10719822894900902,
+      "step": 6144
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "eval_bleu": 0.3286377886298107,
+      "eval_ce_clean_loss": 0.0029047091786412642,
+      "eval_ce_pred_loss": 3.744860372586882,
+      "eval_flow_cos_loss": 0.13526482834187273,
+      "eval_flow_mse_loss": 0.5445779024465988,
+      "eval_loss": 2.4537290106080984,
+      "eval_runtime": 183.0814,
+      "eval_samples_per_second": 152.899,
+      "eval_steps_per_second": 2.392,
+      "flow/cos_sim": 0.8647351836505002,
+      "flow/improvement_ratio": 0.9945465897860593,
+      "flow/mag_ratio_mean": 0.8419829230058139,
+      "flow/mag_ratio_std": 0.10719822894900902,
+      "step": 6144
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "grad_norm": 0.6808182001113892,
+      "learning_rate": 0.00023901765476336221,
+      "loss": 2.4522030353546143,
+      "step": 7168
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "eval_bleu": 0.36087564863431026,
+      "eval_ce_clean_loss": 0.0020003622779784804,
+      "eval_ce_pred_loss": 3.5444279462779495,
+      "eval_flow_cos_loss": 0.1795310482859067,
+      "eval_flow_mse_loss": 0.5883507338696963,
+      "eval_loss": 2.4074478285497727,
+      "flow/cos_sim": 0.8204689798834117,
+      "flow/improvement_ratio": 0.9948837053558054,
+      "flow/mag_ratio_mean": 0.7740725910282571,
+      "flow/mag_ratio_std": 0.12248109917254209,
+      "step": 7168
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "eval_bleu": 0.36087564863431026,
+      "eval_ce_clean_loss": 0.0020003622779784804,
+      "eval_ce_pred_loss": 3.5444279462779495,
+      "eval_flow_cos_loss": 0.1795310482859067,
+      "eval_flow_mse_loss": 0.5883507338696963,
+      "eval_loss": 2.4074478285497727,
+      "eval_runtime": 189.0036,
+      "eval_samples_per_second": 148.108,
+      "eval_steps_per_second": 2.317,
+      "flow/cos_sim": 0.8204689798834117,
+      "flow/improvement_ratio": 0.9948837053558054,
+      "flow/mag_ratio_mean": 0.7740725910282571,
+      "flow/mag_ratio_std": 0.12248109917254209,
+      "step": 7168
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "grad_norm": 0.8293004035949707,
+      "learning_rate": 0.00021918431840685083,
+      "loss": 2.415977954864502,
+      "step": 8192
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "eval_bleu": 0.3942637509636107,
+      "eval_ce_clean_loss": 0.0015604201352160496,
+      "eval_ce_pred_loss": 3.315679441848302,
+      "eval_flow_cos_loss": 0.2214143528649796,
+      "eval_flow_mse_loss": 0.6519387753586791,
+      "eval_loss": 2.3666924983943436,
+      "flow/cos_sim": 0.7785856559124167,
+      "flow/improvement_ratio": 0.9936508686030836,
+      "flow/mag_ratio_mean": 0.7346892050684315,
+      "flow/mag_ratio_std": 0.1360737154234762,
+      "step": 8192
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "eval_bleu": 0.3942637509636107,
+      "eval_ce_clean_loss": 0.0015604201352160496,
+      "eval_ce_pred_loss": 3.315679441848302,
+      "eval_flow_cos_loss": 0.2214143528649796,
+      "eval_flow_mse_loss": 0.6519387753586791,
+      "eval_loss": 2.3666924983943436,
+      "eval_runtime": 185.161,
+      "eval_samples_per_second": 151.182,
+      "eval_steps_per_second": 2.366,
+      "flow/cos_sim": 0.7785856559124167,
+      "flow/improvement_ratio": 0.9936508686030836,
+      "flow/mag_ratio_mean": 0.7346892050684315,
+      "flow/mag_ratio_std": 0.1360737154234762,
+      "step": 8192
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "grad_norm": 0.7353112101554871,
+      "learning_rate": 0.0001976932406046495,
+      "loss": 2.399784803390503,
+      "step": 9216
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "eval_bleu": 0.3998534289221981,
+      "eval_ce_clean_loss": 0.0012517906730888533,
+      "eval_ce_pred_loss": 3.2301429374032913,
+      "eval_flow_cos_loss": 0.2110011734442624,
+      "eval_flow_mse_loss": 0.6800476714084137,
+      "eval_loss": 2.349121227384158,
+      "flow/cos_sim": 0.7889988516291527,
+      "flow/improvement_ratio": 0.9947260159607891,
+      "flow/mag_ratio_mean": 0.7546588749918219,
+      "flow/mag_ratio_std": 0.1336971542640636,
+      "step": 9216
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "eval_bleu": 0.3998534289221981,
+      "eval_ce_clean_loss": 0.0012517906730888533,
+      "eval_ce_pred_loss": 3.2301429374032913,
+      "eval_flow_cos_loss": 0.2110011734442624,
+      "eval_flow_mse_loss": 0.6800476714084137,
+      "eval_loss": 2.349121227384158,
+      "eval_runtime": 183.7009,
+      "eval_samples_per_second": 152.384,
+      "eval_steps_per_second": 2.384,
+      "flow/cos_sim": 0.7889988516291527,
+      "flow/improvement_ratio": 0.9947260159607891,
+      "flow/mag_ratio_mean": 0.7546588749918219,
+      "flow/mag_ratio_std": 0.1336971542640636,
+      "step": 9216
+    },
+    {
+      "epoch": 0.472957369174634,
+      "grad_norm": 0.8551133871078491,
+      "learning_rate": 0.00017502417211070142,
+      "loss": 2.3940861225128174,
+      "step": 10240
+    },
+    {
+      "epoch": 0.472957369174634,
+      "eval_bleu": 0.4267550667680489,
+      "eval_ce_clean_loss": 0.0010826060028708716,
+      "eval_ce_pred_loss": 3.04460582885568,
+      "eval_flow_cos_loss": 0.27040740272634106,
+      "eval_flow_mse_loss": 0.7608528980925747,
+      "eval_loss": 2.351840272341689,
+      "flow/cos_sim": 0.7295926203738609,
+      "flow/improvement_ratio": 0.9943434236800834,
+      "flow/mag_ratio_mean": 0.6815646335686723,
+      "flow/mag_ratio_std": 0.15056844773493946,
+      "step": 10240
+    },
+    {
+      "epoch": 0.472957369174634,
+      "eval_bleu": 0.4267550667680489,
+      "eval_ce_clean_loss": 0.0010826060028708716,
+      "eval_ce_pred_loss": 3.04460582885568,
+      "eval_flow_cos_loss": 0.27040740272634106,
+      "eval_flow_mse_loss": 0.7608528980925747,
+      "eval_loss": 2.351840272341689,
+      "eval_runtime": 183.7516,
+      "eval_samples_per_second": 152.342,
+      "eval_steps_per_second": 2.384,
+      "flow/cos_sim": 0.7295926203738609,
+      "flow/improvement_ratio": 0.9943434236800834,
+      "flow/mag_ratio_mean": 0.6815646335686723,
+      "flow/mag_ratio_std": 0.15056844773493946,
+      "step": 10240
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "grad_norm": 1.0387604236602783,
+      "learning_rate": 0.00015174765889655888,
+      "loss": 2.3844668865203857,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "eval_bleu": 0.4113314804835657,
+      "eval_ce_clean_loss": 0.0009574266697204226,
+      "eval_ce_pred_loss": 3.012512480287247,
+      "eval_flow_cos_loss": 0.2047049244312935,
+      "eval_flow_mse_loss": 0.7820947959542819,
+      "eval_loss": 2.3404846879989587,
+      "flow/cos_sim": 0.7952951010503725,
+      "flow/improvement_ratio": 0.9946630915550336,
+      "flow/mag_ratio_mean": 0.771922899979979,
+      "flow/mag_ratio_std": 0.1377742688541543,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "eval_bleu": 0.4113314804835657,
+      "eval_ce_clean_loss": 0.0009574266697204226,
+      "eval_ce_pred_loss": 3.012512480287247,
+      "eval_flow_cos_loss": 0.2047049244312935,
+      "eval_flow_mse_loss": 0.7820947959542819,
+      "eval_loss": 2.3404846879989587,
+      "eval_runtime": 183.6789,
+      "eval_samples_per_second": 152.402,
+      "eval_steps_per_second": 2.385,
+      "flow/cos_sim": 0.7952951010503725,
+      "flow/improvement_ratio": 0.9946630915550336,
+      "flow/mag_ratio_mean": 0.771922899979979,
+      "flow/mag_ratio_std": 0.1377742688541543,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5675488430095608,
+      "grad_norm": 1.0685348510742188,
+      "learning_rate": 0.0001284287224539987,
+      "loss": 2.3792426586151123,
+      "step": 12288
+    },
+    {
+      "epoch": 0.5675488430095608,
+      "eval_bleu": 0.43554170247266144,
+      "eval_ce_clean_loss": 0.0008726873317618188,
+      "eval_ce_pred_loss": 2.8611030374487787,
+      "eval_flow_cos_loss": 0.1863578746444014,
+      "eval_flow_mse_loss": 0.8412681570336155,
+      "eval_loss": 2.3192818273692373,
+      "flow/cos_sim": 0.8136421497826163,
+      "flow/improvement_ratio": 0.9942930025052806,
+      "flow/mag_ratio_mean": 0.8013964615183878,
+      "flow/mag_ratio_std": 0.1349830417872564,
+      "step": 12288
+    },
+    {
+      "epoch": 0.5675488430095608,
+      "eval_bleu": 0.43554170247266144,
+      "eval_ce_clean_loss": 0.0008726873317618188,
+      "eval_ce_pred_loss": 2.8611030374487787,
+      "eval_flow_cos_loss": 0.1863578746444014,
+      "eval_flow_mse_loss": 0.8412681570336155,
+      "eval_loss": 2.3192818273692373,
+      "eval_runtime": 183.8808,
+      "eval_samples_per_second": 152.234,
+      "eval_steps_per_second": 2.382,
+      "flow/cos_sim": 0.8136421497826163,
+      "flow/improvement_ratio": 0.9942930025052806,
+      "flow/mag_ratio_mean": 0.8013964615183878,
+      "flow/mag_ratio_std": 0.1349830417872564,
+      "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/checkpoint-12288/training_args.bin

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