Upload folder using huggingface_hub
Browse files
checkpoint-24012/config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"MMoEViT"
|
| 4 |
+
],
|
| 5 |
+
"torch_dtype": "float32",
|
| 6 |
+
"transformers_version": "4.47.0"
|
| 7 |
+
}
|
checkpoint-24012/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:71f4cc6af02598af91d8f81fcfc3c54835d9aae0fde639f7c44b8742de4c97eb
|
| 3 |
+
size 359190616
|
checkpoint-24012/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ddf47d0d1e33cabbcd1c9aedb3564e96a4dd0a9983336fccdc0ccc2d279afb3
|
| 3 |
+
size 718509882
|
checkpoint-24012/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:db1918ecd20c1878b0401c6166770f4b5e553e5ae2878e7703adb3ee5f10c831
|
| 3 |
+
size 14244
|
checkpoint-24012/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:547be801b540007f8911172aaba1ef1e48aba1e11b2f98058477f0eaa860f4e1
|
| 3 |
+
size 1064
|
checkpoint-24012/trainer_state.json
ADDED
|
@@ -0,0 +1,642 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9661077074670175,
|
| 3 |
+
"best_model_checkpoint": "/kaggle/working/mmoe_vit_results/checkpoint-24012",
|
| 4 |
+
"epoch": 6.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 24012,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.12493753123438281,
|
| 13 |
+
"grad_norm": 2.205951452255249,
|
| 14 |
+
"learning_rate": 5e-05,
|
| 15 |
+
"loss": 0.7943,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.24987506246876562,
|
| 20 |
+
"grad_norm": 2.1307880878448486,
|
| 21 |
+
"learning_rate": 4.9091371665334015e-05,
|
| 22 |
+
"loss": 0.474,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3748125937031484,
|
| 27 |
+
"grad_norm": 1.7951563596725464,
|
| 28 |
+
"learning_rate": 4.818274333066802e-05,
|
| 29 |
+
"loss": 0.4167,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.49975012493753124,
|
| 34 |
+
"grad_norm": 1.9876887798309326,
|
| 35 |
+
"learning_rate": 4.727411499600204e-05,
|
| 36 |
+
"loss": 0.3916,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.624687656171914,
|
| 41 |
+
"grad_norm": 3.0330495834350586,
|
| 42 |
+
"learning_rate": 4.636548666133605e-05,
|
| 43 |
+
"loss": 0.3704,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.7496251874062968,
|
| 48 |
+
"grad_norm": 2.0157809257507324,
|
| 49 |
+
"learning_rate": 4.545685832667006e-05,
|
| 50 |
+
"loss": 0.3577,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.8745627186406797,
|
| 55 |
+
"grad_norm": 1.7754454612731934,
|
| 56 |
+
"learning_rate": 4.4548229992004074e-05,
|
| 57 |
+
"loss": 0.3447,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.9995002498750625,
|
| 62 |
+
"grad_norm": 1.7371801137924194,
|
| 63 |
+
"learning_rate": 4.3639601657338087e-05,
|
| 64 |
+
"loss": 0.3348,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.0,
|
| 69 |
+
"eval_category_macro_f1": 0.6952318859438807,
|
| 70 |
+
"eval_category_precision": 0.956608825785308,
|
| 71 |
+
"eval_category_recall": 0.9564824339910845,
|
| 72 |
+
"eval_category_weighted_f1": 0.9548015657704888,
|
| 73 |
+
"eval_color_macro_f1": 0.26886041835616176,
|
| 74 |
+
"eval_color_precision": 0.742503302340083,
|
| 75 |
+
"eval_color_recall": 0.7711898749961034,
|
| 76 |
+
"eval_color_weighted_f1": 0.7494002787544956,
|
| 77 |
+
"eval_gender_macro_f1": 0.5551422720580631,
|
| 78 |
+
"eval_gender_precision": 0.9135315056899179,
|
| 79 |
+
"eval_gender_recall": 0.9142429626858692,
|
| 80 |
+
"eval_gender_weighted_f1": 0.9133261341390547,
|
| 81 |
+
"eval_loss": 0.3139565885066986,
|
| 82 |
+
"eval_material_macro_f1": 0.18175933111933457,
|
| 83 |
+
"eval_material_precision": 0.5510795574472717,
|
| 84 |
+
"eval_material_recall": 0.5989276473705539,
|
| 85 |
+
"eval_material_weighted_f1": 0.5328149786359672,
|
| 86 |
+
"eval_neck_macro_f1": 0.17476784999699163,
|
| 87 |
+
"eval_neck_precision": 0.770231966081002,
|
| 88 |
+
"eval_neck_recall": 0.767043860469466,
|
| 89 |
+
"eval_neck_weighted_f1": 0.7547329696141571,
|
| 90 |
+
"eval_pattern_macro_f1": 0.07495456820600468,
|
| 91 |
+
"eval_pattern_precision": 0.500470399822932,
|
| 92 |
+
"eval_pattern_recall": 0.5943763833037189,
|
| 93 |
+
"eval_pattern_weighted_f1": 0.5043529662664364,
|
| 94 |
+
"eval_product_type_macro_f1": 0.5286677539064947,
|
| 95 |
+
"eval_product_type_precision": 0.6574438114334396,
|
| 96 |
+
"eval_product_type_recall": 0.6384238910190467,
|
| 97 |
+
"eval_product_type_weighted_f1": 0.6249330379628022,
|
| 98 |
+
"eval_runtime": 307.6972,
|
| 99 |
+
"eval_samples_per_second": 104.255,
|
| 100 |
+
"eval_sleeve_macro_f1": 0.3129635925848871,
|
| 101 |
+
"eval_sleeve_precision": 0.8408246980460057,
|
| 102 |
+
"eval_sleeve_recall": 0.8423579288631192,
|
| 103 |
+
"eval_sleeve_weighted_f1": 0.830038616651521,
|
| 104 |
+
"eval_steps_per_second": 1.631,
|
| 105 |
+
"eval_style_macro_f1": 0.14293232343466503,
|
| 106 |
+
"eval_style_precision": 0.5728463487218707,
|
| 107 |
+
"eval_style_recall": 0.6262975778546713,
|
| 108 |
+
"eval_style_weighted_f1": 0.5804908017903951,
|
| 109 |
+
"step": 4002
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.1244377811094453,
|
| 113 |
+
"grad_norm": 1.4916481971740723,
|
| 114 |
+
"learning_rate": 4.27309733226721e-05,
|
| 115 |
+
"loss": 0.3134,
|
| 116 |
+
"step": 4500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.249375312343828,
|
| 120 |
+
"grad_norm": 1.7547612190246582,
|
| 121 |
+
"learning_rate": 4.1822344988006106e-05,
|
| 122 |
+
"loss": 0.3112,
|
| 123 |
+
"step": 5000
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 1.3743128435782108,
|
| 127 |
+
"grad_norm": 1.4588732719421387,
|
| 128 |
+
"learning_rate": 4.091371665334012e-05,
|
| 129 |
+
"loss": 0.3051,
|
| 130 |
+
"step": 5500
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.4992503748125938,
|
| 134 |
+
"grad_norm": 1.2639210224151611,
|
| 135 |
+
"learning_rate": 4.000508831867413e-05,
|
| 136 |
+
"loss": 0.3029,
|
| 137 |
+
"step": 6000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.6241879060469766,
|
| 141 |
+
"grad_norm": 1.4789787530899048,
|
| 142 |
+
"learning_rate": 3.9096459984008145e-05,
|
| 143 |
+
"loss": 0.2961,
|
| 144 |
+
"step": 6500
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.7491254372813594,
|
| 148 |
+
"grad_norm": 1.6470601558685303,
|
| 149 |
+
"learning_rate": 3.818783164934215e-05,
|
| 150 |
+
"loss": 0.2873,
|
| 151 |
+
"step": 7000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.8740629685157422,
|
| 155 |
+
"grad_norm": 2.157299280166626,
|
| 156 |
+
"learning_rate": 3.7279203314676164e-05,
|
| 157 |
+
"loss": 0.2847,
|
| 158 |
+
"step": 7500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.999000499750125,
|
| 162 |
+
"grad_norm": 1.4557589292526245,
|
| 163 |
+
"learning_rate": 3.637057498001018e-05,
|
| 164 |
+
"loss": 0.284,
|
| 165 |
+
"step": 8000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.0,
|
| 169 |
+
"eval_category_macro_f1": 0.7731459035999274,
|
| 170 |
+
"eval_category_precision": 0.9621854954403526,
|
| 171 |
+
"eval_category_recall": 0.9620935814707441,
|
| 172 |
+
"eval_category_weighted_f1": 0.9611122266129147,
|
| 173 |
+
"eval_color_macro_f1": 0.33282064622069346,
|
| 174 |
+
"eval_color_precision": 0.7628507936505963,
|
| 175 |
+
"eval_color_recall": 0.7826927273294055,
|
| 176 |
+
"eval_color_weighted_f1": 0.7666531131477731,
|
| 177 |
+
"eval_gender_macro_f1": 0.6529045193901017,
|
| 178 |
+
"eval_gender_precision": 0.9310067709590051,
|
| 179 |
+
"eval_gender_recall": 0.9307023286262041,
|
| 180 |
+
"eval_gender_weighted_f1": 0.9300238040883521,
|
| 181 |
+
"eval_loss": 0.27808678150177,
|
| 182 |
+
"eval_material_macro_f1": 0.2672650569326831,
|
| 183 |
+
"eval_material_precision": 0.5829820463502581,
|
| 184 |
+
"eval_material_recall": 0.6170703575547867,
|
| 185 |
+
"eval_material_weighted_f1": 0.5634078038324131,
|
| 186 |
+
"eval_neck_macro_f1": 0.20473481727652498,
|
| 187 |
+
"eval_neck_precision": 0.7723201257575838,
|
| 188 |
+
"eval_neck_recall": 0.7876180678948845,
|
| 189 |
+
"eval_neck_weighted_f1": 0.768304127766941,
|
| 190 |
+
"eval_pattern_macro_f1": 0.11918967355596528,
|
| 191 |
+
"eval_pattern_precision": 0.5371433805351696,
|
| 192 |
+
"eval_pattern_recall": 0.6123320552386297,
|
| 193 |
+
"eval_pattern_weighted_f1": 0.5328905424758658,
|
| 194 |
+
"eval_product_type_macro_f1": 0.6181561032086245,
|
| 195 |
+
"eval_product_type_precision": 0.6896032942264115,
|
| 196 |
+
"eval_product_type_recall": 0.6778577885844321,
|
| 197 |
+
"eval_product_type_weighted_f1": 0.67171279677715,
|
| 198 |
+
"eval_runtime": 302.2512,
|
| 199 |
+
"eval_samples_per_second": 106.134,
|
| 200 |
+
"eval_sleeve_macro_f1": 0.34627852813600124,
|
| 201 |
+
"eval_sleeve_precision": 0.8389224468125812,
|
| 202 |
+
"eval_sleeve_recall": 0.8489042675893888,
|
| 203 |
+
"eval_sleeve_weighted_f1": 0.8375757784906263,
|
| 204 |
+
"eval_steps_per_second": 1.661,
|
| 205 |
+
"eval_style_macro_f1": 0.18292409377977412,
|
| 206 |
+
"eval_style_precision": 0.5948748921293634,
|
| 207 |
+
"eval_style_recall": 0.6488356868979707,
|
| 208 |
+
"eval_style_weighted_f1": 0.6017236316787142,
|
| 209 |
+
"step": 8004
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.1239380309845077,
|
| 213 |
+
"grad_norm": 1.367946743965149,
|
| 214 |
+
"learning_rate": 3.546194664534419e-05,
|
| 215 |
+
"loss": 0.2605,
|
| 216 |
+
"step": 8500
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.2488755622188905,
|
| 220 |
+
"grad_norm": 1.5504860877990723,
|
| 221 |
+
"learning_rate": 3.45533183106782e-05,
|
| 222 |
+
"loss": 0.2633,
|
| 223 |
+
"step": 9000
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 2.3738130934532733,
|
| 227 |
+
"grad_norm": 1.5913927555084229,
|
| 228 |
+
"learning_rate": 3.364468997601221e-05,
|
| 229 |
+
"loss": 0.2561,
|
| 230 |
+
"step": 9500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.498750624687656,
|
| 234 |
+
"grad_norm": 1.7354265451431274,
|
| 235 |
+
"learning_rate": 3.273606164134623e-05,
|
| 236 |
+
"loss": 0.2571,
|
| 237 |
+
"step": 10000
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 2.623688155922039,
|
| 241 |
+
"grad_norm": 1.5740700960159302,
|
| 242 |
+
"learning_rate": 3.1827433306680235e-05,
|
| 243 |
+
"loss": 0.2533,
|
| 244 |
+
"step": 10500
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 2.7486256871564216,
|
| 248 |
+
"grad_norm": 1.3308007717132568,
|
| 249 |
+
"learning_rate": 3.091880497201425e-05,
|
| 250 |
+
"loss": 0.2539,
|
| 251 |
+
"step": 11000
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 2.873563218390805,
|
| 255 |
+
"grad_norm": 1.4897780418395996,
|
| 256 |
+
"learning_rate": 3.0010176637348258e-05,
|
| 257 |
+
"loss": 0.2471,
|
| 258 |
+
"step": 11500
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.9985007496251876,
|
| 262 |
+
"grad_norm": 1.9843637943267822,
|
| 263 |
+
"learning_rate": 2.9101548302682274e-05,
|
| 264 |
+
"loss": 0.2483,
|
| 265 |
+
"step": 12000
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 3.0,
|
| 269 |
+
"eval_category_macro_f1": 0.7903733270490069,
|
| 270 |
+
"eval_category_precision": 0.9645144186683499,
|
| 271 |
+
"eval_category_recall": 0.9642756943795006,
|
| 272 |
+
"eval_category_weighted_f1": 0.963310209667099,
|
| 273 |
+
"eval_color_macro_f1": 0.3621249155624305,
|
| 274 |
+
"eval_color_precision": 0.7802552212263318,
|
| 275 |
+
"eval_color_recall": 0.7986533246048817,
|
| 276 |
+
"eval_color_weighted_f1": 0.7810869424554918,
|
| 277 |
+
"eval_gender_macro_f1": 0.682238598199015,
|
| 278 |
+
"eval_gender_precision": 0.9267564337966526,
|
| 279 |
+
"eval_gender_recall": 0.9252782193958664,
|
| 280 |
+
"eval_gender_weighted_f1": 0.9249918474296774,
|
| 281 |
+
"eval_loss": 0.25919318199157715,
|
| 282 |
+
"eval_material_macro_f1": 0.3398667069927169,
|
| 283 |
+
"eval_material_precision": 0.5978318966832341,
|
| 284 |
+
"eval_material_recall": 0.6357741824869853,
|
| 285 |
+
"eval_material_weighted_f1": 0.5944051235584699,
|
| 286 |
+
"eval_neck_macro_f1": 0.23924719225254232,
|
| 287 |
+
"eval_neck_precision": 0.7806692719164586,
|
| 288 |
+
"eval_neck_recall": 0.7925122354188098,
|
| 289 |
+
"eval_neck_weighted_f1": 0.7790870348414296,
|
| 290 |
+
"eval_pattern_macro_f1": 0.1641511847711344,
|
| 291 |
+
"eval_pattern_precision": 0.5819393078338299,
|
| 292 |
+
"eval_pattern_recall": 0.6225879859097853,
|
| 293 |
+
"eval_pattern_weighted_f1": 0.5502139317041554,
|
| 294 |
+
"eval_product_type_macro_f1": 0.6583201290239673,
|
| 295 |
+
"eval_product_type_precision": 0.7128711001410035,
|
| 296 |
+
"eval_product_type_recall": 0.6994918794226753,
|
| 297 |
+
"eval_product_type_weighted_f1": 0.6929448155349042,
|
| 298 |
+
"eval_runtime": 301.2799,
|
| 299 |
+
"eval_samples_per_second": 106.476,
|
| 300 |
+
"eval_sleeve_macro_f1": 0.3563735801589765,
|
| 301 |
+
"eval_sleeve_precision": 0.8406013881476055,
|
| 302 |
+
"eval_sleeve_recall": 0.8499018049191059,
|
| 303 |
+
"eval_sleeve_weighted_f1": 0.840146351199092,
|
| 304 |
+
"eval_steps_per_second": 1.666,
|
| 305 |
+
"eval_style_macro_f1": 0.22006801367696224,
|
| 306 |
+
"eval_style_precision": 0.6224502543435393,
|
| 307 |
+
"eval_style_recall": 0.6560678325384208,
|
| 308 |
+
"eval_style_weighted_f1": 0.622977679977427,
|
| 309 |
+
"step": 12006
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 3.1234382808595704,
|
| 313 |
+
"grad_norm": 2.0150625705718994,
|
| 314 |
+
"learning_rate": 2.8192919968016284e-05,
|
| 315 |
+
"loss": 0.2251,
|
| 316 |
+
"step": 12500
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 3.248375812093953,
|
| 320 |
+
"grad_norm": 1.2645131349563599,
|
| 321 |
+
"learning_rate": 2.7284291633350297e-05,
|
| 322 |
+
"loss": 0.2251,
|
| 323 |
+
"step": 13000
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 3.373313343328336,
|
| 327 |
+
"grad_norm": 1.7234032154083252,
|
| 328 |
+
"learning_rate": 2.6375663298684306e-05,
|
| 329 |
+
"loss": 0.2239,
|
| 330 |
+
"step": 13500
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 3.4982508745627188,
|
| 334 |
+
"grad_norm": 1.676547646522522,
|
| 335 |
+
"learning_rate": 2.5467034964018323e-05,
|
| 336 |
+
"loss": 0.2254,
|
| 337 |
+
"step": 14000
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 3.6231884057971016,
|
| 341 |
+
"grad_norm": 1.3010063171386719,
|
| 342 |
+
"learning_rate": 2.4558406629352332e-05,
|
| 343 |
+
"loss": 0.2241,
|
| 344 |
+
"step": 14500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 3.7481259370314843,
|
| 348 |
+
"grad_norm": 1.72947359085083,
|
| 349 |
+
"learning_rate": 2.3649778294686342e-05,
|
| 350 |
+
"loss": 0.2222,
|
| 351 |
+
"step": 15000
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 3.873063468265867,
|
| 355 |
+
"grad_norm": 1.47541081905365,
|
| 356 |
+
"learning_rate": 2.2741149960020355e-05,
|
| 357 |
+
"loss": 0.2177,
|
| 358 |
+
"step": 15500
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 3.99800099950025,
|
| 362 |
+
"grad_norm": 1.4984639883041382,
|
| 363 |
+
"learning_rate": 2.1832521625354364e-05,
|
| 364 |
+
"loss": 0.2146,
|
| 365 |
+
"step": 16000
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 4.0,
|
| 369 |
+
"eval_category_macro_f1": 0.8070428225617673,
|
| 370 |
+
"eval_category_precision": 0.9639301959345535,
|
| 371 |
+
"eval_category_recall": 0.9639951370055176,
|
| 372 |
+
"eval_category_weighted_f1": 0.9633956909548311,
|
| 373 |
+
"eval_color_macro_f1": 0.4160977179764728,
|
| 374 |
+
"eval_color_precision": 0.79551435648077,
|
| 375 |
+
"eval_color_recall": 0.80382804950279,
|
| 376 |
+
"eval_color_weighted_f1": 0.7932115174135331,
|
| 377 |
+
"eval_gender_macro_f1": 0.7138550749176814,
|
| 378 |
+
"eval_gender_precision": 0.9418271715431166,
|
| 379 |
+
"eval_gender_recall": 0.9409894323389133,
|
| 380 |
+
"eval_gender_weighted_f1": 0.9408785911526341,
|
| 381 |
+
"eval_loss": 0.24342668056488037,
|
| 382 |
+
"eval_material_macro_f1": 0.3788597805248959,
|
| 383 |
+
"eval_material_precision": 0.6203377475947315,
|
| 384 |
+
"eval_material_recall": 0.6495214938121513,
|
| 385 |
+
"eval_material_weighted_f1": 0.6137247593163535,
|
| 386 |
+
"eval_neck_macro_f1": 0.26417595340931804,
|
| 387 |
+
"eval_neck_precision": 0.7888860680739177,
|
| 388 |
+
"eval_neck_recall": 0.8017394557186944,
|
| 389 |
+
"eval_neck_weighted_f1": 0.7912357278702308,
|
| 390 |
+
"eval_pattern_macro_f1": 0.21137706069820827,
|
| 391 |
+
"eval_pattern_precision": 0.5896656959683353,
|
| 392 |
+
"eval_pattern_recall": 0.63402849215998,
|
| 393 |
+
"eval_pattern_weighted_f1": 0.5805939336450403,
|
| 394 |
+
"eval_product_type_macro_f1": 0.7027375784160153,
|
| 395 |
+
"eval_product_type_precision": 0.7257145790424636,
|
| 396 |
+
"eval_product_type_recall": 0.7205025094298451,
|
| 397 |
+
"eval_product_type_weighted_f1": 0.7160893066264747,
|
| 398 |
+
"eval_runtime": 301.2031,
|
| 399 |
+
"eval_samples_per_second": 106.503,
|
| 400 |
+
"eval_sleeve_macro_f1": 0.3797203601033849,
|
| 401 |
+
"eval_sleeve_precision": 0.8450184667180418,
|
| 402 |
+
"eval_sleeve_recall": 0.8539543003210823,
|
| 403 |
+
"eval_sleeve_weighted_f1": 0.8447728073486299,
|
| 404 |
+
"eval_steps_per_second": 1.667,
|
| 405 |
+
"eval_style_macro_f1": 0.25776208112470533,
|
| 406 |
+
"eval_style_precision": 0.636768112985809,
|
| 407 |
+
"eval_style_recall": 0.6687552604507622,
|
| 408 |
+
"eval_style_weighted_f1": 0.6370942948799759,
|
| 409 |
+
"step": 16008
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 4.122938530734633,
|
| 413 |
+
"grad_norm": 1.4649860858917236,
|
| 414 |
+
"learning_rate": 2.0923893290688377e-05,
|
| 415 |
+
"loss": 0.1968,
|
| 416 |
+
"step": 16500
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 4.2478760619690155,
|
| 420 |
+
"grad_norm": 1.7522532939910889,
|
| 421 |
+
"learning_rate": 2.001526495602239e-05,
|
| 422 |
+
"loss": 0.1952,
|
| 423 |
+
"step": 17000
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 4.372813593203398,
|
| 427 |
+
"grad_norm": 1.5114365816116333,
|
| 428 |
+
"learning_rate": 1.9106636621356403e-05,
|
| 429 |
+
"loss": 0.198,
|
| 430 |
+
"step": 17500
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 4.497751124437781,
|
| 434 |
+
"grad_norm": 1.7840216159820557,
|
| 435 |
+
"learning_rate": 1.8198008286690413e-05,
|
| 436 |
+
"loss": 0.1954,
|
| 437 |
+
"step": 18000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 4.622688655672164,
|
| 441 |
+
"grad_norm": 1.5770485401153564,
|
| 442 |
+
"learning_rate": 1.7289379952024426e-05,
|
| 443 |
+
"loss": 0.1927,
|
| 444 |
+
"step": 18500
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 4.747626186906547,
|
| 448 |
+
"grad_norm": 1.5942214727401733,
|
| 449 |
+
"learning_rate": 1.6380751617358436e-05,
|
| 450 |
+
"loss": 0.1936,
|
| 451 |
+
"step": 19000
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 4.872563718140929,
|
| 455 |
+
"grad_norm": 1.1230299472808838,
|
| 456 |
+
"learning_rate": 1.547212328269245e-05,
|
| 457 |
+
"loss": 0.1907,
|
| 458 |
+
"step": 19500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 4.997501249375312,
|
| 462 |
+
"grad_norm": 1.336006999015808,
|
| 463 |
+
"learning_rate": 1.456349494802646e-05,
|
| 464 |
+
"loss": 0.1921,
|
| 465 |
+
"step": 20000
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 5.0,
|
| 469 |
+
"eval_category_macro_f1": 0.814575381252187,
|
| 470 |
+
"eval_category_precision": 0.9663094811768377,
|
| 471 |
+
"eval_category_recall": 0.9665824994544717,
|
| 472 |
+
"eval_category_weighted_f1": 0.965923844387987,
|
| 473 |
+
"eval_color_macro_f1": 0.4369174781614063,
|
| 474 |
+
"eval_color_precision": 0.8028535608404194,
|
| 475 |
+
"eval_color_recall": 0.8148321331712335,
|
| 476 |
+
"eval_color_weighted_f1": 0.8028476456898314,
|
| 477 |
+
"eval_gender_macro_f1": 0.7053936478562223,
|
| 478 |
+
"eval_gender_precision": 0.9416777760492416,
|
| 479 |
+
"eval_gender_recall": 0.9407712210480377,
|
| 480 |
+
"eval_gender_weighted_f1": 0.9403681152060941,
|
| 481 |
+
"eval_loss": 0.23490393161773682,
|
| 482 |
+
"eval_material_macro_f1": 0.40133288032222664,
|
| 483 |
+
"eval_material_precision": 0.6313613296699556,
|
| 484 |
+
"eval_material_recall": 0.6625206521400293,
|
| 485 |
+
"eval_material_weighted_f1": 0.6270470446515838,
|
| 486 |
+
"eval_neck_macro_f1": 0.2814408126866265,
|
| 487 |
+
"eval_neck_precision": 0.8008124167021559,
|
| 488 |
+
"eval_neck_recall": 0.8062907197855295,
|
| 489 |
+
"eval_neck_weighted_f1": 0.7975722055581634,
|
| 490 |
+
"eval_pattern_macro_f1": 0.22699854595816307,
|
| 491 |
+
"eval_pattern_precision": 0.6042854492310323,
|
| 492 |
+
"eval_pattern_recall": 0.6455001714517286,
|
| 493 |
+
"eval_pattern_weighted_f1": 0.5909107309557631,
|
| 494 |
+
"eval_product_type_macro_f1": 0.7113036346548861,
|
| 495 |
+
"eval_product_type_precision": 0.7388496345969355,
|
| 496 |
+
"eval_product_type_recall": 0.7295426914804077,
|
| 497 |
+
"eval_product_type_weighted_f1": 0.726666644027167,
|
| 498 |
+
"eval_runtime": 301.3794,
|
| 499 |
+
"eval_samples_per_second": 106.441,
|
| 500 |
+
"eval_sleeve_macro_f1": 0.38500573083937306,
|
| 501 |
+
"eval_sleeve_precision": 0.8532874153559135,
|
| 502 |
+
"eval_sleeve_recall": 0.8566040088531438,
|
| 503 |
+
"eval_sleeve_weighted_f1": 0.8483550688145379,
|
| 504 |
+
"eval_steps_per_second": 1.666,
|
| 505 |
+
"eval_style_macro_f1": 0.2759097194358481,
|
| 506 |
+
"eval_style_precision": 0.636595640254917,
|
| 507 |
+
"eval_style_recall": 0.6738676392655631,
|
| 508 |
+
"eval_style_weighted_f1": 0.6414452136845251,
|
| 509 |
+
"step": 20010
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 5.122438780609695,
|
| 513 |
+
"grad_norm": 1.5047273635864258,
|
| 514 |
+
"learning_rate": 1.3654866613360473e-05,
|
| 515 |
+
"loss": 0.1717,
|
| 516 |
+
"step": 20500
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 5.247376311844078,
|
| 520 |
+
"grad_norm": 1.4382556676864624,
|
| 521 |
+
"learning_rate": 1.2746238278694484e-05,
|
| 522 |
+
"loss": 0.1693,
|
| 523 |
+
"step": 21000
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 5.3723138430784605,
|
| 527 |
+
"grad_norm": 1.5730925798416138,
|
| 528 |
+
"learning_rate": 1.1837609944028495e-05,
|
| 529 |
+
"loss": 0.1724,
|
| 530 |
+
"step": 21500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 5.497251374312843,
|
| 534 |
+
"grad_norm": 1.8472751379013062,
|
| 535 |
+
"learning_rate": 1.0928981609362507e-05,
|
| 536 |
+
"loss": 0.1721,
|
| 537 |
+
"step": 22000
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 5.622188905547226,
|
| 541 |
+
"grad_norm": 1.3081119060516357,
|
| 542 |
+
"learning_rate": 1.0020353274696518e-05,
|
| 543 |
+
"loss": 0.1694,
|
| 544 |
+
"step": 22500
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 5.747126436781609,
|
| 548 |
+
"grad_norm": 1.9150363206863403,
|
| 549 |
+
"learning_rate": 9.111724940030531e-06,
|
| 550 |
+
"loss": 0.1701,
|
| 551 |
+
"step": 23000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 5.872063968015992,
|
| 555 |
+
"grad_norm": 1.6012868881225586,
|
| 556 |
+
"learning_rate": 8.203096605364542e-06,
|
| 557 |
+
"loss": 0.1692,
|
| 558 |
+
"step": 23500
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 5.997001499250375,
|
| 562 |
+
"grad_norm": 2.1426620483398438,
|
| 563 |
+
"learning_rate": 7.294468270698554e-06,
|
| 564 |
+
"loss": 0.1663,
|
| 565 |
+
"step": 24000
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 6.0,
|
| 569 |
+
"eval_category_macro_f1": 0.8135808160841039,
|
| 570 |
+
"eval_category_precision": 0.9662816427280644,
|
| 571 |
+
"eval_category_recall": 0.9666136724960255,
|
| 572 |
+
"eval_category_weighted_f1": 0.9661077074670175,
|
| 573 |
+
"eval_color_macro_f1": 0.4600635003318152,
|
| 574 |
+
"eval_color_precision": 0.8112357172847174,
|
| 575 |
+
"eval_color_recall": 0.8206926649833224,
|
| 576 |
+
"eval_color_weighted_f1": 0.809040612878886,
|
| 577 |
+
"eval_gender_macro_f1": 0.7077574180619189,
|
| 578 |
+
"eval_gender_precision": 0.9483675936419935,
|
| 579 |
+
"eval_gender_recall": 0.9483150971040244,
|
| 580 |
+
"eval_gender_weighted_f1": 0.9479706364362375,
|
| 581 |
+
"eval_loss": 0.22788158059120178,
|
| 582 |
+
"eval_material_macro_f1": 0.41327594900590064,
|
| 583 |
+
"eval_material_precision": 0.6451838446086702,
|
| 584 |
+
"eval_material_recall": 0.6689422987000841,
|
| 585 |
+
"eval_material_weighted_f1": 0.6426236021542818,
|
| 586 |
+
"eval_neck_macro_f1": 0.29755762158064153,
|
| 587 |
+
"eval_neck_precision": 0.8025765460608351,
|
| 588 |
+
"eval_neck_recall": 0.8098444465226472,
|
| 589 |
+
"eval_neck_weighted_f1": 0.8028234665625918,
|
| 590 |
+
"eval_pattern_macro_f1": 0.25250915316096467,
|
| 591 |
+
"eval_pattern_precision": 0.6176590799884946,
|
| 592 |
+
"eval_pattern_recall": 0.6519218180117834,
|
| 593 |
+
"eval_pattern_weighted_f1": 0.601571471109787,
|
| 594 |
+
"eval_product_type_macro_f1": 0.7308082280687357,
|
| 595 |
+
"eval_product_type_precision": 0.7450374918864295,
|
| 596 |
+
"eval_product_type_recall": 0.7416690046447831,
|
| 597 |
+
"eval_product_type_weighted_f1": 0.7391305618836747,
|
| 598 |
+
"eval_runtime": 301.8571,
|
| 599 |
+
"eval_samples_per_second": 106.272,
|
| 600 |
+
"eval_sleeve_macro_f1": 0.40274936834626424,
|
| 601 |
+
"eval_sleeve_precision": 0.855686478933816,
|
| 602 |
+
"eval_sleeve_recall": 0.8593472365098662,
|
| 603 |
+
"eval_sleeve_weighted_f1": 0.851732539861904,
|
| 604 |
+
"eval_steps_per_second": 1.663,
|
| 605 |
+
"eval_style_macro_f1": 0.2996676235011164,
|
| 606 |
+
"eval_style_precision": 0.6484348800461253,
|
| 607 |
+
"eval_style_recall": 0.6805386701580474,
|
| 608 |
+
"eval_style_weighted_f1": 0.6505079281930336,
|
| 609 |
+
"step": 24012
|
| 610 |
+
}
|
| 611 |
+
],
|
| 612 |
+
"logging_steps": 500,
|
| 613 |
+
"max_steps": 28014,
|
| 614 |
+
"num_input_tokens_seen": 0,
|
| 615 |
+
"num_train_epochs": 7,
|
| 616 |
+
"save_steps": 500,
|
| 617 |
+
"stateful_callbacks": {
|
| 618 |
+
"EarlyStoppingCallback": {
|
| 619 |
+
"args": {
|
| 620 |
+
"early_stopping_patience": 2,
|
| 621 |
+
"early_stopping_threshold": 0.0
|
| 622 |
+
},
|
| 623 |
+
"attributes": {
|
| 624 |
+
"early_stopping_patience_counter": 0
|
| 625 |
+
}
|
| 626 |
+
},
|
| 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": 32,
|
| 640 |
+
"trial_name": null,
|
| 641 |
+
"trial_params": null
|
| 642 |
+
}
|
checkpoint-24012/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bfd16e73b326474644ca8ee6974df45fabbdd7162dc44cee9d339f980bcea75f
|
| 3 |
+
size 5368
|