Upload folder using huggingface_hub

.gitattributes

@@ -51,3 +51,4 @@ checkpoints-v4.5/checkpoint-15360/eval_state.json filter=lfs diff=lfs merge=lfs
 checkpoints-v5.9/checkpoint-11264/eval_state.json filter=lfs diff=lfs merge=lfs -text
 checkpoints-v5.10/checkpoint-7168/eval_state.json filter=lfs diff=lfs merge=lfs -text
 checkpoints-v4.5/checkpoint-21504/eval_state.json filter=lfs diff=lfs merge=lfs -text
+checkpoints-v5.11/checkpoint-10240/eval_state.json filter=lfs diff=lfs merge=lfs -text

checkpoints-v5.11/checkpoint-10240/ema.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41499b4d7bb1556af3feff86eda28b00e8d3dba3def7737434f6b80e5ae78f4c
+size 55150648

checkpoints-v5.11/checkpoint-10240/eval_state.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a083769287760925a8dd73ab66868102703115faee1d1657ed250c789373753c
+size 59162021

checkpoints-v5.11/checkpoint-10240/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4038bd5206f65c2615de3198601fbacb5b23e99ca6b2ffc48a43fb2014a82001
+size 55150680

checkpoints-v5.11/checkpoint-10240/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f78465715bb28134a98872ad456f604761b722e3a8dc69cb430807fdb4af763
+size 77724619

checkpoints-v5.11/checkpoint-10240/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:395a99c9379c14b57848072fe8b6246133ccdfbc9ea5f8c57ebc64605d710240
+size 14645

checkpoints-v5.11/checkpoint-10240/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52b7da564e6046cbf689eeaeb9be88fd2c9342eb25ebe600069b5454c6acfd21
+size 1383

checkpoints-v5.11/checkpoint-10240/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:012f7b23a26c3659b3d642258b23d2a9db5c112a8101c3ff10a13ac93f47a91e
+size 1465

checkpoints-v5.11/checkpoint-10240/trainer_state.json

@@ -0,0 +1,414 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.44131273299286744,
+  "eval_steps": 1024,
+  "global_step": 10240,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.04413127329928675,
+      "grad_norm": 2.9264373779296875,
+      "learning_rate": 9.990234375e-05,
+      "loss": 12.596921920776367,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04413127329928675,
+      "eval_bleu": 0.13005541908821117,
+      "eval_ce_clean_loss": 2.286205404602897,
+      "eval_ce_pred_loss": 5.3831145524470285,
+      "eval_flow_consistency_loss": 0.13552273369864867,
+      "eval_flow_mse_loss": 0.9414198791294464,
+      "eval_loss": 7.06356679757775,
+      "flow/cos_sim": 0.35049253765708094,
+      "flow/improvement_ratio": 0.9494910390138118,
+      "flow/mag_ratio_mean": 0.2741034229172827,
+      "flow/mag_ratio_std": 0.0854220198892327,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04413127329928675,
+      "eval_bleu": 0.13005541908821117,
+      "eval_ce_clean_loss": 2.286205404602897,
+      "eval_ce_pred_loss": 5.3831145524470285,
+      "eval_flow_consistency_loss": 0.13552273369864867,
+      "eval_flow_mse_loss": 0.9414198791294464,
+      "eval_loss": 7.06356679757775,
+      "eval_runtime": 212.0249,
+      "eval_samples_per_second": 141.493,
+      "eval_steps_per_second": 2.212,
+      "flow/cos_sim": 0.35049253765708094,
+      "flow/improvement_ratio": 0.9494910390138118,
+      "flow/mag_ratio_mean": 0.2741034229172827,
+      "flow/mag_ratio_std": 0.0854220198892327,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0882625465985735,
+      "grad_norm": 1.9657727479934692,
+      "learning_rate": 9.9476028157316e-05,
+      "loss": 5.091745853424072,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0882625465985735,
+      "eval_bleu": 0.32513472477417255,
+      "eval_ce_clean_loss": 0.36479977980605577,
+      "eval_ce_pred_loss": 3.6980401625765413,
+      "eval_flow_consistency_loss": 0.1417246216427543,
+      "eval_flow_mse_loss": 1.0055452192515961,
+      "eval_loss": 4.0298353888586895,
+      "flow/cos_sim": 0.4995506982813512,
+      "flow/improvement_ratio": 0.9729367145088944,
+      "flow/mag_ratio_mean": 0.4342197393303487,
+      "flow/mag_ratio_std": 0.1229526477772544,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0882625465985735,
+      "eval_bleu": 0.32513472477417255,
+      "eval_ce_clean_loss": 0.36479977980605577,
+      "eval_ce_pred_loss": 3.6980401625765413,
+      "eval_flow_consistency_loss": 0.1417246216427543,
+      "eval_flow_mse_loss": 1.0055452192515961,
+      "eval_loss": 4.0298353888586895,
+      "eval_runtime": 209.9563,
+      "eval_samples_per_second": 142.887,
+      "eval_steps_per_second": 2.234,
+      "flow/cos_sim": 0.4995506982813512,
+      "flow/improvement_ratio": 0.9729367145088944,
+      "flow/mag_ratio_mean": 0.4342197393303487,
+      "flow/mag_ratio_std": 0.1229526477772544,
+      "step": 2048
+    },
+    {
+      "epoch": 0.13239381989786023,
+      "grad_norm": 1.1680883169174194,
+      "learning_rate": 9.791307026072513e-05,
+      "loss": 3.746001720428467,
+      "step": 3072
+    },
+    {
+      "epoch": 0.13239381989786023,
+      "eval_bleu": 0.3841303115629711,
+      "eval_ce_clean_loss": 0.13881633935897336,
+      "eval_ce_pred_loss": 3.082805846037387,
+      "eval_flow_consistency_loss": 0.1823698268262054,
+      "eval_flow_mse_loss": 1.0654335448991006,
+      "eval_loss": 3.45339886466069,
+      "flow/cos_sim": 0.5774906538188584,
+      "flow/improvement_ratio": 0.9910711745209277,
+      "flow/mag_ratio_mean": 0.5197838679559703,
+      "flow/mag_ratio_std": 0.101421576716117,
+      "step": 3072
+    },
+    {
+      "epoch": 0.13239381989786023,
+      "eval_bleu": 0.3841303115629711,
+      "eval_ce_clean_loss": 0.13881633935897336,
+      "eval_ce_pred_loss": 3.082805846037387,
+      "eval_flow_consistency_loss": 0.1823698268262054,
+      "eval_flow_mse_loss": 1.0654335448991006,
+      "eval_loss": 3.45339886466069,
+      "eval_runtime": 212.9337,
+      "eval_samples_per_second": 140.889,
+      "eval_steps_per_second": 2.203,
+      "flow/cos_sim": 0.5774906538188584,
+      "flow/improvement_ratio": 0.9910711745209277,
+      "flow/mag_ratio_mean": 0.5197838679559703,
+      "flow/mag_ratio_std": 0.101421576716117,
+      "step": 3072
+    },
+    {
+      "epoch": 0.176525093197147,
+      "grad_norm": 0.6403291821479797,
+      "learning_rate": 9.53439476074686e-05,
+      "loss": 3.401794672012329,
+      "step": 4096
+    },
+    {
+      "epoch": 0.176525093197147,
+      "eval_bleu": 0.41164963024752277,
+      "eval_ce_clean_loss": 0.075089068999931,
+      "eval_ce_pred_loss": 2.8144510768369826,
+      "eval_flow_consistency_loss": 0.20391849938359088,
+      "eval_flow_mse_loss": 1.0832029329434132,
+      "eval_loss": 3.230366972717903,
+      "flow/cos_sim": 0.6372234959846367,
+      "flow/improvement_ratio": 0.9945152276106226,
+      "flow/mag_ratio_mean": 0.590258006602208,
+      "flow/mag_ratio_std": 0.08975283501308355,
+      "step": 4096
+    },
+    {
+      "epoch": 0.176525093197147,
+      "eval_bleu": 0.41164963024752277,
+      "eval_ce_clean_loss": 0.075089068999931,
+      "eval_ce_pred_loss": 2.8144510768369826,
+      "eval_flow_consistency_loss": 0.20391849938359088,
+      "eval_flow_mse_loss": 1.0832029329434132,
+      "eval_loss": 3.230366972717903,
+      "eval_runtime": 212.6089,
+      "eval_samples_per_second": 141.104,
+      "eval_steps_per_second": 2.206,
+      "flow/cos_sim": 0.6372234959846367,
+      "flow/improvement_ratio": 0.9945152276106226,
+      "flow/mag_ratio_mean": 0.590258006602208,
+      "flow/mag_ratio_std": 0.08975283501308355,
+      "step": 4096
+    },
+    {
+      "epoch": 0.22065636649643372,
+      "grad_norm": 0.6607010364532471,
+      "learning_rate": 9.18264920723673e-05,
+      "loss": 3.239504337310791,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22065636649643372,
+      "eval_bleu": 0.4271325450649275,
+      "eval_ce_clean_loss": 0.04691080389611884,
+      "eval_ce_pred_loss": 2.6821144310904463,
+      "eval_flow_consistency_loss": 0.19756401179314675,
+      "eval_flow_mse_loss": 1.05885077488702,
+      "eval_loss": 3.0820236617822383,
+      "flow/cos_sim": 0.6784213738146622,
+      "flow/improvement_ratio": 0.9945986220069023,
+      "flow/mag_ratio_mean": 0.6420265626805678,
+      "flow/mag_ratio_std": 0.09060588493339543,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22065636649643372,
+      "eval_bleu": 0.4271325450649275,
+      "eval_ce_clean_loss": 0.04691080389611884,
+      "eval_ce_pred_loss": 2.6821144310904463,
+      "eval_flow_consistency_loss": 0.19756401179314675,
+      "eval_flow_mse_loss": 1.05885077488702,
+      "eval_loss": 3.0820236617822383,
+      "eval_runtime": 211.4363,
+      "eval_samples_per_second": 141.887,
+      "eval_steps_per_second": 2.218,
+      "flow/cos_sim": 0.6784213738146622,
+      "flow/improvement_ratio": 0.9945986220069023,
+      "flow/mag_ratio_mean": 0.6420265626805678,
+      "flow/mag_ratio_std": 0.09060588493339543,
+      "step": 5120
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "grad_norm": 1.069400429725647,
+      "learning_rate": 8.742770483354739e-05,
+      "loss": 3.1248061656951904,
+      "step": 6144
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "eval_bleu": 0.4376915429907597,
+      "eval_ce_clean_loss": 0.03250600234754304,
+      "eval_ce_pred_loss": 2.6012275836615166,
+      "eval_flow_consistency_loss": 0.18590141693030848,
+      "eval_flow_mse_loss": 1.0617919451138105,
+      "eval_loss": 3.0081079306124625,
+      "flow/cos_sim": 0.7023057956685389,
+      "flow/improvement_ratio": 0.9948139922705286,
+      "flow/mag_ratio_mean": 0.6713837680023617,
+      "flow/mag_ratio_std": 0.10080839297982422,
+      "step": 6144
+    },
+    {
+      "epoch": 0.26478763979572045,
+      "eval_bleu": 0.4376915429907597,
+      "eval_ce_clean_loss": 0.03250600234754304,
+      "eval_ce_pred_loss": 2.6012275836615166,
+      "eval_flow_consistency_loss": 0.18590141693030848,
+      "eval_flow_mse_loss": 1.0617919451138105,
+      "eval_loss": 3.0081079306124625,
+      "eval_runtime": 213.8497,
+      "eval_samples_per_second": 140.285,
+      "eval_steps_per_second": 2.193,
+      "flow/cos_sim": 0.7023057956685389,
+      "flow/improvement_ratio": 0.9948139922705286,
+      "flow/mag_ratio_mean": 0.6713837680023617,
+      "flow/mag_ratio_std": 0.10080839297982422,
+      "step": 6144
+    },
+    {
+      "epoch": 0.30891891309500724,
+      "grad_norm": 1.219580054283142,
+      "learning_rate": 8.22483558761947e-05,
+      "loss": 3.0330727100372314,
+      "step": 7168
+    },
+    {
+      "epoch": 0.30891891309500724,
+      "eval_bleu": 0.4551949901017014,
+      "eval_ce_clean_loss": 0.02424235607205487,
+      "eval_ce_pred_loss": 2.500773771993641,
+      "eval_flow_consistency_loss": 0.17524592572056663,
+      "eval_flow_mse_loss": 1.0443366509256586,
+      "eval_loss": 2.9067435889864273,
+      "flow/cos_sim": 0.7204479055363995,
+      "flow/improvement_ratio": 0.9950863978247653,
+      "flow/mag_ratio_mean": 0.6897140079215646,
+      "flow/mag_ratio_std": 0.10839694198260684,
+      "step": 7168
+    },
+    {
+      "epoch": 0.30891891309500724,
+      "eval_bleu": 0.4551949901017014,
+      "eval_ce_clean_loss": 0.02424235607205487,
+      "eval_ce_pred_loss": 2.500773771993641,
+      "eval_flow_consistency_loss": 0.17524592572056663,
+      "eval_flow_mse_loss": 1.0443366509256586,
+      "eval_loss": 2.9067435889864273,
+      "eval_runtime": 215.0284,
+      "eval_samples_per_second": 139.516,
+      "eval_steps_per_second": 2.181,
+      "flow/cos_sim": 0.7204479055363995,
+      "flow/improvement_ratio": 0.9950863978247653,
+      "flow/mag_ratio_mean": 0.6897140079215646,
+      "flow/mag_ratio_std": 0.10839694198260684,
+      "step": 7168
+    },
+    {
+      "epoch": 0.353050186394294,
+      "grad_norm": 2.246631145477295,
+      "learning_rate": 7.638710244802891e-05,
+      "loss": 2.95896315574646,
+      "step": 8192
+    },
+    {
+      "epoch": 0.353050186394294,
+      "eval_bleu": 0.46141542061426044,
+      "eval_ce_clean_loss": 0.01877150087674925,
+      "eval_ce_pred_loss": 2.4870052022466274,
+      "eval_flow_consistency_loss": 0.16635312886634615,
+      "eval_flow_mse_loss": 1.0426371824512604,
+      "eval_loss": 2.8854888682680597,
+      "flow/cos_sim": 0.7327545807559861,
+      "flow/improvement_ratio": 0.9935595951100656,
+      "flow/mag_ratio_mean": 0.7086031040403126,
+      "flow/mag_ratio_std": 0.11536222337278476,
+      "step": 8192
+    },
+    {
+      "epoch": 0.353050186394294,
+      "eval_bleu": 0.46141542061426044,
+      "eval_ce_clean_loss": 0.01877150087674925,
+      "eval_ce_pred_loss": 2.4870052022466274,
+      "eval_flow_consistency_loss": 0.16635312886634615,
+      "eval_flow_mse_loss": 1.0426371824512604,
+      "eval_loss": 2.8854888682680597,
+      "eval_runtime": 212.0987,
+      "eval_samples_per_second": 141.444,
+      "eval_steps_per_second": 2.211,
+      "flow/cos_sim": 0.7327545807559861,
+      "flow/improvement_ratio": 0.9935595951100656,
+      "flow/mag_ratio_mean": 0.7086031040403126,
+      "flow/mag_ratio_std": 0.11536222337278476,
+      "step": 8192
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "grad_norm": 1.2935556173324585,
+      "learning_rate": 6.997821756319211e-05,
+      "loss": 2.9052517414093018,
+      "step": 9216
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "eval_bleu": 0.46944041426382027,
+      "eval_ce_clean_loss": 0.014995425907787737,
+      "eval_ce_pred_loss": 2.4030428840153255,
+      "eval_flow_consistency_loss": 0.1594380565098862,
+      "eval_flow_mse_loss": 1.0466467982162035,
+      "eval_loss": 2.8234912571368187,
+      "flow/cos_sim": 0.744667167983838,
+      "flow/improvement_ratio": 0.993155452361239,
+      "flow/mag_ratio_mean": 0.7179579849182162,
+      "flow/mag_ratio_std": 0.11927568821955337,
+      "step": 9216
+    },
+    {
+      "epoch": 0.3971814596935807,
+      "eval_bleu": 0.46944041426382027,
+      "eval_ce_clean_loss": 0.014995425907787737,
+      "eval_ce_pred_loss": 2.4030428840153255,
+      "eval_flow_consistency_loss": 0.1594380565098862,
+      "eval_flow_mse_loss": 1.0466467982162035,
+      "eval_loss": 2.8234912571368187,
+      "eval_runtime": 210.1213,
+      "eval_samples_per_second": 142.775,
+      "eval_steps_per_second": 2.232,
+      "flow/cos_sim": 0.744667167983838,
+      "flow/improvement_ratio": 0.993155452361239,
+      "flow/mag_ratio_mean": 0.7179579849182162,
+      "flow/mag_ratio_std": 0.11927568821955337,
+      "step": 9216
+    },
+    {
+      "epoch": 0.44131273299286744,
+      "grad_norm": 0.7428159117698669,
+      "learning_rate": 6.314377890922702e-05,
+      "loss": 2.8689093589782715,
+      "step": 10240
+    },
+    {
+      "epoch": 0.44131273299286744,
+      "eval_bleu": 0.4685034093487891,
+      "eval_ce_clean_loss": 0.01237281793450464,
+      "eval_ce_pred_loss": 2.384004484361677,
+      "eval_flow_consistency_loss": 0.1589052606302538,
+      "eval_flow_mse_loss": 1.0222321362383584,
+      "eval_loss": 2.7828606944094334,
+      "flow/cos_sim": 0.7520584338255274,
+      "flow/improvement_ratio": 0.9938689404204964,
+      "flow/mag_ratio_mean": 0.7304074718499742,
+      "flow/mag_ratio_std": 0.12232410612263914,
+      "step": 10240
+    },
+    {
+      "epoch": 0.44131273299286744,
+      "eval_bleu": 0.4685034093487891,
+      "eval_ce_clean_loss": 0.01237281793450464,
+      "eval_ce_pred_loss": 2.384004484361677,
+      "eval_flow_consistency_loss": 0.1589052606302538,
+      "eval_flow_mse_loss": 1.0222321362383584,
+      "eval_loss": 2.7828606944094334,
+      "eval_runtime": 211.3456,
+      "eval_samples_per_second": 141.948,
+      "eval_steps_per_second": 2.219,
+      "flow/cos_sim": 0.7520584338255274,
+      "flow/improvement_ratio": 0.9938689404204964,
+      "flow/mag_ratio_mean": 0.7304074718499742,
+      "flow/mag_ratio_std": 0.12232410612263914,
+      "step": 10240
+    }
+  ],
+  "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-v5.11/checkpoint-10240/training_args.bin

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