Upload folder using huggingface_hub

.gitattributes

@@ -38,3 +38,4 @@ checkpoints-v5.1/checkpoint-12288/eval_state.json filter=lfs diff=lfs merge=lfs
 checkpoints-v5.2/checkpoint-11264/eval_state.json filter=lfs diff=lfs merge=lfs -text
 checkpoints-v5.3/checkpoint-11264/eval_state.json filter=lfs diff=lfs merge=lfs -text
 checkpoints-v5.4/checkpoint-12288/eval_state.json filter=lfs diff=lfs merge=lfs -text
+checkpoints-v5.5/checkpoint-10240/eval_state.json filter=lfs diff=lfs merge=lfs -text

checkpoints-v5.5/checkpoint-10240/ema.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53c32f304e06a664ced6b5a01dc36f41c6efe354be2b757f556465a1c6d2dc5c
+size 54599592

checkpoints-v5.5/checkpoint-10240/eval_state.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9dbe0c785ada6ae8ecb6cf96b30f8ee28b18ca8039146e1d8734df7733554bbf
+size 19984661

checkpoints-v5.5/checkpoint-10240/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67bfcd99265e0a7f198a1da6183a2f5095b406251bd7dbc61a0ea3f3322685a9
+size 54599624

checkpoints-v5.5/checkpoint-10240/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cde09398522a16da97b8e3489802a486d243f6a2fd19b6085e0df696ca97cd63
+size 76551435

checkpoints-v5.5/checkpoint-10240/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:056ea97ad2f35feda7dc1d25414ba26f8d40ede84bb42878aa5f2ad51b1b0166
+size 14645

checkpoints-v5.5/checkpoint-10240/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1613ca5e69882f12c8eb8015e0d6aa8d2c1789f21a7954b955d73a0d7994fc88
+size 1383

checkpoints-v5.5/checkpoint-10240/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20959e6dc06f08fff3842ea99d603e06a6a51caeb422e8680bff7397ce65afef
+size 1465

checkpoints-v5.5/checkpoint-10240/trainer_state.json

@@ -0,0 +1,414 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.3238735535587303,
+  "eval_steps": 1024,
+  "global_step": 10240,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.13239381989786023,
+      "grad_norm": 7.3368072509765625,
+      "learning_rate": 9.990234375e-05,
+      "loss": 13.833250999450684,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13239381989786023,
+      "eval_bleu": 0.08843010093383347,
+      "eval_ce_clean_loss": 3.785046512154257,
+      "eval_ce_pred_loss": 6.1898151750017885,
+      "eval_flow_cos_loss": 0.39729490059955863,
+      "eval_flow_mse_loss": 1.2504585077808161,
+      "eval_loss": 9.467699330323821,
+      "flow/cos_sim": 0.6027051043358578,
+      "flow/improvement_ratio": 0.9944390654563904,
+      "flow/mag_ratio_mean": 0.6029561613775363,
+      "flow/mag_ratio_std": 0.06967356720357944,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13239381989786023,
+      "eval_bleu": 0.08843010093383347,
+      "eval_ce_clean_loss": 3.785046512154257,
+      "eval_ce_pred_loss": 6.1898151750017885,
+      "eval_flow_cos_loss": 0.39729490059955863,
+      "eval_flow_mse_loss": 1.2504585077808161,
+      "eval_loss": 9.467699330323821,
+      "eval_runtime": 69.1426,
+      "eval_samples_per_second": 144.629,
+      "eval_steps_per_second": 2.271,
+      "flow/cos_sim": 0.6027051043358578,
+      "flow/improvement_ratio": 0.9944390654563904,
+      "flow/mag_ratio_mean": 0.6029561613775363,
+      "flow/mag_ratio_std": 0.06967356720357944,
+      "step": 1024
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "grad_norm": 1.5976208448410034,
+      "learning_rate": 9.971175203561169e-05,
+      "loss": 6.556396484375,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "eval_bleu": 0.2647666813502405,
+      "eval_ce_clean_loss": 0.6921462336922907,
+      "eval_ce_pred_loss": 4.159270631279915,
+      "eval_flow_cos_loss": 0.33016856043202103,
+      "eval_flow_mse_loss": 1.083006814786583,
+      "eval_loss": 4.769184565088551,
+      "flow/cos_sim": 0.6698314437441005,
+      "flow/improvement_ratio": 0.9946717445258122,
+      "flow/mag_ratio_mean": 0.6456947994839614,
+      "flow/mag_ratio_std": 0.08488734332239552,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "eval_bleu": 0.2647666813502405,
+      "eval_ce_clean_loss": 0.6921462336922907,
+      "eval_ce_pred_loss": 4.159270631279915,
+      "eval_flow_cos_loss": 0.33016856043202103,
+      "eval_flow_mse_loss": 1.083006814786583,
+      "eval_loss": 4.769184565088551,
+      "eval_runtime": 67.8586,
+      "eval_samples_per_second": 147.365,
+      "eval_steps_per_second": 2.314,
+      "flow/cos_sim": 0.6698314437441005,
+      "flow/improvement_ratio": 0.9946717445258122,
+      "flow/mag_ratio_mean": 0.6456947994839614,
+      "flow/mag_ratio_std": 0.08488734332239552,
+      "step": 2048
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "grad_norm": 1.0349175930023193,
+      "learning_rate": 9.885033161800567e-05,
+      "loss": 4.22868013381958,
+      "step": 3072
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "eval_bleu": 0.3307477170153146,
+      "eval_ce_clean_loss": 0.21776276523140586,
+      "eval_ce_pred_loss": 3.5308286536271405,
+      "eval_flow_cos_loss": 0.3201853291244264,
+      "eval_flow_mse_loss": 1.0728299655732076,
+      "eval_loss": 3.8422190869689747,
+      "flow/cos_sim": 0.6798147354156349,
+      "flow/improvement_ratio": 0.9928556031482235,
+      "flow/mag_ratio_mean": 0.6495526474752243,
+      "flow/mag_ratio_std": 0.08990857878308388,
+      "step": 3072
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "eval_bleu": 0.3307477170153146,
+      "eval_ce_clean_loss": 0.21776276523140586,
+      "eval_ce_pred_loss": 3.5308286536271405,
+      "eval_flow_cos_loss": 0.3201853291244264,
+      "eval_flow_mse_loss": 1.0728299655732076,
+      "eval_loss": 3.8422190869689747,
+      "eval_runtime": 67.9151,
+      "eval_samples_per_second": 147.243,
+      "eval_steps_per_second": 2.312,
+      "flow/cos_sim": 0.6798147354156349,
+      "flow/improvement_ratio": 0.9928556031482235,
+      "flow/mag_ratio_mean": 0.6495526474752243,
+      "flow/mag_ratio_std": 0.08990857878308388,
+      "step": 3072
+    },
+    {
+      "epoch": 0.5295752795914409,
+      "grad_norm": 2.284677028656006,
+      "learning_rate": 9.742400750550229e-05,
+      "loss": 3.7094979286193848,
+      "step": 4096
+    },
+    {
+      "epoch": 0.5295752795914409,
+      "eval_bleu": 0.3527590985954599,
+      "eval_ce_clean_loss": 0.10325595935818496,
+      "eval_ce_pred_loss": 3.3037764221240002,
+      "eval_flow_cos_loss": 0.29567099870390195,
+      "eval_flow_mse_loss": 1.0334580939286833,
+      "eval_loss": 3.5232752614719853,
+      "flow/cos_sim": 0.7043290168616423,
+      "flow/improvement_ratio": 0.9942314150227103,
+      "flow/mag_ratio_mean": 0.6710645951283206,
+      "flow/mag_ratio_std": 0.08415729860970929,
+      "step": 4096
+    },
+    {
+      "epoch": 0.5295752795914409,
+      "eval_bleu": 0.3527590985954599,
+      "eval_ce_clean_loss": 0.10325595935818496,
+      "eval_ce_pred_loss": 3.3037764221240002,
+      "eval_flow_cos_loss": 0.29567099870390195,
+      "eval_flow_mse_loss": 1.0334580939286833,
+      "eval_loss": 3.5232752614719853,
+      "eval_runtime": 67.527,
+      "eval_samples_per_second": 148.089,
+      "eval_steps_per_second": 2.325,
+      "flow/cos_sim": 0.7043290168616423,
+      "flow/improvement_ratio": 0.9942314150227103,
+      "flow/mag_ratio_mean": 0.6710645951283206,
+      "flow/mag_ratio_std": 0.08415729860970929,
+      "step": 4096
+    },
+    {
+      "epoch": 0.6619690994893012,
+      "grad_norm": 1.0625219345092773,
+      "learning_rate": 9.544981995345226e-05,
+      "loss": 3.4675893783569336,
+      "step": 5120
+    },
+    {
+      "epoch": 0.6619690994893012,
+      "eval_bleu": 0.37785369114831263,
+      "eval_ce_clean_loss": 0.0570975638404014,
+      "eval_ce_pred_loss": 3.1086114409622874,
+      "eval_flow_cos_loss": 0.2890773550340324,
+      "eval_flow_mse_loss": 1.0530509029983715,
+      "eval_loss": 3.3584457916818606,
+      "flow/cos_sim": 0.7109226474336757,
+      "flow/improvement_ratio": 0.9944721566643685,
+      "flow/mag_ratio_mean": 0.6796372863137798,
+      "flow/mag_ratio_std": 0.08220032687969268,
+      "step": 5120
+    },
+    {
+      "epoch": 0.6619690994893012,
+      "eval_bleu": 0.37785369114831263,
+      "eval_ce_clean_loss": 0.0570975638404014,
+      "eval_ce_pred_loss": 3.1086114409622874,
+      "eval_flow_cos_loss": 0.2890773550340324,
+      "eval_flow_mse_loss": 1.0530509029983715,
+      "eval_loss": 3.3584457916818606,
+      "eval_runtime": 67.5831,
+      "eval_samples_per_second": 147.966,
+      "eval_steps_per_second": 2.323,
+      "flow/cos_sim": 0.7109226474336757,
+      "flow/improvement_ratio": 0.9944721566643685,
+      "flow/mag_ratio_mean": 0.6796372863137798,
+      "flow/mag_ratio_std": 0.08220032687969268,
+      "step": 5120
+    },
+    {
+      "epoch": 0.7943629193871614,
+      "grad_norm": 1.390210747718811,
+      "learning_rate": 9.295057566334431e-05,
+      "loss": 3.3179638385772705,
+      "step": 6144
+    },
+    {
+      "epoch": 0.7943629193871614,
+      "eval_bleu": 0.40284924593911503,
+      "eval_ce_clean_loss": 0.03560329580999863,
+      "eval_ce_pred_loss": 2.935079864635589,
+      "eval_flow_cos_loss": 0.2695348363393431,
+      "eval_flow_mse_loss": 1.037700882383213,
+      "eval_loss": 3.195243762556914,
+      "flow/cos_sim": 0.7304651706841341,
+      "flow/improvement_ratio": 0.9945835288922498,
+      "flow/mag_ratio_mean": 0.7005888793119199,
+      "flow/mag_ratio_std": 0.0835665136480787,
+      "step": 6144
+    },
+    {
+      "epoch": 0.7943629193871614,
+      "eval_bleu": 0.40284924593911503,
+      "eval_ce_clean_loss": 0.03560329580999863,
+      "eval_ce_pred_loss": 2.935079864635589,
+      "eval_flow_cos_loss": 0.2695348363393431,
+      "eval_flow_mse_loss": 1.037700882383213,
+      "eval_loss": 3.195243762556914,
+      "eval_runtime": 67.6818,
+      "eval_samples_per_second": 147.75,
+      "eval_steps_per_second": 2.32,
+      "flow/cos_sim": 0.7304651706841341,
+      "flow/improvement_ratio": 0.9945835288922498,
+      "flow/mag_ratio_mean": 0.7005888793119199,
+      "flow/mag_ratio_std": 0.0835665136480787,
+      "step": 6144
+    },
+    {
+      "epoch": 0.9267567392850217,
+      "grad_norm": 1.7730882167816162,
+      "learning_rate": 8.995830349195804e-05,
+      "loss": 3.2147014141082764,
+      "step": 7168
+    },
+    {
+      "epoch": 0.9267567392850217,
+      "eval_bleu": 0.4108774093856199,
+      "eval_ce_clean_loss": 0.025004512588879105,
+      "eval_ce_pred_loss": 2.8478709542827243,
+      "eval_flow_cos_loss": 0.25457756676871307,
+      "eval_flow_mse_loss": 1.0318333833080948,
+      "eval_loss": 3.113991931745201,
+      "flow/cos_sim": 0.745422419469068,
+      "flow/improvement_ratio": 0.9949804590006542,
+      "flow/mag_ratio_mean": 0.7171509364607987,
+      "flow/mag_ratio_std": 0.08899391105600224,
+      "step": 7168
+    },
+    {
+      "epoch": 0.9267567392850217,
+      "eval_bleu": 0.4108774093856199,
+      "eval_ce_clean_loss": 0.025004512588879105,
+      "eval_ce_pred_loss": 2.8478709542827243,
+      "eval_flow_cos_loss": 0.25457756676871307,
+      "eval_flow_mse_loss": 1.0318333833080948,
+      "eval_loss": 3.113991931745201,
+      "eval_runtime": 67.7148,
+      "eval_samples_per_second": 147.678,
+      "eval_steps_per_second": 2.319,
+      "flow/cos_sim": 0.745422419469068,
+      "flow/improvement_ratio": 0.9949804590006542,
+      "flow/mag_ratio_mean": 0.7171509364607987,
+      "flow/mag_ratio_std": 0.08899391105600224,
+      "step": 7168
+    },
+    {
+      "epoch": 1.05908591376301,
+      "grad_norm": 2.208953619003296,
+      "learning_rate": 8.650172716103233e-05,
+      "loss": 3.119405508041382,
+      "step": 8192
+    },
+    {
+      "epoch": 1.05908591376301,
+      "eval_bleu": 0.4100243274079566,
+      "eval_ce_clean_loss": 0.018418982221632248,
+      "eval_ce_pred_loss": 2.801309669853016,
+      "eval_flow_cos_loss": 0.24036190160520518,
+      "eval_flow_mse_loss": 1.0228236701078475,
+      "eval_loss": 3.062249883724626,
+      "flow/cos_sim": 0.7596381280072935,
+      "flow/improvement_ratio": 0.9959618415042852,
+      "flow/mag_ratio_mean": 0.7364322915198697,
+      "flow/mag_ratio_std": 0.0917528191949152,
+      "step": 8192
+    },
+    {
+      "epoch": 1.05908591376301,
+      "eval_bleu": 0.4100243274079566,
+      "eval_ce_clean_loss": 0.018418982221632248,
+      "eval_ce_pred_loss": 2.801309669853016,
+      "eval_flow_cos_loss": 0.24036190160520518,
+      "eval_flow_mse_loss": 1.0228236701078475,
+      "eval_loss": 3.062249883724626,
+      "eval_runtime": 69.3159,
+      "eval_samples_per_second": 144.267,
+      "eval_steps_per_second": 2.265,
+      "flow/cos_sim": 0.7596381280072935,
+      "flow/improvement_ratio": 0.9959618415042852,
+      "flow/mag_ratio_mean": 0.7364322915198697,
+      "flow/mag_ratio_std": 0.0917528191949152,
+      "step": 8192
+    },
+    {
+      "epoch": 1.19147973366087,
+      "grad_norm": 1.0985896587371826,
+      "learning_rate": 8.263142386444264e-05,
+      "loss": 3.0675039291381836,
+      "step": 9216
+    },
+    {
+      "epoch": 1.19147973366087,
+      "eval_bleu": 0.42913809326340735,
+      "eval_ce_clean_loss": 0.014243369497904543,
+      "eval_ce_pred_loss": 2.679083149903899,
+      "eval_flow_cos_loss": 0.2306169035138598,
+      "eval_flow_mse_loss": 1.0239125793906534,
+      "eval_loss": 2.97116835861449,
+      "flow/cos_sim": 0.7693831127160674,
+      "flow/improvement_ratio": 0.9939722635184124,
+      "flow/mag_ratio_mean": 0.748531128190885,
+      "flow/mag_ratio_std": 0.0947970964821281,
+      "step": 9216
+    },
+    {
+      "epoch": 1.19147973366087,
+      "eval_bleu": 0.42913809326340735,
+      "eval_ce_clean_loss": 0.014243369497904543,
+      "eval_ce_pred_loss": 2.679083149903899,
+      "eval_flow_cos_loss": 0.2306169035138598,
+      "eval_flow_mse_loss": 1.0239125793906534,
+      "eval_loss": 2.97116835861449,
+      "eval_runtime": 67.1108,
+      "eval_samples_per_second": 149.007,
+      "eval_steps_per_second": 2.339,
+      "flow/cos_sim": 0.7693831127160674,
+      "flow/improvement_ratio": 0.9939722635184124,
+      "flow/mag_ratio_mean": 0.748531128190885,
+      "flow/mag_ratio_std": 0.0947970964821281,
+      "step": 9216
+    },
+    {
+      "epoch": 1.3238735535587303,
+      "grad_norm": 2.3166847229003906,
+      "learning_rate": 7.837697175482903e-05,
+      "loss": 3.002436399459839,
+      "step": 10240
+    },
+    {
+      "epoch": 1.3238735535587303,
+      "eval_bleu": 0.42768151912552244,
+      "eval_ce_clean_loss": 0.011440879840308883,
+      "eval_ce_pred_loss": 2.672361337455215,
+      "eval_flow_cos_loss": 0.2179829450739417,
+      "eval_flow_mse_loss": 1.0098259748926588,
+      "eval_loss": 2.946415506350766,
+      "flow/cos_sim": 0.782017103426016,
+      "flow/improvement_ratio": 0.9939965146362402,
+      "flow/mag_ratio_mean": 0.7625711093283003,
+      "flow/mag_ratio_std": 0.09684707447411908,
+      "step": 10240
+    },
+    {
+      "epoch": 1.3238735535587303,
+      "eval_bleu": 0.42768151912552244,
+      "eval_ce_clean_loss": 0.011440879840308883,
+      "eval_ce_pred_loss": 2.672361337455215,
+      "eval_flow_cos_loss": 0.2179829450739417,
+      "eval_flow_mse_loss": 1.0098259748926588,
+      "eval_loss": 2.946415506350766,
+      "eval_runtime": 67.7743,
+      "eval_samples_per_second": 147.549,
+      "eval_steps_per_second": 2.317,
+      "flow/cos_sim": 0.782017103426016,
+      "flow/improvement_ratio": 0.9939965146362402,
+      "flow/mag_ratio_mean": 0.7625711093283003,
+      "flow/mag_ratio_std": 0.09684707447411908,
+      "step": 10240
+    }
+  ],
+  "logging_steps": 1024,
+  "max_steps": 30940,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "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-v5.5/checkpoint-10240/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7be4d80b8499f3b5f618b042dcec062719328222caddac0d4e4ce11d371480d
+size 5137