Upload folder using huggingface_hub
Browse files- config.json +7 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +742 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"MMoEViT"
|
| 4 |
+
],
|
| 5 |
+
"torch_dtype": "float32",
|
| 6 |
+
"transformers_version": "4.47.0"
|
| 7 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:06c46b31a595c9e3cacc33ae9ff0d04745cee24796af58038220f67a5fa33c36
|
| 3 |
+
size 359190616
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2582a72cfe28b6154411dfdf8976e7882bbd5caf3fa407d4a12c3bf9a2eeaeb6
|
| 3 |
+
size 718509882
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9ecdb2aab127fdd4a50e882dca2103373b1eb980735e992800a6e69e40807b5c
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:53c92f011cd3caa037f9d6a8e340e59b8a7fe10331eb357c4ea4d3f3cdba14fb
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,742 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.967008960660055,
|
| 3 |
+
"best_model_checkpoint": "/kaggle/working/mmoe_vit_results/checkpoint-24012",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 28014,
|
| 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": 57.01617431640625,
|
| 14 |
+
"learning_rate": 5e-05,
|
| 15 |
+
"loss": 16.5766,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.24987506246876562,
|
| 20 |
+
"grad_norm": 36.94090270996094,
|
| 21 |
+
"learning_rate": 4.9367408906882594e-05,
|
| 22 |
+
"loss": 10.9165,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3748125937031484,
|
| 27 |
+
"grad_norm": 30.66377067565918,
|
| 28 |
+
"learning_rate": 4.8734817813765186e-05,
|
| 29 |
+
"loss": 9.8167,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.49975012493753124,
|
| 34 |
+
"grad_norm": 42.035491943359375,
|
| 35 |
+
"learning_rate": 4.810222672064778e-05,
|
| 36 |
+
"loss": 9.3138,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.624687656171914,
|
| 41 |
+
"grad_norm": 51.97323989868164,
|
| 42 |
+
"learning_rate": 4.746963562753037e-05,
|
| 43 |
+
"loss": 8.8995,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.7496251874062968,
|
| 48 |
+
"grad_norm": 30.12773895263672,
|
| 49 |
+
"learning_rate": 4.683704453441296e-05,
|
| 50 |
+
"loss": 8.6501,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.8745627186406797,
|
| 55 |
+
"grad_norm": 35.8332405090332,
|
| 56 |
+
"learning_rate": 4.6204453441295545e-05,
|
| 57 |
+
"loss": 8.4094,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.9995002498750625,
|
| 62 |
+
"grad_norm": 33.1800422668457,
|
| 63 |
+
"learning_rate": 4.557186234817814e-05,
|
| 64 |
+
"loss": 8.2048,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.0,
|
| 69 |
+
"eval_category_macro_f1": 0.7056917901689866,
|
| 70 |
+
"eval_category_precision": 0.959686334325201,
|
| 71 |
+
"eval_category_recall": 0.9583528164843044,
|
| 72 |
+
"eval_category_weighted_f1": 0.9569810583065036,
|
| 73 |
+
"eval_color_macro_f1": 0.2688303275737633,
|
| 74 |
+
"eval_color_precision": 0.7518445947091915,
|
| 75 |
+
"eval_color_recall": 0.7785778858443219,
|
| 76 |
+
"eval_color_weighted_f1": 0.7567459270340884,
|
| 77 |
+
"eval_gender_macro_f1": 0.61691692266805,
|
| 78 |
+
"eval_gender_precision": 0.924795901785245,
|
| 79 |
+
"eval_gender_recall": 0.9251535272296518,
|
| 80 |
+
"eval_gender_weighted_f1": 0.9247660825888746,
|
| 81 |
+
"eval_loss": 7.85603666305542,
|
| 82 |
+
"eval_material_macro_f1": 0.20719595562175594,
|
| 83 |
+
"eval_material_precision": 0.5500768038697358,
|
| 84 |
+
"eval_material_recall": 0.6039153340191402,
|
| 85 |
+
"eval_material_weighted_f1": 0.5466956697100748,
|
| 86 |
+
"eval_neck_macro_f1": 0.19113506916293588,
|
| 87 |
+
"eval_neck_precision": 0.7622121166320025,
|
| 88 |
+
"eval_neck_recall": 0.7717509897440693,
|
| 89 |
+
"eval_neck_weighted_f1": 0.7581205711853191,
|
| 90 |
+
"eval_pattern_macro_f1": 0.07869956737005358,
|
| 91 |
+
"eval_pattern_precision": 0.5012835314156145,
|
| 92 |
+
"eval_pattern_recall": 0.5952804015087753,
|
| 93 |
+
"eval_pattern_weighted_f1": 0.5053200171793594,
|
| 94 |
+
"eval_product_type_macro_f1": 0.5397272380061237,
|
| 95 |
+
"eval_product_type_precision": 0.659065702055643,
|
| 96 |
+
"eval_product_type_recall": 0.6470899965709654,
|
| 97 |
+
"eval_product_type_weighted_f1": 0.6335857444723578,
|
| 98 |
+
"eval_runtime": 319.4683,
|
| 99 |
+
"eval_samples_per_second": 100.414,
|
| 100 |
+
"eval_sleeve_macro_f1": 0.32806975143028533,
|
| 101 |
+
"eval_sleeve_precision": 0.8325346017171937,
|
| 102 |
+
"eval_sleeve_recall": 0.8443530035225537,
|
| 103 |
+
"eval_sleeve_weighted_f1": 0.8322418774164811,
|
| 104 |
+
"eval_steps_per_second": 1.571,
|
| 105 |
+
"eval_style_macro_f1": 0.14638281117715193,
|
| 106 |
+
"eval_style_precision": 0.5779767242059484,
|
| 107 |
+
"eval_style_recall": 0.6343402225755167,
|
| 108 |
+
"eval_style_weighted_f1": 0.5854913490611046,
|
| 109 |
+
"step": 4002
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.1244377811094453,
|
| 113 |
+
"grad_norm": 29.43435287475586,
|
| 114 |
+
"learning_rate": 4.4939271255060735e-05,
|
| 115 |
+
"loss": 7.7873,
|
| 116 |
+
"step": 4500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.249375312343828,
|
| 120 |
+
"grad_norm": 36.80842971801758,
|
| 121 |
+
"learning_rate": 4.430668016194332e-05,
|
| 122 |
+
"loss": 7.7689,
|
| 123 |
+
"step": 5000
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 1.3743128435782108,
|
| 127 |
+
"grad_norm": 29.0100040435791,
|
| 128 |
+
"learning_rate": 4.367408906882591e-05,
|
| 129 |
+
"loss": 7.6509,
|
| 130 |
+
"step": 5500
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.4992503748125938,
|
| 134 |
+
"grad_norm": 43.44118118286133,
|
| 135 |
+
"learning_rate": 4.3041497975708504e-05,
|
| 136 |
+
"loss": 7.6131,
|
| 137 |
+
"step": 6000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.6241879060469766,
|
| 141 |
+
"grad_norm": 27.290224075317383,
|
| 142 |
+
"learning_rate": 4.2408906882591095e-05,
|
| 143 |
+
"loss": 7.4747,
|
| 144 |
+
"step": 6500
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.7491254372813594,
|
| 148 |
+
"grad_norm": 32.5240364074707,
|
| 149 |
+
"learning_rate": 4.177631578947369e-05,
|
| 150 |
+
"loss": 7.307,
|
| 151 |
+
"step": 7000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.8740629685157422,
|
| 155 |
+
"grad_norm": 30.75071907043457,
|
| 156 |
+
"learning_rate": 4.114372469635628e-05,
|
| 157 |
+
"loss": 7.2819,
|
| 158 |
+
"step": 7500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.999000499750125,
|
| 162 |
+
"grad_norm": 38.38020706176758,
|
| 163 |
+
"learning_rate": 4.051113360323887e-05,
|
| 164 |
+
"loss": 7.2531,
|
| 165 |
+
"step": 8000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.0,
|
| 169 |
+
"eval_category_macro_f1": 0.7602602884544463,
|
| 170 |
+
"eval_category_precision": 0.9629087982936062,
|
| 171 |
+
"eval_category_recall": 0.9625300040524954,
|
| 172 |
+
"eval_category_weighted_f1": 0.9614384751352656,
|
| 173 |
+
"eval_color_macro_f1": 0.32999615687202,
|
| 174 |
+
"eval_color_precision": 0.7657516039866678,
|
| 175 |
+
"eval_color_recall": 0.7893637582218898,
|
| 176 |
+
"eval_color_weighted_f1": 0.7702899342050024,
|
| 177 |
+
"eval_gender_macro_f1": 0.65351315015835,
|
| 178 |
+
"eval_gender_precision": 0.9344584897700611,
|
| 179 |
+
"eval_gender_recall": 0.9329779606596216,
|
| 180 |
+
"eval_gender_weighted_f1": 0.9321715496558077,
|
| 181 |
+
"eval_loss": 7.258897304534912,
|
| 182 |
+
"eval_material_macro_f1": 0.2746169076986001,
|
| 183 |
+
"eval_material_precision": 0.5915997907196028,
|
| 184 |
+
"eval_material_recall": 0.6195953739206335,
|
| 185 |
+
"eval_material_weighted_f1": 0.5668523075222497,
|
| 186 |
+
"eval_neck_macro_f1": 0.20049848494224712,
|
| 187 |
+
"eval_neck_precision": 0.7708207481094091,
|
| 188 |
+
"eval_neck_recall": 0.7876492409364382,
|
| 189 |
+
"eval_neck_weighted_f1": 0.7684817623431263,
|
| 190 |
+
"eval_pattern_macro_f1": 0.11355145063472773,
|
| 191 |
+
"eval_pattern_precision": 0.5245684834877903,
|
| 192 |
+
"eval_pattern_recall": 0.6085289441690825,
|
| 193 |
+
"eval_pattern_weighted_f1": 0.5261034858766623,
|
| 194 |
+
"eval_product_type_macro_f1": 0.613166417428187,
|
| 195 |
+
"eval_product_type_precision": 0.6931319121735884,
|
| 196 |
+
"eval_product_type_recall": 0.6792917484959008,
|
| 197 |
+
"eval_product_type_weighted_f1": 0.6700524758340307,
|
| 198 |
+
"eval_runtime": 305.4934,
|
| 199 |
+
"eval_samples_per_second": 105.007,
|
| 200 |
+
"eval_sleeve_macro_f1": 0.3328543256219435,
|
| 201 |
+
"eval_sleeve_precision": 0.8471446159191267,
|
| 202 |
+
"eval_sleeve_recall": 0.8456622712678076,
|
| 203 |
+
"eval_sleeve_weighted_f1": 0.8332191903962787,
|
| 204 |
+
"eval_steps_per_second": 1.643,
|
| 205 |
+
"eval_style_macro_f1": 0.19206324944658193,
|
| 206 |
+
"eval_style_precision": 0.6055367377975819,
|
| 207 |
+
"eval_style_recall": 0.6526699710090713,
|
| 208 |
+
"eval_style_weighted_f1": 0.6042691229527709,
|
| 209 |
+
"step": 8004
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.1239380309845077,
|
| 213 |
+
"grad_norm": 34.519927978515625,
|
| 214 |
+
"learning_rate": 3.9878542510121455e-05,
|
| 215 |
+
"loss": 6.7502,
|
| 216 |
+
"step": 8500
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.2488755622188905,
|
| 220 |
+
"grad_norm": 34.661766052246094,
|
| 221 |
+
"learning_rate": 3.924595141700405e-05,
|
| 222 |
+
"loss": 6.8356,
|
| 223 |
+
"step": 9000
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 2.3738130934532733,
|
| 227 |
+
"grad_norm": 33.54723358154297,
|
| 228 |
+
"learning_rate": 3.8613360323886645e-05,
|
| 229 |
+
"loss": 6.6994,
|
| 230 |
+
"step": 9500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.498750624687656,
|
| 234 |
+
"grad_norm": 36.0948486328125,
|
| 235 |
+
"learning_rate": 3.798076923076923e-05,
|
| 236 |
+
"loss": 6.6972,
|
| 237 |
+
"step": 10000
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 2.623688155922039,
|
| 241 |
+
"grad_norm": 36.17250442504883,
|
| 242 |
+
"learning_rate": 3.734817813765182e-05,
|
| 243 |
+
"loss": 6.6196,
|
| 244 |
+
"step": 10500
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 2.7486256871564216,
|
| 248 |
+
"grad_norm": 34.5025634765625,
|
| 249 |
+
"learning_rate": 3.671558704453441e-05,
|
| 250 |
+
"loss": 6.6215,
|
| 251 |
+
"step": 11000
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 2.873563218390805,
|
| 255 |
+
"grad_norm": 33.60887908935547,
|
| 256 |
+
"learning_rate": 3.6082995951417005e-05,
|
| 257 |
+
"loss": 6.5135,
|
| 258 |
+
"step": 11500
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.9985007496251876,
|
| 262 |
+
"grad_norm": 34.10890197753906,
|
| 263 |
+
"learning_rate": 3.5450404858299596e-05,
|
| 264 |
+
"loss": 6.5109,
|
| 265 |
+
"step": 12000
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 3.0,
|
| 269 |
+
"eval_category_macro_f1": 0.787512350264236,
|
| 270 |
+
"eval_category_precision": 0.9635144196428028,
|
| 271 |
+
"eval_category_recall": 0.9634028492159981,
|
| 272 |
+
"eval_category_weighted_f1": 0.9624150257086617,
|
| 273 |
+
"eval_color_macro_f1": 0.34627790250554974,
|
| 274 |
+
"eval_color_precision": 0.7784139120798733,
|
| 275 |
+
"eval_color_recall": 0.7970634994856448,
|
| 276 |
+
"eval_color_weighted_f1": 0.7762200635616149,
|
| 277 |
+
"eval_gender_macro_f1": 0.6754785001926855,
|
| 278 |
+
"eval_gender_precision": 0.9226440250759032,
|
| 279 |
+
"eval_gender_recall": 0.9195112067084386,
|
| 280 |
+
"eval_gender_weighted_f1": 0.9188428393913575,
|
| 281 |
+
"eval_loss": 6.8828935623168945,
|
| 282 |
+
"eval_material_macro_f1": 0.34439436651770383,
|
| 283 |
+
"eval_material_precision": 0.6044803466129843,
|
| 284 |
+
"eval_material_recall": 0.6432557124598647,
|
| 285 |
+
"eval_material_weighted_f1": 0.5987268283398588,
|
| 286 |
+
"eval_neck_macro_f1": 0.2636374805840866,
|
| 287 |
+
"eval_neck_precision": 0.7859040450058413,
|
| 288 |
+
"eval_neck_recall": 0.7974375759842888,
|
| 289 |
+
"eval_neck_weighted_f1": 0.785483280031206,
|
| 290 |
+
"eval_pattern_macro_f1": 0.15190625172777147,
|
| 291 |
+
"eval_pattern_precision": 0.5791524062798472,
|
| 292 |
+
"eval_pattern_recall": 0.6267340004364226,
|
| 293 |
+
"eval_pattern_weighted_f1": 0.5539934146060564,
|
| 294 |
+
"eval_product_type_macro_f1": 0.6638223544457101,
|
| 295 |
+
"eval_product_type_precision": 0.7147695152999568,
|
| 296 |
+
"eval_product_type_recall": 0.7016428192898782,
|
| 297 |
+
"eval_product_type_weighted_f1": 0.6952318435900471,
|
| 298 |
+
"eval_runtime": 300.8363,
|
| 299 |
+
"eval_samples_per_second": 106.633,
|
| 300 |
+
"eval_sleeve_macro_f1": 0.3475305353320075,
|
| 301 |
+
"eval_sleeve_precision": 0.8334833047421953,
|
| 302 |
+
"eval_sleeve_recall": 0.8485925371738521,
|
| 303 |
+
"eval_sleeve_weighted_f1": 0.8360923435643329,
|
| 304 |
+
"eval_steps_per_second": 1.669,
|
| 305 |
+
"eval_style_macro_f1": 0.22495804935114228,
|
| 306 |
+
"eval_style_precision": 0.622297675917607,
|
| 307 |
+
"eval_style_recall": 0.6627388634309049,
|
| 308 |
+
"eval_style_weighted_f1": 0.6259413091730935,
|
| 309 |
+
"step": 12006
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 3.1234382808595704,
|
| 313 |
+
"grad_norm": 37.55683517456055,
|
| 314 |
+
"learning_rate": 3.481781376518219e-05,
|
| 315 |
+
"loss": 6.0245,
|
| 316 |
+
"step": 12500
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 3.248375812093953,
|
| 320 |
+
"grad_norm": 31.214113235473633,
|
| 321 |
+
"learning_rate": 3.418522267206478e-05,
|
| 322 |
+
"loss": 6.0258,
|
| 323 |
+
"step": 13000
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 3.373313343328336,
|
| 327 |
+
"grad_norm": 42.683753967285156,
|
| 328 |
+
"learning_rate": 3.355263157894737e-05,
|
| 329 |
+
"loss": 6.0101,
|
| 330 |
+
"step": 13500
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 3.4982508745627188,
|
| 334 |
+
"grad_norm": 33.178009033203125,
|
| 335 |
+
"learning_rate": 3.292004048582996e-05,
|
| 336 |
+
"loss": 6.0655,
|
| 337 |
+
"step": 14000
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 3.6231884057971016,
|
| 341 |
+
"grad_norm": 33.86723709106445,
|
| 342 |
+
"learning_rate": 3.2287449392712554e-05,
|
| 343 |
+
"loss": 6.0188,
|
| 344 |
+
"step": 14500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 3.7481259370314843,
|
| 348 |
+
"grad_norm": 49.44440460205078,
|
| 349 |
+
"learning_rate": 3.165485829959514e-05,
|
| 350 |
+
"loss": 6.0043,
|
| 351 |
+
"step": 15000
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 3.873063468265867,
|
| 355 |
+
"grad_norm": 26.736482620239258,
|
| 356 |
+
"learning_rate": 3.102226720647773e-05,
|
| 357 |
+
"loss": 5.873,
|
| 358 |
+
"step": 15500
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 3.99800099950025,
|
| 362 |
+
"grad_norm": 33.90946578979492,
|
| 363 |
+
"learning_rate": 3.0389676113360326e-05,
|
| 364 |
+
"loss": 5.7982,
|
| 365 |
+
"step": 16000
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 4.0,
|
| 369 |
+
"eval_category_macro_f1": 0.7955569681071116,
|
| 370 |
+
"eval_category_precision": 0.9633120393270671,
|
| 371 |
+
"eval_category_recall": 0.9633405031328907,
|
| 372 |
+
"eval_category_weighted_f1": 0.962656221421562,
|
| 373 |
+
"eval_color_macro_f1": 0.39246945574897335,
|
| 374 |
+
"eval_color_precision": 0.7946069660631935,
|
| 375 |
+
"eval_color_recall": 0.8061036815362075,
|
| 376 |
+
"eval_color_weighted_f1": 0.7926817060026948,
|
| 377 |
+
"eval_gender_macro_f1": 0.7037573639471852,
|
| 378 |
+
"eval_gender_precision": 0.9465872847377934,
|
| 379 |
+
"eval_gender_recall": 0.9456342155304093,
|
| 380 |
+
"eval_gender_weighted_f1": 0.9454295788332787,
|
| 381 |
+
"eval_loss": 6.5368194580078125,
|
| 382 |
+
"eval_material_macro_f1": 0.38069446247038957,
|
| 383 |
+
"eval_material_precision": 0.6231463458041611,
|
| 384 |
+
"eval_material_recall": 0.6512671841391564,
|
| 385 |
+
"eval_material_weighted_f1": 0.6147786520859259,
|
| 386 |
+
"eval_neck_macro_f1": 0.27672490278496564,
|
| 387 |
+
"eval_neck_precision": 0.7914577242024867,
|
| 388 |
+
"eval_neck_recall": 0.8040774338352193,
|
| 389 |
+
"eval_neck_weighted_f1": 0.792155013664771,
|
| 390 |
+
"eval_pattern_macro_f1": 0.19187411544892444,
|
| 391 |
+
"eval_pattern_precision": 0.586085709618732,
|
| 392 |
+
"eval_pattern_recall": 0.6352130677390193,
|
| 393 |
+
"eval_pattern_weighted_f1": 0.5795812233806185,
|
| 394 |
+
"eval_product_type_macro_f1": 0.7027178432304306,
|
| 395 |
+
"eval_product_type_precision": 0.7296781379497822,
|
| 396 |
+
"eval_product_type_recall": 0.7236198135852115,
|
| 397 |
+
"eval_product_type_weighted_f1": 0.7179279185541192,
|
| 398 |
+
"eval_runtime": 302.0245,
|
| 399 |
+
"eval_samples_per_second": 106.213,
|
| 400 |
+
"eval_sleeve_macro_f1": 0.3677025507155895,
|
| 401 |
+
"eval_sleeve_precision": 0.8508002629167147,
|
| 402 |
+
"eval_sleeve_recall": 0.8566663549362511,
|
| 403 |
+
"eval_sleeve_weighted_f1": 0.8455598759505015,
|
| 404 |
+
"eval_steps_per_second": 1.662,
|
| 405 |
+
"eval_style_macro_f1": 0.2618223911719054,
|
| 406 |
+
"eval_style_precision": 0.6421653958993869,
|
| 407 |
+
"eval_style_recall": 0.6738988123071168,
|
| 408 |
+
"eval_style_weighted_f1": 0.6433547708269393,
|
| 409 |
+
"step": 16008
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 4.122938530734633,
|
| 413 |
+
"grad_norm": 40.41415023803711,
|
| 414 |
+
"learning_rate": 2.9757085020242914e-05,
|
| 415 |
+
"loss": 5.3993,
|
| 416 |
+
"step": 16500
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 4.2478760619690155,
|
| 420 |
+
"grad_norm": 35.530174255371094,
|
| 421 |
+
"learning_rate": 2.9124493927125506e-05,
|
| 422 |
+
"loss": 5.3668,
|
| 423 |
+
"step": 17000
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 4.372813593203398,
|
| 427 |
+
"grad_norm": 37.02978515625,
|
| 428 |
+
"learning_rate": 2.84919028340081e-05,
|
| 429 |
+
"loss": 5.4427,
|
| 430 |
+
"step": 17500
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 4.497751124437781,
|
| 434 |
+
"grad_norm": 31.028322219848633,
|
| 435 |
+
"learning_rate": 2.7859311740890692e-05,
|
| 436 |
+
"loss": 5.4199,
|
| 437 |
+
"step": 18000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 4.622688655672164,
|
| 441 |
+
"grad_norm": 39.566871643066406,
|
| 442 |
+
"learning_rate": 2.722672064777328e-05,
|
| 443 |
+
"loss": 5.3339,
|
| 444 |
+
"step": 18500
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 4.747626186906547,
|
| 448 |
+
"grad_norm": 29.115066528320312,
|
| 449 |
+
"learning_rate": 2.6594129554655872e-05,
|
| 450 |
+
"loss": 5.3469,
|
| 451 |
+
"step": 19000
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 4.872563718140929,
|
| 455 |
+
"grad_norm": 27.22176742553711,
|
| 456 |
+
"learning_rate": 2.5961538461538464e-05,
|
| 457 |
+
"loss": 5.3245,
|
| 458 |
+
"step": 19500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 4.997501249375312,
|
| 462 |
+
"grad_norm": 30.805620193481445,
|
| 463 |
+
"learning_rate": 2.5328947368421052e-05,
|
| 464 |
+
"loss": 5.3437,
|
| 465 |
+
"step": 20000
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 5.0,
|
| 469 |
+
"eval_category_macro_f1": 0.8085353223473223,
|
| 470 |
+
"eval_category_precision": 0.9670311680024095,
|
| 471 |
+
"eval_category_recall": 0.9672059602855451,
|
| 472 |
+
"eval_category_weighted_f1": 0.9663742766459245,
|
| 473 |
+
"eval_color_macro_f1": 0.4238649051693472,
|
| 474 |
+
"eval_color_precision": 0.7981474301781062,
|
| 475 |
+
"eval_color_recall": 0.8128994045949063,
|
| 476 |
+
"eval_color_weighted_f1": 0.8002174907155432,
|
| 477 |
+
"eval_gender_macro_f1": 0.7196717213524463,
|
| 478 |
+
"eval_gender_precision": 0.9467508045608919,
|
| 479 |
+
"eval_gender_recall": 0.9464447146108046,
|
| 480 |
+
"eval_gender_weighted_f1": 0.9462666634376086,
|
| 481 |
+
"eval_loss": 6.324069499969482,
|
| 482 |
+
"eval_material_macro_f1": 0.40655027085392736,
|
| 483 |
+
"eval_material_precision": 0.6409986821944779,
|
| 484 |
+
"eval_material_recall": 0.6640169581346052,
|
| 485 |
+
"eval_material_weighted_f1": 0.6282823649788557,
|
| 486 |
+
"eval_neck_macro_f1": 0.28417588679207595,
|
| 487 |
+
"eval_neck_precision": 0.8012300473656296,
|
| 488 |
+
"eval_neck_recall": 0.8069141806166028,
|
| 489 |
+
"eval_neck_weighted_f1": 0.7976287506977606,
|
| 490 |
+
"eval_pattern_macro_f1": 0.2187785306978239,
|
| 491 |
+
"eval_pattern_precision": 0.6129304703362072,
|
| 492 |
+
"eval_pattern_recall": 0.6472770348202874,
|
| 493 |
+
"eval_pattern_weighted_f1": 0.5881639092284188,
|
| 494 |
+
"eval_product_type_macro_f1": 0.7169662707860948,
|
| 495 |
+
"eval_product_type_precision": 0.739852886082883,
|
| 496 |
+
"eval_product_type_recall": 0.7304778827270176,
|
| 497 |
+
"eval_product_type_weighted_f1": 0.7280465570285056,
|
| 498 |
+
"eval_runtime": 300.8243,
|
| 499 |
+
"eval_samples_per_second": 106.637,
|
| 500 |
+
"eval_sleeve_macro_f1": 0.39617711979641707,
|
| 501 |
+
"eval_sleeve_precision": 0.8573303931972993,
|
| 502 |
+
"eval_sleeve_recall": 0.858069141806166,
|
| 503 |
+
"eval_sleeve_weighted_f1": 0.8493273999182868,
|
| 504 |
+
"eval_steps_per_second": 1.669,
|
| 505 |
+
"eval_style_macro_f1": 0.2753572234324515,
|
| 506 |
+
"eval_style_precision": 0.6415522634195322,
|
| 507 |
+
"eval_style_recall": 0.6796658249945448,
|
| 508 |
+
"eval_style_weighted_f1": 0.644725501006127,
|
| 509 |
+
"step": 20010
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 5.122438780609695,
|
| 513 |
+
"grad_norm": 32.560482025146484,
|
| 514 |
+
"learning_rate": 2.4696356275303644e-05,
|
| 515 |
+
"loss": 4.7955,
|
| 516 |
+
"step": 20500
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 5.247376311844078,
|
| 520 |
+
"grad_norm": 31.55764389038086,
|
| 521 |
+
"learning_rate": 2.4063765182186235e-05,
|
| 522 |
+
"loss": 4.7692,
|
| 523 |
+
"step": 21000
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 5.3723138430784605,
|
| 527 |
+
"grad_norm": 31.34699821472168,
|
| 528 |
+
"learning_rate": 2.3431174089068827e-05,
|
| 529 |
+
"loss": 4.8501,
|
| 530 |
+
"step": 21500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 5.497251374312843,
|
| 534 |
+
"grad_norm": 35.56719207763672,
|
| 535 |
+
"learning_rate": 2.279858299595142e-05,
|
| 536 |
+
"loss": 4.8357,
|
| 537 |
+
"step": 22000
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 5.622188905547226,
|
| 541 |
+
"grad_norm": 27.214921951293945,
|
| 542 |
+
"learning_rate": 2.216599190283401e-05,
|
| 543 |
+
"loss": 4.7804,
|
| 544 |
+
"step": 22500
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 5.747126436781609,
|
| 548 |
+
"grad_norm": 47.55564880371094,
|
| 549 |
+
"learning_rate": 2.15334008097166e-05,
|
| 550 |
+
"loss": 4.8164,
|
| 551 |
+
"step": 23000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 5.872063968015992,
|
| 555 |
+
"grad_norm": 27.112457275390625,
|
| 556 |
+
"learning_rate": 2.090080971659919e-05,
|
| 557 |
+
"loss": 4.8067,
|
| 558 |
+
"step": 23500
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 5.997001499250375,
|
| 562 |
+
"grad_norm": 37.71267318725586,
|
| 563 |
+
"learning_rate": 2.0268218623481782e-05,
|
| 564 |
+
"loss": 4.7547,
|
| 565 |
+
"step": 24000
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 6.0,
|
| 569 |
+
"eval_category_macro_f1": 0.8067492828396051,
|
| 570 |
+
"eval_category_precision": 0.9675417493566353,
|
| 571 |
+
"eval_category_recall": 0.967767075033511,
|
| 572 |
+
"eval_category_weighted_f1": 0.967008960660055,
|
| 573 |
+
"eval_color_macro_f1": 0.46170642529262285,
|
| 574 |
+
"eval_color_precision": 0.808572630866145,
|
| 575 |
+
"eval_color_recall": 0.8167025156644534,
|
| 576 |
+
"eval_color_weighted_f1": 0.8052156787210335,
|
| 577 |
+
"eval_gender_macro_f1": 0.7253099260026424,
|
| 578 |
+
"eval_gender_precision": 0.9516655675023045,
|
| 579 |
+
"eval_gender_recall": 0.9511206708438542,
|
| 580 |
+
"eval_gender_weighted_f1": 0.9509797608509165,
|
| 581 |
+
"eval_loss": 6.1924147605896,
|
| 582 |
+
"eval_material_macro_f1": 0.4418062098856238,
|
| 583 |
+
"eval_material_precision": 0.6532578684941821,
|
| 584 |
+
"eval_material_recall": 0.6747716574706194,
|
| 585 |
+
"eval_material_weighted_f1": 0.6468422079358449,
|
| 586 |
+
"eval_neck_macro_f1": 0.30263560541401846,
|
| 587 |
+
"eval_neck_precision": 0.8066569574286967,
|
| 588 |
+
"eval_neck_recall": 0.812556501137816,
|
| 589 |
+
"eval_neck_weighted_f1": 0.8053312212408406,
|
| 590 |
+
"eval_pattern_macro_f1": 0.24683911824062632,
|
| 591 |
+
"eval_pattern_precision": 0.6203592848139506,
|
| 592 |
+
"eval_pattern_recall": 0.6539792387543253,
|
| 593 |
+
"eval_pattern_weighted_f1": 0.6034023280202176,
|
| 594 |
+
"eval_product_type_macro_f1": 0.7286701507504901,
|
| 595 |
+
"eval_product_type_precision": 0.7477061999565924,
|
| 596 |
+
"eval_product_type_recall": 0.7434146949717884,
|
| 597 |
+
"eval_product_type_weighted_f1": 0.7404621958524741,
|
| 598 |
+
"eval_runtime": 301.5697,
|
| 599 |
+
"eval_samples_per_second": 106.373,
|
| 600 |
+
"eval_sleeve_macro_f1": 0.4149438597919067,
|
| 601 |
+
"eval_sleeve_precision": 0.8619098226901072,
|
| 602 |
+
"eval_sleeve_recall": 0.8607500233797811,
|
| 603 |
+
"eval_sleeve_weighted_f1": 0.8529440106549326,
|
| 604 |
+
"eval_steps_per_second": 1.665,
|
| 605 |
+
"eval_style_macro_f1": 0.28852599697346254,
|
| 606 |
+
"eval_style_precision": 0.6561689572939597,
|
| 607 |
+
"eval_style_recall": 0.6877396427569438,
|
| 608 |
+
"eval_style_weighted_f1": 0.6590960855736121,
|
| 609 |
+
"step": 24012
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 6.121939030484757,
|
| 613 |
+
"grad_norm": 31.159828186035156,
|
| 614 |
+
"learning_rate": 1.9635627530364373e-05,
|
| 615 |
+
"loss": 4.2872,
|
| 616 |
+
"step": 24500
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 6.246876561719141,
|
| 620 |
+
"grad_norm": 36.643314361572266,
|
| 621 |
+
"learning_rate": 1.9003036437246965e-05,
|
| 622 |
+
"loss": 4.2826,
|
| 623 |
+
"step": 25000
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 6.371814092953524,
|
| 627 |
+
"grad_norm": 31.81490707397461,
|
| 628 |
+
"learning_rate": 1.8370445344129557e-05,
|
| 629 |
+
"loss": 4.2782,
|
| 630 |
+
"step": 25500
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 6.496751624187906,
|
| 634 |
+
"grad_norm": 41.419471740722656,
|
| 635 |
+
"learning_rate": 1.7737854251012145e-05,
|
| 636 |
+
"loss": 4.3151,
|
| 637 |
+
"step": 26000
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 6.621689155422289,
|
| 641 |
+
"grad_norm": 36.339202880859375,
|
| 642 |
+
"learning_rate": 1.7105263157894737e-05,
|
| 643 |
+
"loss": 4.2912,
|
| 644 |
+
"step": 26500
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 6.746626686656672,
|
| 648 |
+
"grad_norm": 35.872161865234375,
|
| 649 |
+
"learning_rate": 1.6472672064777328e-05,
|
| 650 |
+
"loss": 4.321,
|
| 651 |
+
"step": 27000
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 6.871564217891055,
|
| 655 |
+
"grad_norm": 31.54119873046875,
|
| 656 |
+
"learning_rate": 1.584008097165992e-05,
|
| 657 |
+
"loss": 4.2605,
|
| 658 |
+
"step": 27500
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 6.9965017491254375,
|
| 662 |
+
"grad_norm": 29.37108039855957,
|
| 663 |
+
"learning_rate": 1.5207489878542511e-05,
|
| 664 |
+
"loss": 4.1995,
|
| 665 |
+
"step": 28000
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 7.0,
|
| 669 |
+
"eval_category_macro_f1": 0.8019924698000542,
|
| 670 |
+
"eval_category_precision": 0.9653353385671121,
|
| 671 |
+
"eval_category_recall": 0.9655849621247545,
|
| 672 |
+
"eval_category_weighted_f1": 0.9651467292153788,
|
| 673 |
+
"eval_color_macro_f1": 0.4778177910015613,
|
| 674 |
+
"eval_color_precision": 0.8132908708208361,
|
| 675 |
+
"eval_color_recall": 0.821846067520808,
|
| 676 |
+
"eval_color_weighted_f1": 0.8108879503190392,
|
| 677 |
+
"eval_gender_macro_f1": 0.7304886482949747,
|
| 678 |
+
"eval_gender_precision": 0.9575237181327425,
|
| 679 |
+
"eval_gender_recall": 0.9568253374481748,
|
| 680 |
+
"eval_gender_weighted_f1": 0.9565473974994894,
|
| 681 |
+
"eval_loss": 6.094598770141602,
|
| 682 |
+
"eval_material_macro_f1": 0.4445061914881263,
|
| 683 |
+
"eval_material_precision": 0.659882151181023,
|
| 684 |
+
"eval_material_recall": 0.6815050344462109,
|
| 685 |
+
"eval_material_weighted_f1": 0.6551634350927689,
|
| 686 |
+
"eval_neck_macro_f1": 0.3312326690246815,
|
| 687 |
+
"eval_neck_precision": 0.8066317752184025,
|
| 688 |
+
"eval_neck_recall": 0.8143333645063748,
|
| 689 |
+
"eval_neck_weighted_f1": 0.8067876189718791,
|
| 690 |
+
"eval_pattern_macro_f1": 0.26541974190021195,
|
| 691 |
+
"eval_pattern_precision": 0.6297518110690976,
|
| 692 |
+
"eval_pattern_recall": 0.6604632313974874,
|
| 693 |
+
"eval_pattern_weighted_f1": 0.6108385812217497,
|
| 694 |
+
"eval_product_type_macro_f1": 0.7399005420683133,
|
| 695 |
+
"eval_product_type_precision": 0.7515415781015994,
|
| 696 |
+
"eval_product_type_recall": 0.7487141120359113,
|
| 697 |
+
"eval_product_type_weighted_f1": 0.7451143261096842,
|
| 698 |
+
"eval_runtime": 299.9165,
|
| 699 |
+
"eval_samples_per_second": 106.96,
|
| 700 |
+
"eval_sleeve_macro_f1": 0.43326486489655175,
|
| 701 |
+
"eval_sleeve_precision": 0.8584343914083422,
|
| 702 |
+
"eval_sleeve_recall": 0.8612799650861934,
|
| 703 |
+
"eval_sleeve_weighted_f1": 0.8542297807421783,
|
| 704 |
+
"eval_steps_per_second": 1.674,
|
| 705 |
+
"eval_style_macro_f1": 0.3162728761585937,
|
| 706 |
+
"eval_style_precision": 0.6621445575945633,
|
| 707 |
+
"eval_style_recall": 0.6907946008292029,
|
| 708 |
+
"eval_style_weighted_f1": 0.6607505081198878,
|
| 709 |
+
"step": 28014
|
| 710 |
+
}
|
| 711 |
+
],
|
| 712 |
+
"logging_steps": 500,
|
| 713 |
+
"max_steps": 40020,
|
| 714 |
+
"num_input_tokens_seen": 0,
|
| 715 |
+
"num_train_epochs": 10,
|
| 716 |
+
"save_steps": 500,
|
| 717 |
+
"stateful_callbacks": {
|
| 718 |
+
"EarlyStoppingCallback": {
|
| 719 |
+
"args": {
|
| 720 |
+
"early_stopping_patience": 3,
|
| 721 |
+
"early_stopping_threshold": 0.0
|
| 722 |
+
},
|
| 723 |
+
"attributes": {
|
| 724 |
+
"early_stopping_patience_counter": 1
|
| 725 |
+
}
|
| 726 |
+
},
|
| 727 |
+
"TrainerControl": {
|
| 728 |
+
"args": {
|
| 729 |
+
"should_epoch_stop": false,
|
| 730 |
+
"should_evaluate": false,
|
| 731 |
+
"should_log": false,
|
| 732 |
+
"should_save": true,
|
| 733 |
+
"should_training_stop": false
|
| 734 |
+
},
|
| 735 |
+
"attributes": {}
|
| 736 |
+
}
|
| 737 |
+
},
|
| 738 |
+
"total_flos": 0.0,
|
| 739 |
+
"train_batch_size": 32,
|
| 740 |
+
"trial_name": null,
|
| 741 |
+
"trial_params": null
|
| 742 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:605dc8dba1eb34fc917712a11a1287e988480ab3615e9e36153256c99cdb78cc
|
| 3 |
+
size 5368
|