Upload folder using huggingface_hub

.gitattributes

@@ -44,3 +44,4 @@ checkpoints-v4.4+/checkpoint-7168/eval_state.json filter=lfs diff=lfs merge=lfs
 checkpoints-v5.5/checkpoint-24576/eval_state.json filter=lfs diff=lfs merge=lfs -text
 checkpoints-v5.6/checkpoint-4096/eval_state.json filter=lfs diff=lfs merge=lfs -text
 checkpoints-v5.6/checkpoint-14336/eval_state.json filter=lfs diff=lfs merge=lfs -text
+checkpoints-v5.7/checkpoint-11264/eval_state.json filter=lfs diff=lfs merge=lfs -text

checkpoints-v5.7/checkpoint-11264/ema.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfc5af0310f2166d9003e76c133aeba67efed19fcb458295ca1dc9a1a612c620
+size 54599592

checkpoints-v5.7/checkpoint-11264/eval_state.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79201eb36a32062afb99bd12f11f01a97b2ead8283ae7d6c75b8c34efc186176
+size 59985045

checkpoints-v5.7/checkpoint-11264/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e6d75bd60fc632256a58d92778398dc8f1ee608e7057b89e36759c6ba050ed2
+size 54599624

checkpoints-v5.7/checkpoint-11264/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c832b88f473165d599a5184fd54e164ebd0a61b9be83c8bd7757fda5ce0b1e29
+size 76550347

checkpoints-v5.7/checkpoint-11264/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3445e2029753be8bf32790ad2df6092dc63f2028e7603a9903f650b4f19cbab4
+size 14645

checkpoints-v5.7/checkpoint-11264/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79701bbdc0c49714e0e085b5c67881fa167d2bf0d010e2195ebdf057057d72dd
+size 1383

checkpoints-v5.7/checkpoint-11264/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19ad511dc53ec97f3489767f5cff37a551a58fb728bc82956d6ea6fdb5514518
+size 1465

checkpoints-v5.7/checkpoint-11264/trainer_state.json

@@ -0,0 +1,452 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4854440062921542,
+  "eval_steps": 1024,
+  "global_step": 11264,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.04413127329928675,
+      "grad_norm": 3.5420610904693604,
+      "learning_rate": 9.990234375e-05,
+      "loss": 15.012842178344727,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04413127329928675,
+      "eval_bleu": 0.1625731794882785,
+      "eval_ce_clean_loss": 2.213519910759509,
+      "eval_ce_pred_loss": 5.054043449572663,
+      "eval_flow_cos_loss": 0.47666819663698484,
+      "eval_flow_mse_loss": 1.3785955913539634,
+      "eval_loss": 8.765326018018255,
+      "flow/cos_sim": 0.523331833546604,
+      "flow/improvement_ratio": 0.9235056383269173,
+      "flow/mag_ratio_mean": 0.0070004605495535745,
+      "flow/mag_ratio_std": 0.0032371597157032695,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04413127329928675,
+      "eval_bleu": 0.1625731794882785,
+      "eval_ce_clean_loss": 2.213519910759509,
+      "eval_ce_pred_loss": 5.054043449572663,
+      "eval_flow_cos_loss": 0.47666819663698484,
+      "eval_flow_mse_loss": 1.3785955913539634,
+      "eval_loss": 8.765326018018255,
+      "eval_runtime": 203.2491,
+      "eval_samples_per_second": 147.602,
+      "eval_steps_per_second": 2.308,
+      "flow/cos_sim": 0.523331833546604,
+      "flow/improvement_ratio": 0.9235056383269173,
+      "flow/mag_ratio_mean": 0.0070004605495535745,
+      "flow/mag_ratio_std": 0.0032371597157032695,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0882625465985735,
+      "grad_norm": 1.748165249824524,
+      "learning_rate": 9.9476028157316e-05,
+      "loss": 6.209901809692383,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0882625465985735,
+      "eval_bleu": 0.3936336657643431,
+      "eval_ce_clean_loss": 0.36314720633441705,
+      "eval_ce_pred_loss": 3.07268785362813,
+      "eval_flow_cos_loss": 0.4477332118096382,
+      "eval_flow_mse_loss": 1.4534405088628024,
+      "eval_loss": 5.001208877766818,
+      "flow/cos_sim": 0.552266787110107,
+      "flow/improvement_ratio": 0.9922580608426889,
+      "flow/mag_ratio_mean": 0.4709286067023206,
+      "flow/mag_ratio_std": 0.1402487441230176,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0882625465985735,
+      "eval_bleu": 0.3936336657643431,
+      "eval_ce_clean_loss": 0.36314720633441705,
+      "eval_ce_pred_loss": 3.07268785362813,
+      "eval_flow_cos_loss": 0.4477332118096382,
+      "eval_flow_mse_loss": 1.4534405088628024,
+      "eval_loss": 5.001208877766818,
+      "eval_runtime": 202.5752,
+      "eval_samples_per_second": 148.093,
+      "eval_steps_per_second": 2.315,
+      "flow/cos_sim": 0.552266787110107,
+      "flow/improvement_ratio": 0.9922580608426889,
+      "flow/mag_ratio_mean": 0.4709286067023206,
+      "flow/mag_ratio_std": 0.1402487441230176,
+      "step": 2048
+    },
+    {
+      "epoch": 0.13239381989786023,
+      "grad_norm": 2.414154529571533,
+      "learning_rate": 9.791307026072513e-05,
+      "loss": 4.706303596496582,
+      "step": 3072
+    },
+    {
+      "epoch": 0.13239381989786023,
+      "eval_bleu": 0.441628158711516,
+      "eval_ce_clean_loss": 0.1436088167464555,
+      "eval_ce_pred_loss": 2.6581369013166123,
+      "eval_flow_cos_loss": 0.308133079616754,
+      "eval_flow_mse_loss": 1.4948640911818059,
+      "eval_loss": 4.373643074462663,
+      "flow/cos_sim": 0.6918669179050144,
+      "flow/improvement_ratio": 0.9946666356088765,
+      "flow/mag_ratio_mean": 0.6559606152556853,
+      "flow/mag_ratio_std": 0.1298362000632896,
+      "step": 3072
+    },
+    {
+      "epoch": 0.13239381989786023,
+      "eval_bleu": 0.441628158711516,
+      "eval_ce_clean_loss": 0.1436088167464555,
+      "eval_ce_pred_loss": 2.6581369013166123,
+      "eval_flow_cos_loss": 0.308133079616754,
+      "eval_flow_mse_loss": 1.4948640911818059,
+      "eval_loss": 4.373643074462663,
+      "eval_runtime": 204.4075,
+      "eval_samples_per_second": 146.766,
+      "eval_steps_per_second": 2.294,
+      "flow/cos_sim": 0.6918669179050144,
+      "flow/improvement_ratio": 0.9946666356088765,
+      "flow/mag_ratio_mean": 0.6559606152556853,
+      "flow/mag_ratio_std": 0.1298362000632896,
+      "step": 3072
+    },
+    {
+      "epoch": 0.176525093197147,
+      "grad_norm": 2.3635289669036865,
+      "learning_rate": 9.53439476074686e-05,
+      "loss": 4.239753723144531,
+      "step": 4096
+    },
+    {
+      "epoch": 0.176525093197147,
+      "eval_bleu": 0.4615621334542706,
+      "eval_ce_clean_loss": 0.07557619481420974,
+      "eval_ce_pred_loss": 2.4599488297505165,
+      "eval_flow_cos_loss": 0.19299609037731758,
+      "eval_flow_mse_loss": 1.4604684709231737,
+      "eval_loss": 4.044242526930787,
+      "flow/cos_sim": 0.8070039445403289,
+      "flow/improvement_ratio": 0.9945209180114112,
+      "flow/mag_ratio_mean": 0.7952993317707769,
+      "flow/mag_ratio_std": 0.10881571144437485,
+      "step": 4096
+    },
+    {
+      "epoch": 0.176525093197147,
+      "eval_bleu": 0.4615621334542706,
+      "eval_ce_clean_loss": 0.07557619481420974,
+      "eval_ce_pred_loss": 2.4599488297505165,
+      "eval_flow_cos_loss": 0.19299609037731758,
+      "eval_flow_mse_loss": 1.4604684709231737,
+      "eval_loss": 4.044242526930787,
+      "eval_runtime": 202.0002,
+      "eval_samples_per_second": 148.515,
+      "eval_steps_per_second": 2.322,
+      "flow/cos_sim": 0.8070039445403289,
+      "flow/improvement_ratio": 0.9945209180114112,
+      "flow/mag_ratio_mean": 0.7952993317707769,
+      "flow/mag_ratio_std": 0.10881571144437485,
+      "step": 4096
+    },
+    {
+      "epoch": 0.22065636649643372,
+      "grad_norm": 2.678138494491577,
+      "learning_rate": 9.18264920723673e-05,
+      "loss": 4.034201622009277,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22065636649643372,
+      "eval_bleu": 0.4705916689295048,
+      "eval_ce_clean_loss": 0.0494267504447813,
+      "eval_ce_pred_loss": 2.3516790477960092,
+      "eval_flow_cos_loss": 0.15000832287360355,
+      "eval_flow_mse_loss": 1.4484400370481934,
+      "eval_loss": 3.88704792268749,
+      "flow/cos_sim": 0.8499916934255344,
+      "flow/improvement_ratio": 0.994823846608591,
+      "flow/mag_ratio_mean": 0.8384634782510526,
+      "flow/mag_ratio_std": 0.10320697241881763,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22065636649643372,
+      "eval_bleu": 0.4705916689295048,
+      "eval_ce_clean_loss": 0.0494267504447813,
+      "eval_ce_pred_loss": 2.3516790477960092,
+      "eval_flow_cos_loss": 0.15000832287360355,
+      "eval_flow_mse_loss": 1.4484400370481934,
+      "eval_loss": 3.88704792268749,
+      "eval_runtime": 203.0161,
+      "eval_samples_per_second": 147.772,
+      "eval_steps_per_second": 2.31,
+      "flow/cos_sim": 0.8499916934255344,
+      "flow/improvement_ratio": 0.994823846608591,
+      "flow/mag_ratio_mean": 0.8384634782510526,
+      "flow/mag_ratio_std": 0.10320697241881763,
+      "step": 5120
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "grad_norm": 2.697188377380371,
+      "learning_rate": 8.742770483354739e-05,
+      "loss": 3.91947078704834,
+      "step": 6144
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "eval_bleu": 0.4864538614651346,
+      "eval_ce_clean_loss": 0.035138009219313225,
+      "eval_ce_pred_loss": 2.2508087689434286,
+      "eval_flow_cos_loss": 0.12960540185541486,
+      "eval_flow_mse_loss": 1.4621114108099866,
+      "eval_loss": 3.780459539722532,
+      "flow/cos_sim": 0.8703946246267128,
+      "flow/improvement_ratio": 0.9945963007300648,
+      "flow/mag_ratio_mean": 0.8639802461239829,
+      "flow/mag_ratio_std": 0.10180102466647305,
+      "step": 6144
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "eval_bleu": 0.4864538614651346,
+      "eval_ce_clean_loss": 0.035138009219313225,
+      "eval_ce_pred_loss": 2.2508087689434286,
+      "eval_flow_cos_loss": 0.12960540185541486,
+      "eval_flow_mse_loss": 1.4621114108099866,
+      "eval_loss": 3.780459539722532,
+      "eval_runtime": 205.5395,
+      "eval_samples_per_second": 145.957,
+      "eval_steps_per_second": 2.282,
+      "flow/cos_sim": 0.8703946246267128,
+      "flow/improvement_ratio": 0.9945963007300648,
+      "flow/mag_ratio_mean": 0.8639802461239829,
+      "flow/mag_ratio_std": 0.10180102466647305,
+      "step": 6144
+    },
+    {
+      "epoch": 0.30891891309500724,
+      "grad_norm": 2.7849202156066895,
+      "learning_rate": 8.22483558761947e-05,
+      "loss": 3.8042945861816406,
+      "step": 7168
+    },
+    {
+      "epoch": 0.30891891309500724,
+      "eval_bleu": 0.4923328702286439,
+      "eval_ce_clean_loss": 0.025836026905648617,
+      "eval_ce_pred_loss": 2.201877651946631,
+      "eval_flow_cos_loss": 0.11639371145762868,
+      "eval_flow_mse_loss": 1.4438700551417336,
+      "eval_loss": 3.700682167813722,
+      "flow/cos_sim": 0.8836063147862074,
+      "flow/improvement_ratio": 0.9956886801384152,
+      "flow/mag_ratio_mean": 0.8825320526480929,
+      "flow/mag_ratio_std": 0.09993369791553473,
+      "step": 7168
+    },
+    {
+      "epoch": 0.30891891309500724,
+      "eval_bleu": 0.4923328702286439,
+      "eval_ce_clean_loss": 0.025836026905648617,
+      "eval_ce_pred_loss": 2.201877651946631,
+      "eval_flow_cos_loss": 0.11639371145762868,
+      "eval_flow_mse_loss": 1.4438700551417336,
+      "eval_loss": 3.700682167813722,
+      "eval_runtime": 204.1105,
+      "eval_samples_per_second": 146.979,
+      "eval_steps_per_second": 2.298,
+      "flow/cos_sim": 0.8836063147862074,
+      "flow/improvement_ratio": 0.9956886801384152,
+      "flow/mag_ratio_mean": 0.8825320526480929,
+      "flow/mag_ratio_std": 0.09993369791553473,
+      "step": 7168
+    },
+    {
+      "epoch": 0.353050186394294,
+      "grad_norm": 3.4521219730377197,
+      "learning_rate": 7.639311770076283e-05,
+      "loss": 3.742882013320923,
+      "step": 8192
+    },
+    {
+      "epoch": 0.353050186394294,
+      "eval_bleu": 0.4985979565576243,
+      "eval_ce_clean_loss": 0.019649004948927143,
+      "eval_ce_pred_loss": 2.1497555045939203,
+      "eval_flow_cos_loss": 0.10761768184999414,
+      "eval_flow_mse_loss": 1.4486083501437579,
+      "eval_loss": 3.6449172842477178,
+      "flow/cos_sim": 0.8923823411530777,
+      "flow/improvement_ratio": 0.994830364865789,
+      "flow/mag_ratio_mean": 0.887576200560466,
+      "flow/mag_ratio_std": 0.09879913681478643,
+      "step": 8192
+    },
+    {
+      "epoch": 0.353050186394294,
+      "eval_bleu": 0.4985979565576243,
+      "eval_ce_clean_loss": 0.019649004948927143,
+      "eval_ce_pred_loss": 2.1497555045939203,
+      "eval_flow_cos_loss": 0.10761768184999414,
+      "eval_flow_mse_loss": 1.4486083501437579,
+      "eval_loss": 3.6449172842477178,
+      "eval_runtime": 202.46,
+      "eval_samples_per_second": 148.177,
+      "eval_steps_per_second": 2.317,
+      "flow/cos_sim": 0.8923823411530777,
+      "flow/improvement_ratio": 0.994830364865789,
+      "flow/mag_ratio_mean": 0.887576200560466,
+      "flow/mag_ratio_std": 0.09879913681478643,
+      "step": 8192
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "grad_norm": 3.2107231616973877,
+      "learning_rate": 6.998470950469718e-05,
+      "loss": 3.6923625469207764,
+      "step": 9216
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "eval_bleu": 0.5066862633379221,
+      "eval_ce_clean_loss": 0.015679127411611045,
+      "eval_ce_pred_loss": 2.0771032885701928,
+      "eval_flow_cos_loss": 0.09987032024273232,
+      "eval_flow_mse_loss": 1.4281837706372682,
+      "eval_loss": 3.5459337570011487,
+      "flow/cos_sim": 0.9001296995036892,
+      "flow/improvement_ratio": 0.9953078100167866,
+      "flow/mag_ratio_mean": 0.8994739131886822,
+      "flow/mag_ratio_std": 0.09903026111662261,
+      "step": 9216
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "eval_bleu": 0.5066862633379221,
+      "eval_ce_clean_loss": 0.015679127411611045,
+      "eval_ce_pred_loss": 2.0771032885701928,
+      "eval_flow_cos_loss": 0.09987032024273232,
+      "eval_flow_mse_loss": 1.4281837706372682,
+      "eval_loss": 3.5459337570011487,
+      "eval_runtime": 202.3533,
+      "eval_samples_per_second": 148.256,
+      "eval_steps_per_second": 2.318,
+      "flow/cos_sim": 0.9001296995036892,
+      "flow/improvement_ratio": 0.9953078100167866,
+      "flow/mag_ratio_mean": 0.8994739131886822,
+      "flow/mag_ratio_std": 0.09903026111662261,
+      "step": 9216
+    },
+    {
+      "epoch": 0.44131273299286744,
+      "grad_norm": 2.0100369453430176,
+      "learning_rate": 6.315061173955019e-05,
+      "loss": 3.6312835216522217,
+      "step": 10240
+    },
+    {
+      "epoch": 0.44131273299286744,
+      "eval_bleu": 0.5093028892911015,
+      "eval_ce_clean_loss": 0.01291672647225736,
+      "eval_ce_pred_loss": 2.0760930476666513,
+      "eval_flow_cos_loss": 0.09497748156473326,
+      "eval_flow_mse_loss": 1.4274813465472223,
+      "eval_loss": 3.54023549094129,
+      "flow/cos_sim": 0.9050225401992229,
+      "flow/improvement_ratio": 0.9958795127329796,
+      "flow/mag_ratio_mean": 0.9021421949238156,
+      "flow/mag_ratio_std": 0.09747769917125132,
+      "step": 10240
+    },
+    {
+      "epoch": 0.44131273299286744,
+      "eval_bleu": 0.5093028892911015,
+      "eval_ce_clean_loss": 0.01291672647225736,
+      "eval_ce_pred_loss": 2.0760930476666513,
+      "eval_flow_cos_loss": 0.09497748156473326,
+      "eval_flow_mse_loss": 1.4274813465472223,
+      "eval_loss": 3.54023549094129,
+      "eval_runtime": 201.9329,
+      "eval_samples_per_second": 148.564,
+      "eval_steps_per_second": 2.323,
+      "flow/cos_sim": 0.9050225401992229,
+      "flow/improvement_ratio": 0.9958795127329796,
+      "flow/mag_ratio_mean": 0.9021421949238156,
+      "flow/mag_ratio_std": 0.09747769917125132,
+      "step": 10240
+    },
+    {
+      "epoch": 0.4854440062921542,
+      "grad_norm": 1.2468712329864502,
+      "learning_rate": 5.604738390528452e-05,
+      "loss": 3.59726881980896,
+      "step": 11264
+    },
+    {
+      "epoch": 0.4854440062921542,
+      "eval_bleu": 0.5041383836806987,
+      "eval_ce_clean_loss": 0.010839646158894814,
+      "eval_ce_pred_loss": 2.0983083942043246,
+      "eval_flow_cos_loss": 0.08914897600407286,
+      "eval_flow_mse_loss": 1.3907782823038,
+      "eval_loss": 3.5222135632277043,
+      "flow/cos_sim": 0.9108510378327197,
+      "flow/improvement_ratio": 0.9943067862280904,
+      "flow/mag_ratio_mean": 0.9097654617417341,
+      "flow/mag_ratio_std": 0.09677679185419957,
+      "step": 11264
+    },
+    {
+      "epoch": 0.4854440062921542,
+      "eval_bleu": 0.5041383836806987,
+      "eval_ce_clean_loss": 0.010839646158894814,
+      "eval_ce_pred_loss": 2.0983083942043246,
+      "eval_flow_cos_loss": 0.08914897600407286,
+      "eval_flow_mse_loss": 1.3907782823038,
+      "eval_loss": 3.5222135632277043,
+      "eval_runtime": 203.9153,
+      "eval_samples_per_second": 147.12,
+      "eval_steps_per_second": 2.3,
+      "flow/cos_sim": 0.9108510378327197,
+      "flow/improvement_ratio": 0.9943067862280904,
+      "flow/mag_ratio_mean": 0.9097654617417341,
+      "flow/mag_ratio_std": 0.09677679185419957,
+      "step": 11264
+    }
+  ],
+  "logging_steps": 1024,
+  "max_steps": 23204,
+  "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-v5.7/checkpoint-11264/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8469bbc81a2ba0be2f5b44007faafd15c75615abe30f4f4e56171816d31caa5b
+size 5137