Upload folder using huggingface_hub
Browse files- checkpoints-v2.6-d/checkpoint-19456/ema.safetensors +3 -0
- checkpoints-v2.6-d/checkpoint-19456/model.safetensors +3 -0
- checkpoints-v2.6-d/checkpoint-19456/optimizer.pt +3 -0
- checkpoints-v2.6-d/checkpoint-19456/rng_state.pth +3 -0
- checkpoints-v2.6-d/checkpoint-19456/scaler.pt +3 -0
- checkpoints-v2.6-d/checkpoint-19456/scheduler.pt +3 -0
- checkpoints-v2.6-d/checkpoint-19456/trainer_state.json +642 -0
- checkpoints-v2.6-d/checkpoint-19456/training_args.bin +3 -0
checkpoints-v2.6-d/checkpoint-19456/ema.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b3da1a4017ea1e76d94f8d8f1c33151acc6328f022fbdd23676ebfbc037d22d3
|
| 3 |
+
size 685416
|
checkpoints-v2.6-d/checkpoint-19456/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:07f7babf4ee2bfb0ee668e44b3dcfc6147dd40eb35bb3f09bf2b94e0372701f4
|
| 3 |
+
size 19466896
|
checkpoints-v2.6-d/checkpoint-19456/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d05291d04ce348f05bd13a6eba2805c32fa386cb7e866ede8d5851e0ea002098
|
| 3 |
+
size 1463627
|
checkpoints-v2.6-d/checkpoint-19456/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b99619c15c9452e4c40ddfa70e473c21d82ef5758607cce5410080cbc1ed8f16
|
| 3 |
+
size 14645
|
checkpoints-v2.6-d/checkpoint-19456/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b074bd8f525c816f8d0a345f3a310eb604d0320fe709722b506b3e2568883c3
|
| 3 |
+
size 1383
|
checkpoints-v2.6-d/checkpoint-19456/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cae8472ee801fc147116dcf4548b90fad1c7e2970f7f7944b2df576fd00f3a03
|
| 3 |
+
size 1465
|
checkpoints-v2.6-d/checkpoint-19456/trainer_state.json
ADDED
|
@@ -0,0 +1,642 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.8986190014318045,
|
| 6 |
+
"eval_steps": 1024,
|
| 7 |
+
"global_step": 19456,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.047295736917463395,
|
| 14 |
+
"grad_norm": 0.47292250394821167,
|
| 15 |
+
"learning_rate": 0.00016650390625,
|
| 16 |
+
"loss": 2.742399215698242,
|
| 17 |
+
"step": 1024
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.047295736917463395,
|
| 21 |
+
"eval_cos_loss": 0.5961006330028517,
|
| 22 |
+
"eval_loss": 2.449579068514854,
|
| 23 |
+
"eval_mse_loss": 1.8534784379615088,
|
| 24 |
+
"flow/cos_sim": 0.4038993830550207,
|
| 25 |
+
"flow/improvement_ratio": 0.9530568593713247,
|
| 26 |
+
"flow/mag_ratio_mean": 0.08299264077046146,
|
| 27 |
+
"flow/mag_ratio_std": 0.04246563146307588,
|
| 28 |
+
"step": 1024
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.047295736917463395,
|
| 32 |
+
"eval_cos_loss": 0.5961006330028517,
|
| 33 |
+
"eval_loss": 2.449579068514854,
|
| 34 |
+
"eval_mse_loss": 1.8534784379615088,
|
| 35 |
+
"eval_runtime": 39.604,
|
| 36 |
+
"eval_samples_per_second": 706.823,
|
| 37 |
+
"eval_steps_per_second": 11.059,
|
| 38 |
+
"flow/cos_sim": 0.4038993830550207,
|
| 39 |
+
"flow/improvement_ratio": 0.9530568593713247,
|
| 40 |
+
"flow/mag_ratio_mean": 0.08299264077046146,
|
| 41 |
+
"flow/mag_ratio_std": 0.04246563146307588,
|
| 42 |
+
"step": 1024
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.09459147383492679,
|
| 46 |
+
"grad_norm": 0.3767559826374054,
|
| 47 |
+
"learning_rate": 0.0003331705729166667,
|
| 48 |
+
"loss": 2.077958345413208,
|
| 49 |
+
"step": 2048
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.09459147383492679,
|
| 53 |
+
"eval_cos_loss": 0.4873075728138832,
|
| 54 |
+
"eval_loss": 1.9004751797680441,
|
| 55 |
+
"eval_mse_loss": 1.413167606477868,
|
| 56 |
+
"flow/cos_sim": 0.5126924207221427,
|
| 57 |
+
"flow/improvement_ratio": 0.96191185848898,
|
| 58 |
+
"flow/mag_ratio_mean": 0.5089414368753564,
|
| 59 |
+
"flow/mag_ratio_std": 0.17938318715928353,
|
| 60 |
+
"step": 2048
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.09459147383492679,
|
| 64 |
+
"eval_cos_loss": 0.4873075728138832,
|
| 65 |
+
"eval_loss": 1.9004751797680441,
|
| 66 |
+
"eval_mse_loss": 1.413167606477868,
|
| 67 |
+
"eval_runtime": 40.5135,
|
| 68 |
+
"eval_samples_per_second": 690.955,
|
| 69 |
+
"eval_steps_per_second": 10.811,
|
| 70 |
+
"flow/cos_sim": 0.5126924207221427,
|
| 71 |
+
"flow/improvement_ratio": 0.96191185848898,
|
| 72 |
+
"flow/mag_ratio_mean": 0.5089414368753564,
|
| 73 |
+
"flow/mag_ratio_std": 0.17938318715928353,
|
| 74 |
+
"step": 2048
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.1418872107523902,
|
| 78 |
+
"grad_norm": 0.32334470748901367,
|
| 79 |
+
"learning_rate": 0.0004998372395833334,
|
| 80 |
+
"loss": 1.8623729944229126,
|
| 81 |
+
"step": 3072
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.1418872107523902,
|
| 85 |
+
"eval_cos_loss": 0.45344613587747423,
|
| 86 |
+
"eval_loss": 1.7837548544417778,
|
| 87 |
+
"eval_mse_loss": 1.3303087190405962,
|
| 88 |
+
"flow/cos_sim": 0.5465538722195037,
|
| 89 |
+
"flow/improvement_ratio": 0.9649395484086041,
|
| 90 |
+
"flow/mag_ratio_mean": 0.545519895885633,
|
| 91 |
+
"flow/mag_ratio_std": 0.2078168946722327,
|
| 92 |
+
"step": 3072
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.1418872107523902,
|
| 96 |
+
"eval_cos_loss": 0.45344613587747423,
|
| 97 |
+
"eval_loss": 1.7837548544417778,
|
| 98 |
+
"eval_mse_loss": 1.3303087190405962,
|
| 99 |
+
"eval_runtime": 39.4807,
|
| 100 |
+
"eval_samples_per_second": 709.03,
|
| 101 |
+
"eval_steps_per_second": 11.094,
|
| 102 |
+
"flow/cos_sim": 0.5465538722195037,
|
| 103 |
+
"flow/improvement_ratio": 0.9649395484086041,
|
| 104 |
+
"flow/mag_ratio_mean": 0.545519895885633,
|
| 105 |
+
"flow/mag_ratio_std": 0.2078168946722327,
|
| 106 |
+
"step": 3072
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.18918294766985358,
|
| 110 |
+
"grad_norm": 0.3577847480773926,
|
| 111 |
+
"learning_rate": 0.0004962689322628078,
|
| 112 |
+
"loss": 1.775568962097168,
|
| 113 |
+
"step": 4096
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 0.18918294766985358,
|
| 117 |
+
"eval_cos_loss": 0.4333892414967219,
|
| 118 |
+
"eval_loss": 1.716203493342552,
|
| 119 |
+
"eval_mse_loss": 1.282814250689119,
|
| 120 |
+
"flow/cos_sim": 0.5666107895983953,
|
| 121 |
+
"flow/improvement_ratio": 0.9660467817086608,
|
| 122 |
+
"flow/mag_ratio_mean": 0.5606984513807515,
|
| 123 |
+
"flow/mag_ratio_std": 0.21290324090822646,
|
| 124 |
+
"step": 4096
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.18918294766985358,
|
| 128 |
+
"eval_cos_loss": 0.4333892414967219,
|
| 129 |
+
"eval_loss": 1.716203493342552,
|
| 130 |
+
"eval_mse_loss": 1.282814250689119,
|
| 131 |
+
"eval_runtime": 39.672,
|
| 132 |
+
"eval_samples_per_second": 705.612,
|
| 133 |
+
"eval_steps_per_second": 11.041,
|
| 134 |
+
"flow/cos_sim": 0.5666107895983953,
|
| 135 |
+
"flow/improvement_ratio": 0.9660467817086608,
|
| 136 |
+
"flow/mag_ratio_mean": 0.5606984513807515,
|
| 137 |
+
"flow/mag_ratio_std": 0.21290324090822646,
|
| 138 |
+
"step": 4096
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.236478684587317,
|
| 142 |
+
"grad_norm": 0.3629797697067261,
|
| 143 |
+
"learning_rate": 0.00048517275746994595,
|
| 144 |
+
"loss": 1.7335060834884644,
|
| 145 |
+
"step": 5120
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.236478684587317,
|
| 149 |
+
"eval_cos_loss": 0.42527746058762345,
|
| 150 |
+
"eval_loss": 1.6876028531218228,
|
| 151 |
+
"eval_mse_loss": 1.2623253905609861,
|
| 152 |
+
"flow/cos_sim": 0.5747225568991273,
|
| 153 |
+
"flow/improvement_ratio": 0.9670584807385049,
|
| 154 |
+
"flow/mag_ratio_mean": 0.569412647861324,
|
| 155 |
+
"flow/mag_ratio_std": 0.22070125653727413,
|
| 156 |
+
"step": 5120
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.236478684587317,
|
| 160 |
+
"eval_cos_loss": 0.42527746058762345,
|
| 161 |
+
"eval_loss": 1.6876028531218228,
|
| 162 |
+
"eval_mse_loss": 1.2623253905609861,
|
| 163 |
+
"eval_runtime": 39.6097,
|
| 164 |
+
"eval_samples_per_second": 706.721,
|
| 165 |
+
"eval_steps_per_second": 11.058,
|
| 166 |
+
"flow/cos_sim": 0.5747225568991273,
|
| 167 |
+
"flow/improvement_ratio": 0.9670584807385049,
|
| 168 |
+
"flow/mag_ratio_mean": 0.569412647861324,
|
| 169 |
+
"flow/mag_ratio_std": 0.22070125653727413,
|
| 170 |
+
"step": 5120
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.2837744215047804,
|
| 174 |
+
"grad_norm": 0.41703319549560547,
|
| 175 |
+
"learning_rate": 0.0004670433228990193,
|
| 176 |
+
"loss": 1.706570029258728,
|
| 177 |
+
"step": 6144
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.2837744215047804,
|
| 181 |
+
"eval_cos_loss": 0.41543219704606216,
|
| 182 |
+
"eval_loss": 1.6554871824778379,
|
| 183 |
+
"eval_mse_loss": 1.2400549827101024,
|
| 184 |
+
"flow/cos_sim": 0.5845678245912403,
|
| 185 |
+
"flow/improvement_ratio": 0.9684235336573701,
|
| 186 |
+
"flow/mag_ratio_mean": 0.580940374636759,
|
| 187 |
+
"flow/mag_ratio_std": 0.21690591768314849,
|
| 188 |
+
"step": 6144
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.2837744215047804,
|
| 192 |
+
"eval_cos_loss": 0.41543219704606216,
|
| 193 |
+
"eval_loss": 1.6554871824778379,
|
| 194 |
+
"eval_mse_loss": 1.2400549827101024,
|
| 195 |
+
"eval_runtime": 39.4039,
|
| 196 |
+
"eval_samples_per_second": 710.412,
|
| 197 |
+
"eval_steps_per_second": 11.116,
|
| 198 |
+
"flow/cos_sim": 0.5845678245912403,
|
| 199 |
+
"flow/improvement_ratio": 0.9684235336573701,
|
| 200 |
+
"flow/mag_ratio_mean": 0.580940374636759,
|
| 201 |
+
"flow/mag_ratio_std": 0.21690591768314849,
|
| 202 |
+
"step": 6144
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.3310701584222438,
|
| 206 |
+
"grad_norm": 0.4814172387123108,
|
| 207 |
+
"learning_rate": 0.0004424228215503503,
|
| 208 |
+
"loss": 1.6873632669448853,
|
| 209 |
+
"step": 7168
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.3310701584222438,
|
| 213 |
+
"eval_cos_loss": 0.41339083641903585,
|
| 214 |
+
"eval_loss": 1.6469332964997314,
|
| 215 |
+
"eval_mse_loss": 1.2335424589239843,
|
| 216 |
+
"flow/cos_sim": 0.5866091871914798,
|
| 217 |
+
"flow/improvement_ratio": 0.9676676742290253,
|
| 218 |
+
"flow/mag_ratio_mean": 0.5795940211102298,
|
| 219 |
+
"flow/mag_ratio_std": 0.21971321344103442,
|
| 220 |
+
"step": 7168
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.3310701584222438,
|
| 224 |
+
"eval_cos_loss": 0.41339083641903585,
|
| 225 |
+
"eval_loss": 1.6469332964997314,
|
| 226 |
+
"eval_mse_loss": 1.2335424589239843,
|
| 227 |
+
"eval_runtime": 39.4159,
|
| 228 |
+
"eval_samples_per_second": 710.196,
|
| 229 |
+
"eval_steps_per_second": 11.112,
|
| 230 |
+
"flow/cos_sim": 0.5866091871914798,
|
| 231 |
+
"flow/improvement_ratio": 0.9676676742290253,
|
| 232 |
+
"flow/mag_ratio_mean": 0.5795940211102298,
|
| 233 |
+
"flow/mag_ratio_std": 0.21971321344103442,
|
| 234 |
+
"step": 7168
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.37836589533970716,
|
| 238 |
+
"grad_norm": 0.3746698498725891,
|
| 239 |
+
"learning_rate": 0.0004120475733264409,
|
| 240 |
+
"loss": 1.6745531558990479,
|
| 241 |
+
"step": 8192
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.37836589533970716,
|
| 245 |
+
"eval_cos_loss": 0.40882552949260903,
|
| 246 |
+
"eval_loss": 1.6312648664870764,
|
| 247 |
+
"eval_mse_loss": 1.2224393353614633,
|
| 248 |
+
"flow/cos_sim": 0.5911744925529445,
|
| 249 |
+
"flow/improvement_ratio": 0.9690862153763096,
|
| 250 |
+
"flow/mag_ratio_mean": 0.5859045668007576,
|
| 251 |
+
"flow/mag_ratio_std": 0.2201043140806564,
|
| 252 |
+
"step": 8192
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.37836589533970716,
|
| 256 |
+
"eval_cos_loss": 0.40882552949260903,
|
| 257 |
+
"eval_loss": 1.6312648664870764,
|
| 258 |
+
"eval_mse_loss": 1.2224393353614633,
|
| 259 |
+
"eval_runtime": 39.6627,
|
| 260 |
+
"eval_samples_per_second": 705.777,
|
| 261 |
+
"eval_steps_per_second": 11.043,
|
| 262 |
+
"flow/cos_sim": 0.5911744925529445,
|
| 263 |
+
"flow/improvement_ratio": 0.9690862153763096,
|
| 264 |
+
"flow/mag_ratio_mean": 0.5859045668007576,
|
| 265 |
+
"flow/mag_ratio_std": 0.2201043140806564,
|
| 266 |
+
"step": 8192
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.4256616322571706,
|
| 270 |
+
"grad_norm": 0.47942957282066345,
|
| 271 |
+
"learning_rate": 0.0003768624321203836,
|
| 272 |
+
"loss": 1.6640042066574097,
|
| 273 |
+
"step": 9216
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.4256616322571706,
|
| 277 |
+
"eval_cos_loss": 0.4063904314688896,
|
| 278 |
+
"eval_loss": 1.6224081018199659,
|
| 279 |
+
"eval_mse_loss": 1.2160176700108671,
|
| 280 |
+
"flow/cos_sim": 0.593609596224136,
|
| 281 |
+
"flow/improvement_ratio": 0.9695234370830397,
|
| 282 |
+
"flow/mag_ratio_mean": 0.5825079079360178,
|
| 283 |
+
"flow/mag_ratio_std": 0.21707500255407264,
|
| 284 |
+
"step": 9216
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.4256616322571706,
|
| 288 |
+
"eval_cos_loss": 0.4063904314688896,
|
| 289 |
+
"eval_loss": 1.6224081018199659,
|
| 290 |
+
"eval_mse_loss": 1.2160176700108671,
|
| 291 |
+
"eval_runtime": 39.7053,
|
| 292 |
+
"eval_samples_per_second": 705.019,
|
| 293 |
+
"eval_steps_per_second": 11.031,
|
| 294 |
+
"flow/cos_sim": 0.593609596224136,
|
| 295 |
+
"flow/improvement_ratio": 0.9695234370830397,
|
| 296 |
+
"flow/mag_ratio_mean": 0.5825079079360178,
|
| 297 |
+
"flow/mag_ratio_std": 0.21707500255407264,
|
| 298 |
+
"step": 9216
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.472957369174634,
|
| 302 |
+
"grad_norm": 0.35636550188064575,
|
| 303 |
+
"learning_rate": 0.00033785105615189,
|
| 304 |
+
"loss": 1.6540666818618774,
|
| 305 |
+
"step": 10240
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.472957369174634,
|
| 309 |
+
"eval_cos_loss": 0.40575925862952456,
|
| 310 |
+
"eval_loss": 1.6192435720195508,
|
| 311 |
+
"eval_mse_loss": 1.213484312301357,
|
| 312 |
+
"flow/cos_sim": 0.5942407566118458,
|
| 313 |
+
"flow/improvement_ratio": 0.9697231067914397,
|
| 314 |
+
"flow/mag_ratio_mean": 0.5884986310244695,
|
| 315 |
+
"flow/mag_ratio_std": 0.2199755361703433,
|
| 316 |
+
"step": 10240
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.472957369174634,
|
| 320 |
+
"eval_cos_loss": 0.40575925862952456,
|
| 321 |
+
"eval_loss": 1.6192435720195508,
|
| 322 |
+
"eval_mse_loss": 1.213484312301357,
|
| 323 |
+
"eval_runtime": 39.5214,
|
| 324 |
+
"eval_samples_per_second": 708.3,
|
| 325 |
+
"eval_steps_per_second": 11.083,
|
| 326 |
+
"flow/cos_sim": 0.5942407566118458,
|
| 327 |
+
"flow/improvement_ratio": 0.9697231067914397,
|
| 328 |
+
"flow/mag_ratio_mean": 0.5884986310244695,
|
| 329 |
+
"flow/mag_ratio_std": 0.2199755361703433,
|
| 330 |
+
"step": 10240
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.5202531060920974,
|
| 334 |
+
"grad_norm": 0.35981205105781555,
|
| 335 |
+
"learning_rate": 0.0002962123380271837,
|
| 336 |
+
"loss": 1.6480820178985596,
|
| 337 |
+
"step": 11264
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.5202531060920974,
|
| 341 |
+
"eval_cos_loss": 0.4020878194265714,
|
| 342 |
+
"eval_loss": 1.6067346075898437,
|
| 343 |
+
"eval_mse_loss": 1.2046467928581586,
|
| 344 |
+
"flow/cos_sim": 0.5979121977880121,
|
| 345 |
+
"flow/improvement_ratio": 0.9696458948257307,
|
| 346 |
+
"flow/mag_ratio_mean": 0.5931163091365605,
|
| 347 |
+
"flow/mag_ratio_std": 0.2266381677563332,
|
| 348 |
+
"step": 11264
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.5202531060920974,
|
| 352 |
+
"eval_cos_loss": 0.4020878194265714,
|
| 353 |
+
"eval_loss": 1.6067346075898437,
|
| 354 |
+
"eval_mse_loss": 1.2046467928581586,
|
| 355 |
+
"eval_runtime": 40.0623,
|
| 356 |
+
"eval_samples_per_second": 698.736,
|
| 357 |
+
"eval_steps_per_second": 10.933,
|
| 358 |
+
"flow/cos_sim": 0.5979121977880121,
|
| 359 |
+
"flow/improvement_ratio": 0.9696458948257307,
|
| 360 |
+
"flow/mag_ratio_mean": 0.5931163091365605,
|
| 361 |
+
"flow/mag_ratio_std": 0.2266381677563332,
|
| 362 |
+
"step": 11264
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.5675488430095608,
|
| 366 |
+
"grad_norm": 0.5238872766494751,
|
| 367 |
+
"learning_rate": 0.00025323382764865535,
|
| 368 |
+
"loss": 1.6410928964614868,
|
| 369 |
+
"step": 12288
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.5675488430095608,
|
| 373 |
+
"eval_cos_loss": 0.4016706557714776,
|
| 374 |
+
"eval_loss": 1.6051203358118937,
|
| 375 |
+
"eval_mse_loss": 1.2034496806527926,
|
| 376 |
+
"flow/cos_sim": 0.5983293527337514,
|
| 377 |
+
"flow/improvement_ratio": 0.9703199659308342,
|
| 378 |
+
"flow/mag_ratio_mean": 0.5952222302351913,
|
| 379 |
+
"flow/mag_ratio_std": 0.2251834258709324,
|
| 380 |
+
"step": 12288
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.5675488430095608,
|
| 384 |
+
"eval_cos_loss": 0.4016706557714776,
|
| 385 |
+
"eval_loss": 1.6051203358118937,
|
| 386 |
+
"eval_mse_loss": 1.2034496806527926,
|
| 387 |
+
"eval_runtime": 39.8137,
|
| 388 |
+
"eval_samples_per_second": 703.099,
|
| 389 |
+
"eval_steps_per_second": 11.001,
|
| 390 |
+
"flow/cos_sim": 0.5983293527337514,
|
| 391 |
+
"flow/improvement_ratio": 0.9703199659308342,
|
| 392 |
+
"flow/mag_ratio_mean": 0.5952222302351913,
|
| 393 |
+
"flow/mag_ratio_std": 0.2251834258709324,
|
| 394 |
+
"step": 12288
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.6148445799270241,
|
| 398 |
+
"grad_norm": 0.26784732937812805,
|
| 399 |
+
"learning_rate": 0.000210117059628806,
|
| 400 |
+
"loss": 1.63731050491333,
|
| 401 |
+
"step": 13312
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.6148445799270241,
|
| 405 |
+
"eval_cos_loss": 0.4014841335140951,
|
| 406 |
+
"eval_loss": 1.604530786814755,
|
| 407 |
+
"eval_mse_loss": 1.2030466533687016,
|
| 408 |
+
"flow/cos_sim": 0.5985158796179785,
|
| 409 |
+
"flow/improvement_ratio": 0.9693736963348302,
|
| 410 |
+
"flow/mag_ratio_mean": 0.5919299127032224,
|
| 411 |
+
"flow/mag_ratio_std": 0.22532628976727184,
|
| 412 |
+
"step": 13312
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 0.6148445799270241,
|
| 416 |
+
"eval_cos_loss": 0.4014841335140951,
|
| 417 |
+
"eval_loss": 1.604530786814755,
|
| 418 |
+
"eval_mse_loss": 1.2030466533687016,
|
| 419 |
+
"eval_runtime": 40.496,
|
| 420 |
+
"eval_samples_per_second": 691.254,
|
| 421 |
+
"eval_steps_per_second": 10.816,
|
| 422 |
+
"flow/cos_sim": 0.5985158796179785,
|
| 423 |
+
"flow/improvement_ratio": 0.9693736963348302,
|
| 424 |
+
"flow/mag_ratio_mean": 0.5919299127032224,
|
| 425 |
+
"flow/mag_ratio_std": 0.22532628976727184,
|
| 426 |
+
"step": 13312
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.6621403168444876,
|
| 430 |
+
"grad_norm": 0.43763285875320435,
|
| 431 |
+
"learning_rate": 0.0001682330091706446,
|
| 432 |
+
"loss": 1.6302827596664429,
|
| 433 |
+
"step": 14336
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.6621403168444876,
|
| 437 |
+
"eval_cos_loss": 0.3985591554614507,
|
| 438 |
+
"eval_loss": 1.593907880184313,
|
| 439 |
+
"eval_mse_loss": 1.1953487265599918,
|
| 440 |
+
"flow/cos_sim": 0.6014408363054876,
|
| 441 |
+
"flow/improvement_ratio": 0.9697505197568571,
|
| 442 |
+
"flow/mag_ratio_mean": 0.5954256657868215,
|
| 443 |
+
"flow/mag_ratio_std": 0.2259266352340511,
|
| 444 |
+
"step": 14336
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.6621403168444876,
|
| 448 |
+
"eval_cos_loss": 0.3985591554614507,
|
| 449 |
+
"eval_loss": 1.593907880184313,
|
| 450 |
+
"eval_mse_loss": 1.1953487265599918,
|
| 451 |
+
"eval_runtime": 40.0371,
|
| 452 |
+
"eval_samples_per_second": 699.177,
|
| 453 |
+
"eval_steps_per_second": 10.94,
|
| 454 |
+
"flow/cos_sim": 0.6014408363054876,
|
| 455 |
+
"flow/improvement_ratio": 0.9697505197568571,
|
| 456 |
+
"flow/mag_ratio_mean": 0.5954256657868215,
|
| 457 |
+
"flow/mag_ratio_std": 0.2259266352340511,
|
| 458 |
+
"step": 14336
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.709436053761951,
|
| 462 |
+
"grad_norm": 0.341217041015625,
|
| 463 |
+
"learning_rate": 0.0001287526134679914,
|
| 464 |
+
"loss": 1.6316653490066528,
|
| 465 |
+
"step": 15360
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.709436053761951,
|
| 469 |
+
"eval_cos_loss": 0.3979599736460812,
|
| 470 |
+
"eval_loss": 1.5915418874183203,
|
| 471 |
+
"eval_mse_loss": 1.1935819131598624,
|
| 472 |
+
"flow/cos_sim": 0.6020400329539765,
|
| 473 |
+
"flow/improvement_ratio": 0.9696243418950469,
|
| 474 |
+
"flow/mag_ratio_mean": 0.595387307612319,
|
| 475 |
+
"flow/mag_ratio_std": 0.22555491620819318,
|
| 476 |
+
"step": 15360
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.709436053761951,
|
| 480 |
+
"eval_cos_loss": 0.3979599736460812,
|
| 481 |
+
"eval_loss": 1.5915418874183203,
|
| 482 |
+
"eval_mse_loss": 1.1935819131598624,
|
| 483 |
+
"eval_runtime": 39.8928,
|
| 484 |
+
"eval_samples_per_second": 701.706,
|
| 485 |
+
"eval_steps_per_second": 10.979,
|
| 486 |
+
"flow/cos_sim": 0.6020400329539765,
|
| 487 |
+
"flow/improvement_ratio": 0.9696243418950469,
|
| 488 |
+
"flow/mag_ratio_mean": 0.595387307612319,
|
| 489 |
+
"flow/mag_ratio_std": 0.22555491620819318,
|
| 490 |
+
"step": 15360
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.7567317906794143,
|
| 494 |
+
"grad_norm": 0.25786092877388,
|
| 495 |
+
"learning_rate": 9.293122272447036e-05,
|
| 496 |
+
"loss": 1.6261003017425537,
|
| 497 |
+
"step": 16384
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.7567317906794143,
|
| 501 |
+
"eval_cos_loss": 0.3959788013947064,
|
| 502 |
+
"eval_loss": 1.5846927321120485,
|
| 503 |
+
"eval_mse_loss": 1.1887139288802124,
|
| 504 |
+
"flow/cos_sim": 0.6040212045249329,
|
| 505 |
+
"flow/improvement_ratio": 0.9709213373323554,
|
| 506 |
+
"flow/mag_ratio_mean": 0.5975744849470652,
|
| 507 |
+
"flow/mag_ratio_std": 0.2268577309912198,
|
| 508 |
+
"step": 16384
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.7567317906794143,
|
| 512 |
+
"eval_cos_loss": 0.3959788013947064,
|
| 513 |
+
"eval_loss": 1.5846927321120485,
|
| 514 |
+
"eval_mse_loss": 1.1887139288802124,
|
| 515 |
+
"eval_runtime": 40.2532,
|
| 516 |
+
"eval_samples_per_second": 695.423,
|
| 517 |
+
"eval_steps_per_second": 10.881,
|
| 518 |
+
"flow/cos_sim": 0.6040212045249329,
|
| 519 |
+
"flow/improvement_ratio": 0.9709213373323554,
|
| 520 |
+
"flow/mag_ratio_mean": 0.5975744849470652,
|
| 521 |
+
"flow/mag_ratio_std": 0.2268577309912198,
|
| 522 |
+
"step": 16384
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.8040275275968778,
|
| 526 |
+
"grad_norm": 0.30163654685020447,
|
| 527 |
+
"learning_rate": 6.177028255713927e-05,
|
| 528 |
+
"loss": 1.6237930059432983,
|
| 529 |
+
"step": 17408
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.8040275275968778,
|
| 533 |
+
"eval_cos_loss": 0.397802642967603,
|
| 534 |
+
"eval_loss": 1.5906421003276354,
|
| 535 |
+
"eval_mse_loss": 1.1928394578363253,
|
| 536 |
+
"flow/cos_sim": 0.6021973686675502,
|
| 537 |
+
"flow/improvement_ratio": 0.9696774086723589,
|
| 538 |
+
"flow/mag_ratio_mean": 0.5974947977555941,
|
| 539 |
+
"flow/mag_ratio_std": 0.22862732230120053,
|
| 540 |
+
"step": 17408
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.8040275275968778,
|
| 544 |
+
"eval_cos_loss": 0.397802642967603,
|
| 545 |
+
"eval_loss": 1.5906421003276354,
|
| 546 |
+
"eval_mse_loss": 1.1928394578363253,
|
| 547 |
+
"eval_runtime": 39.4241,
|
| 548 |
+
"eval_samples_per_second": 710.047,
|
| 549 |
+
"eval_steps_per_second": 11.11,
|
| 550 |
+
"flow/cos_sim": 0.6021973686675502,
|
| 551 |
+
"flow/improvement_ratio": 0.9696774086723589,
|
| 552 |
+
"flow/mag_ratio_mean": 0.5974947977555941,
|
| 553 |
+
"flow/mag_ratio_std": 0.22862732230120053,
|
| 554 |
+
"step": 17408
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.8513232645143411,
|
| 558 |
+
"grad_norm": 0.3143230974674225,
|
| 559 |
+
"learning_rate": 3.623868702551317e-05,
|
| 560 |
+
"loss": 1.623117446899414,
|
| 561 |
+
"step": 18432
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.8513232645143411,
|
| 565 |
+
"eval_cos_loss": 0.39859252644183973,
|
| 566 |
+
"eval_loss": 1.5927672549469831,
|
| 567 |
+
"eval_mse_loss": 1.1941747259875954,
|
| 568 |
+
"flow/cos_sim": 0.6014074842407279,
|
| 569 |
+
"flow/improvement_ratio": 0.9700041171078269,
|
| 570 |
+
"flow/mag_ratio_mean": 0.5950140223655527,
|
| 571 |
+
"flow/mag_ratio_std": 0.22678067565780796,
|
| 572 |
+
"step": 18432
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.8513232645143411,
|
| 576 |
+
"eval_cos_loss": 0.39859252644183973,
|
| 577 |
+
"eval_loss": 1.5927672549469831,
|
| 578 |
+
"eval_mse_loss": 1.1941747259875954,
|
| 579 |
+
"eval_runtime": 40.1556,
|
| 580 |
+
"eval_samples_per_second": 697.113,
|
| 581 |
+
"eval_steps_per_second": 10.908,
|
| 582 |
+
"flow/cos_sim": 0.6014074842407279,
|
| 583 |
+
"flow/improvement_ratio": 0.9700041171078269,
|
| 584 |
+
"flow/mag_ratio_mean": 0.5950140223655527,
|
| 585 |
+
"flow/mag_ratio_std": 0.22678067565780796,
|
| 586 |
+
"step": 18432
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.8986190014318045,
|
| 590 |
+
"grad_norm": 0.23344755172729492,
|
| 591 |
+
"learning_rate": 1.710000392483449e-05,
|
| 592 |
+
"loss": 1.6220356225967407,
|
| 593 |
+
"step": 19456
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.8986190014318045,
|
| 597 |
+
"eval_cos_loss": 0.3966924757717951,
|
| 598 |
+
"eval_loss": 1.5871477426459255,
|
| 599 |
+
"eval_mse_loss": 1.1904552660576284,
|
| 600 |
+
"flow/cos_sim": 0.6033075423273322,
|
| 601 |
+
"flow/improvement_ratio": 0.9706315692156962,
|
| 602 |
+
"flow/mag_ratio_mean": 0.5972542343618663,
|
| 603 |
+
"flow/mag_ratio_std": 0.22712956321293906,
|
| 604 |
+
"step": 19456
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.8986190014318045,
|
| 608 |
+
"eval_cos_loss": 0.3966924757717951,
|
| 609 |
+
"eval_loss": 1.5871477426459255,
|
| 610 |
+
"eval_mse_loss": 1.1904552660576284,
|
| 611 |
+
"eval_runtime": 39.2746,
|
| 612 |
+
"eval_samples_per_second": 712.751,
|
| 613 |
+
"eval_steps_per_second": 11.152,
|
| 614 |
+
"flow/cos_sim": 0.6033075423273322,
|
| 615 |
+
"flow/improvement_ratio": 0.9706315692156962,
|
| 616 |
+
"flow/mag_ratio_mean": 0.5972542343618663,
|
| 617 |
+
"flow/mag_ratio_std": 0.22712956321293906,
|
| 618 |
+
"step": 19456
|
| 619 |
+
}
|
| 620 |
+
],
|
| 621 |
+
"logging_steps": 1024,
|
| 622 |
+
"max_steps": 21651,
|
| 623 |
+
"num_input_tokens_seen": 0,
|
| 624 |
+
"num_train_epochs": 1,
|
| 625 |
+
"save_steps": 1024,
|
| 626 |
+
"stateful_callbacks": {
|
| 627 |
+
"TrainerControl": {
|
| 628 |
+
"args": {
|
| 629 |
+
"should_epoch_stop": false,
|
| 630 |
+
"should_evaluate": false,
|
| 631 |
+
"should_log": false,
|
| 632 |
+
"should_save": true,
|
| 633 |
+
"should_training_stop": false
|
| 634 |
+
},
|
| 635 |
+
"attributes": {}
|
| 636 |
+
}
|
| 637 |
+
},
|
| 638 |
+
"total_flos": 0.0,
|
| 639 |
+
"train_batch_size": 64,
|
| 640 |
+
"trial_name": null,
|
| 641 |
+
"trial_params": null
|
| 642 |
+
}
|
checkpoints-v2.6-d/checkpoint-19456/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:35ff62bd01803980b2adbf28bfd2750093914b32565b65c1bc28d36ccd64d3ae
|
| 3 |
+
size 5137
|