Upload folder using huggingface_hub
Browse files- drawing/config.json +40 -0
- drawing/optimizer.pt +3 -0
- drawing/preprocessor_config.json +24 -0
- drawing/pytorch_model.bin +3 -0
- drawing/rng_state.pth +3 -0
- drawing/scheduler.pt +3 -0
- drawing/trainer_state.json +548 -0
- drawing/training_args.bin +3 -0
- edge_tip/config.json +40 -0
- edge_tip/optimizer.pt +3 -0
- edge_tip/preprocessor_config.json +24 -0
- edge_tip/pytorch_model.bin +3 -0
- edge_tip/rng_state.pth +3 -0
- edge_tip/scheduler.pt +3 -0
- edge_tip/trainer_state.json +184 -0
- edge_tip/training_args.bin +3 -0
- node/config.json +40 -0
- node/optimizer.pt +3 -0
- node/preprocessor_config.json +24 -0
- node/pytorch_model.bin +3 -0
- node/rng_state.pth +3 -0
- node/scheduler.pt +3 -0
- node/trainer_state.json +240 -0
- node/training_args.bin +3 -0
drawing/config.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "hustvl/yolos-base",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"YolosForObjectDetection"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"auxiliary_loss": false,
|
| 8 |
+
"bbox_cost": 5,
|
| 9 |
+
"bbox_loss_coefficient": 5,
|
| 10 |
+
"class_cost": 1,
|
| 11 |
+
"eos_coefficient": 0.1,
|
| 12 |
+
"giou_cost": 2,
|
| 13 |
+
"giou_loss_coefficient": 2,
|
| 14 |
+
"hidden_act": "gelu",
|
| 15 |
+
"hidden_dropout_prob": 0.0,
|
| 16 |
+
"hidden_size": 768,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "drawing"
|
| 19 |
+
},
|
| 20 |
+
"image_size": [
|
| 21 |
+
800,
|
| 22 |
+
1344
|
| 23 |
+
],
|
| 24 |
+
"initializer_range": 0.02,
|
| 25 |
+
"intermediate_size": 3072,
|
| 26 |
+
"label2id": {
|
| 27 |
+
"drawing": 0
|
| 28 |
+
},
|
| 29 |
+
"layer_norm_eps": 1e-12,
|
| 30 |
+
"model_type": "yolos",
|
| 31 |
+
"num_attention_heads": 12,
|
| 32 |
+
"num_channels": 3,
|
| 33 |
+
"num_detection_tokens": 100,
|
| 34 |
+
"num_hidden_layers": 12,
|
| 35 |
+
"patch_size": 16,
|
| 36 |
+
"qkv_bias": true,
|
| 37 |
+
"torch_dtype": "float32",
|
| 38 |
+
"transformers_version": "4.31.0",
|
| 39 |
+
"use_mid_position_embeddings": true
|
| 40 |
+
}
|
drawing/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1b24526c80855e6180a536f9ecf49f6cc03c867e67114360b5132835dd258068
|
| 3 |
+
size 1021957701
|
drawing/preprocessor_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_pad": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"format": "coco_detection",
|
| 7 |
+
"image_mean": [
|
| 8 |
+
0.485,
|
| 9 |
+
0.456,
|
| 10 |
+
0.406
|
| 11 |
+
],
|
| 12 |
+
"image_processor_type": "YolosImageProcessor",
|
| 13 |
+
"image_std": [
|
| 14 |
+
0.229,
|
| 15 |
+
0.224,
|
| 16 |
+
0.225
|
| 17 |
+
],
|
| 18 |
+
"resample": 2,
|
| 19 |
+
"rescale_factor": 0.00392156862745098,
|
| 20 |
+
"size": {
|
| 21 |
+
"longest_edge": 1333,
|
| 22 |
+
"shortest_edge": 1024
|
| 23 |
+
}
|
| 24 |
+
}
|
drawing/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3ab8f84f66f00c6e711f2ea059f262d58f20b94f1151f85dec0e20968cf8ec2
|
| 3 |
+
size 510988573
|
drawing/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d587abdc6a1e31d1bbc43b01ace013df177fc83177946c5aebb06dda6ba3f18f
|
| 3 |
+
size 14575
|
drawing/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2cd0bfd61c12013f588fd366fa15b1e6179589727616549da05345e8ee4e9bf1
|
| 3 |
+
size 627
|
drawing/trainer_state.json
ADDED
|
@@ -0,0 +1,548 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 73.07692307692308,
|
| 5 |
+
"global_step": 3800,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.92,
|
| 12 |
+
"learning_rate": 9.940384615384614e-07,
|
| 13 |
+
"loss": 2.1803,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.92,
|
| 18 |
+
"eval_loss": 1.6282658576965332,
|
| 19 |
+
"eval_runtime": 4.7479,
|
| 20 |
+
"eval_samples_per_second": 4.634,
|
| 21 |
+
"eval_steps_per_second": 0.632,
|
| 22 |
+
"step": 100
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 3.85,
|
| 26 |
+
"learning_rate": 9.87628205128205e-07,
|
| 27 |
+
"loss": 1.9556,
|
| 28 |
+
"step": 200
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 3.85,
|
| 32 |
+
"eval_loss": 1.6041302680969238,
|
| 33 |
+
"eval_runtime": 4.2335,
|
| 34 |
+
"eval_samples_per_second": 5.197,
|
| 35 |
+
"eval_steps_per_second": 0.709,
|
| 36 |
+
"step": 200
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 5.77,
|
| 40 |
+
"learning_rate": 9.812179487179486e-07,
|
| 41 |
+
"loss": 1.8415,
|
| 42 |
+
"step": 300
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 5.77,
|
| 46 |
+
"eval_loss": 1.5072983503341675,
|
| 47 |
+
"eval_runtime": 4.2682,
|
| 48 |
+
"eval_samples_per_second": 5.154,
|
| 49 |
+
"eval_steps_per_second": 0.703,
|
| 50 |
+
"step": 300
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 7.69,
|
| 54 |
+
"learning_rate": 9.748076923076923e-07,
|
| 55 |
+
"loss": 1.849,
|
| 56 |
+
"step": 400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 7.69,
|
| 60 |
+
"eval_loss": 1.5139409303665161,
|
| 61 |
+
"eval_runtime": 4.1966,
|
| 62 |
+
"eval_samples_per_second": 5.242,
|
| 63 |
+
"eval_steps_per_second": 0.715,
|
| 64 |
+
"step": 400
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 9.62,
|
| 68 |
+
"learning_rate": 9.683974358974359e-07,
|
| 69 |
+
"loss": 1.7466,
|
| 70 |
+
"step": 500
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 9.62,
|
| 74 |
+
"eval_loss": 1.4191316366195679,
|
| 75 |
+
"eval_runtime": 4.1991,
|
| 76 |
+
"eval_samples_per_second": 5.239,
|
| 77 |
+
"eval_steps_per_second": 0.714,
|
| 78 |
+
"step": 500
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 11.54,
|
| 82 |
+
"learning_rate": 9.62051282051282e-07,
|
| 83 |
+
"loss": 1.7443,
|
| 84 |
+
"step": 600
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 11.54,
|
| 88 |
+
"eval_loss": 1.4041415452957153,
|
| 89 |
+
"eval_runtime": 4.3488,
|
| 90 |
+
"eval_samples_per_second": 5.059,
|
| 91 |
+
"eval_steps_per_second": 0.69,
|
| 92 |
+
"step": 600
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 13.46,
|
| 96 |
+
"learning_rate": 9.556410256410255e-07,
|
| 97 |
+
"loss": 1.6662,
|
| 98 |
+
"step": 700
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 13.46,
|
| 102 |
+
"eval_loss": 1.3958640098571777,
|
| 103 |
+
"eval_runtime": 4.2134,
|
| 104 |
+
"eval_samples_per_second": 5.221,
|
| 105 |
+
"eval_steps_per_second": 0.712,
|
| 106 |
+
"step": 700
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 15.38,
|
| 110 |
+
"learning_rate": 9.492307692307691e-07,
|
| 111 |
+
"loss": 1.6388,
|
| 112 |
+
"step": 800
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 15.38,
|
| 116 |
+
"eval_loss": 1.364356517791748,
|
| 117 |
+
"eval_runtime": 4.4313,
|
| 118 |
+
"eval_samples_per_second": 4.965,
|
| 119 |
+
"eval_steps_per_second": 0.677,
|
| 120 |
+
"step": 800
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 17.31,
|
| 124 |
+
"learning_rate": 9.428205128205127e-07,
|
| 125 |
+
"loss": 1.5375,
|
| 126 |
+
"step": 900
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 17.31,
|
| 130 |
+
"eval_loss": 1.3546158075332642,
|
| 131 |
+
"eval_runtime": 4.2939,
|
| 132 |
+
"eval_samples_per_second": 5.124,
|
| 133 |
+
"eval_steps_per_second": 0.699,
|
| 134 |
+
"step": 900
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 19.23,
|
| 138 |
+
"learning_rate": 9.364102564102564e-07,
|
| 139 |
+
"loss": 1.7777,
|
| 140 |
+
"step": 1000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 19.23,
|
| 144 |
+
"eval_loss": 1.345603585243225,
|
| 145 |
+
"eval_runtime": 4.2959,
|
| 146 |
+
"eval_samples_per_second": 5.121,
|
| 147 |
+
"eval_steps_per_second": 0.698,
|
| 148 |
+
"step": 1000
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 21.15,
|
| 152 |
+
"learning_rate": 9.3e-07,
|
| 153 |
+
"loss": 1.562,
|
| 154 |
+
"step": 1100
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 21.15,
|
| 158 |
+
"eval_loss": 1.3052722215652466,
|
| 159 |
+
"eval_runtime": 4.3299,
|
| 160 |
+
"eval_samples_per_second": 5.081,
|
| 161 |
+
"eval_steps_per_second": 0.693,
|
| 162 |
+
"step": 1100
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 23.08,
|
| 166 |
+
"learning_rate": 9.235897435897435e-07,
|
| 167 |
+
"loss": 1.4714,
|
| 168 |
+
"step": 1200
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 23.08,
|
| 172 |
+
"eval_loss": 1.2963144779205322,
|
| 173 |
+
"eval_runtime": 4.2895,
|
| 174 |
+
"eval_samples_per_second": 5.129,
|
| 175 |
+
"eval_steps_per_second": 0.699,
|
| 176 |
+
"step": 1200
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 25.0,
|
| 180 |
+
"learning_rate": 9.172435897435897e-07,
|
| 181 |
+
"loss": 1.5037,
|
| 182 |
+
"step": 1300
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 25.0,
|
| 186 |
+
"eval_loss": 1.2800103425979614,
|
| 187 |
+
"eval_runtime": 4.2403,
|
| 188 |
+
"eval_samples_per_second": 5.188,
|
| 189 |
+
"eval_steps_per_second": 0.707,
|
| 190 |
+
"step": 1300
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 26.92,
|
| 194 |
+
"learning_rate": 9.108333333333333e-07,
|
| 195 |
+
"loss": 1.3721,
|
| 196 |
+
"step": 1400
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 26.92,
|
| 200 |
+
"eval_loss": 1.276384711265564,
|
| 201 |
+
"eval_runtime": 4.3022,
|
| 202 |
+
"eval_samples_per_second": 5.114,
|
| 203 |
+
"eval_steps_per_second": 0.697,
|
| 204 |
+
"step": 1400
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 28.85,
|
| 208 |
+
"learning_rate": 9.044230769230768e-07,
|
| 209 |
+
"loss": 1.4898,
|
| 210 |
+
"step": 1500
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 28.85,
|
| 214 |
+
"eval_loss": 1.29233980178833,
|
| 215 |
+
"eval_runtime": 4.2189,
|
| 216 |
+
"eval_samples_per_second": 5.215,
|
| 217 |
+
"eval_steps_per_second": 0.711,
|
| 218 |
+
"step": 1500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 30.77,
|
| 222 |
+
"learning_rate": 8.980128205128205e-07,
|
| 223 |
+
"loss": 1.4936,
|
| 224 |
+
"step": 1600
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 30.77,
|
| 228 |
+
"eval_loss": 1.2993799448013306,
|
| 229 |
+
"eval_runtime": 4.1785,
|
| 230 |
+
"eval_samples_per_second": 5.265,
|
| 231 |
+
"eval_steps_per_second": 0.718,
|
| 232 |
+
"step": 1600
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 32.69,
|
| 236 |
+
"learning_rate": 8.916025641025641e-07,
|
| 237 |
+
"loss": 1.4178,
|
| 238 |
+
"step": 1700
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 32.69,
|
| 242 |
+
"eval_loss": 1.253838062286377,
|
| 243 |
+
"eval_runtime": 4.2708,
|
| 244 |
+
"eval_samples_per_second": 5.151,
|
| 245 |
+
"eval_steps_per_second": 0.702,
|
| 246 |
+
"step": 1700
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 34.62,
|
| 250 |
+
"learning_rate": 8.851923076923077e-07,
|
| 251 |
+
"loss": 1.2665,
|
| 252 |
+
"step": 1800
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 34.62,
|
| 256 |
+
"eval_loss": 1.2696892023086548,
|
| 257 |
+
"eval_runtime": 4.2383,
|
| 258 |
+
"eval_samples_per_second": 5.191,
|
| 259 |
+
"eval_steps_per_second": 0.708,
|
| 260 |
+
"step": 1800
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 36.54,
|
| 264 |
+
"learning_rate": 8.787820512820513e-07,
|
| 265 |
+
"loss": 1.3008,
|
| 266 |
+
"step": 1900
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 36.54,
|
| 270 |
+
"eval_loss": 1.2657532691955566,
|
| 271 |
+
"eval_runtime": 4.3241,
|
| 272 |
+
"eval_samples_per_second": 5.088,
|
| 273 |
+
"eval_steps_per_second": 0.694,
|
| 274 |
+
"step": 1900
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 38.46,
|
| 278 |
+
"learning_rate": 8.723717948717948e-07,
|
| 279 |
+
"loss": 1.4281,
|
| 280 |
+
"step": 2000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 38.46,
|
| 284 |
+
"eval_loss": 1.2740833759307861,
|
| 285 |
+
"eval_runtime": 4.3835,
|
| 286 |
+
"eval_samples_per_second": 5.019,
|
| 287 |
+
"eval_steps_per_second": 0.684,
|
| 288 |
+
"step": 2000
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 40.38,
|
| 292 |
+
"learning_rate": 8.659615384615384e-07,
|
| 293 |
+
"loss": 1.2315,
|
| 294 |
+
"step": 2100
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 40.38,
|
| 298 |
+
"eval_loss": 1.2776740789413452,
|
| 299 |
+
"eval_runtime": 4.3172,
|
| 300 |
+
"eval_samples_per_second": 5.096,
|
| 301 |
+
"eval_steps_per_second": 0.695,
|
| 302 |
+
"step": 2100
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 42.31,
|
| 306 |
+
"learning_rate": 8.59551282051282e-07,
|
| 307 |
+
"loss": 1.2442,
|
| 308 |
+
"step": 2200
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 42.31,
|
| 312 |
+
"eval_loss": 1.2726629972457886,
|
| 313 |
+
"eval_runtime": 4.2859,
|
| 314 |
+
"eval_samples_per_second": 5.133,
|
| 315 |
+
"eval_steps_per_second": 0.7,
|
| 316 |
+
"step": 2200
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 44.23,
|
| 320 |
+
"learning_rate": 8.531410256410256e-07,
|
| 321 |
+
"loss": 1.2882,
|
| 322 |
+
"step": 2300
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 44.23,
|
| 326 |
+
"eval_loss": 1.2753421068191528,
|
| 327 |
+
"eval_runtime": 4.2108,
|
| 328 |
+
"eval_samples_per_second": 5.225,
|
| 329 |
+
"eval_steps_per_second": 0.712,
|
| 330 |
+
"step": 2300
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 46.15,
|
| 334 |
+
"learning_rate": 8.467307692307691e-07,
|
| 335 |
+
"loss": 1.194,
|
| 336 |
+
"step": 2400
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 46.15,
|
| 340 |
+
"eval_loss": 1.2462925910949707,
|
| 341 |
+
"eval_runtime": 4.1887,
|
| 342 |
+
"eval_samples_per_second": 5.252,
|
| 343 |
+
"eval_steps_per_second": 0.716,
|
| 344 |
+
"step": 2400
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 48.08,
|
| 348 |
+
"learning_rate": 8.403205128205128e-07,
|
| 349 |
+
"loss": 1.2223,
|
| 350 |
+
"step": 2500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 48.08,
|
| 354 |
+
"eval_loss": 1.2566279172897339,
|
| 355 |
+
"eval_runtime": 4.2232,
|
| 356 |
+
"eval_samples_per_second": 5.209,
|
| 357 |
+
"eval_steps_per_second": 0.71,
|
| 358 |
+
"step": 2500
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 50.0,
|
| 362 |
+
"learning_rate": 8.339102564102564e-07,
|
| 363 |
+
"loss": 1.1984,
|
| 364 |
+
"step": 2600
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 50.0,
|
| 368 |
+
"eval_loss": 1.2548397779464722,
|
| 369 |
+
"eval_runtime": 4.2193,
|
| 370 |
+
"eval_samples_per_second": 5.214,
|
| 371 |
+
"eval_steps_per_second": 0.711,
|
| 372 |
+
"step": 2600
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 51.92,
|
| 376 |
+
"learning_rate": 8.275e-07,
|
| 377 |
+
"loss": 1.1202,
|
| 378 |
+
"step": 2700
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 51.92,
|
| 382 |
+
"eval_loss": 1.2726110219955444,
|
| 383 |
+
"eval_runtime": 4.197,
|
| 384 |
+
"eval_samples_per_second": 5.242,
|
| 385 |
+
"eval_steps_per_second": 0.715,
|
| 386 |
+
"step": 2700
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 53.85,
|
| 390 |
+
"learning_rate": 8.210897435897435e-07,
|
| 391 |
+
"loss": 1.1128,
|
| 392 |
+
"step": 2800
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 53.85,
|
| 396 |
+
"eval_loss": 1.2732219696044922,
|
| 397 |
+
"eval_runtime": 4.189,
|
| 398 |
+
"eval_samples_per_second": 5.252,
|
| 399 |
+
"eval_steps_per_second": 0.716,
|
| 400 |
+
"step": 2800
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 55.77,
|
| 404 |
+
"learning_rate": 8.146794871794871e-07,
|
| 405 |
+
"loss": 1.1649,
|
| 406 |
+
"step": 2900
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 55.77,
|
| 410 |
+
"eval_loss": 1.2652595043182373,
|
| 411 |
+
"eval_runtime": 4.2028,
|
| 412 |
+
"eval_samples_per_second": 5.235,
|
| 413 |
+
"eval_steps_per_second": 0.714,
|
| 414 |
+
"step": 2900
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 57.69,
|
| 418 |
+
"learning_rate": 8.082692307692307e-07,
|
| 419 |
+
"loss": 1.1066,
|
| 420 |
+
"step": 3000
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 57.69,
|
| 424 |
+
"eval_loss": 1.254478931427002,
|
| 425 |
+
"eval_runtime": 4.2195,
|
| 426 |
+
"eval_samples_per_second": 5.214,
|
| 427 |
+
"eval_steps_per_second": 0.711,
|
| 428 |
+
"step": 3000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 59.62,
|
| 432 |
+
"learning_rate": 8.018589743589743e-07,
|
| 433 |
+
"loss": 1.0629,
|
| 434 |
+
"step": 3100
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 59.62,
|
| 438 |
+
"eval_loss": 1.2273532152175903,
|
| 439 |
+
"eval_runtime": 4.2147,
|
| 440 |
+
"eval_samples_per_second": 5.22,
|
| 441 |
+
"eval_steps_per_second": 0.712,
|
| 442 |
+
"step": 3100
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 61.54,
|
| 446 |
+
"learning_rate": 7.955128205128204e-07,
|
| 447 |
+
"loss": 1.093,
|
| 448 |
+
"step": 3200
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 61.54,
|
| 452 |
+
"eval_loss": 1.2593165636062622,
|
| 453 |
+
"eval_runtime": 4.2295,
|
| 454 |
+
"eval_samples_per_second": 5.202,
|
| 455 |
+
"eval_steps_per_second": 0.709,
|
| 456 |
+
"step": 3200
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 63.46,
|
| 460 |
+
"learning_rate": 7.89102564102564e-07,
|
| 461 |
+
"loss": 0.9813,
|
| 462 |
+
"step": 3300
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 63.46,
|
| 466 |
+
"eval_loss": 1.2721446752548218,
|
| 467 |
+
"eval_runtime": 4.1692,
|
| 468 |
+
"eval_samples_per_second": 5.277,
|
| 469 |
+
"eval_steps_per_second": 0.72,
|
| 470 |
+
"step": 3300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 65.38,
|
| 474 |
+
"learning_rate": 7.826923076923076e-07,
|
| 475 |
+
"loss": 1.0538,
|
| 476 |
+
"step": 3400
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 65.38,
|
| 480 |
+
"eval_loss": 1.2700778245925903,
|
| 481 |
+
"eval_runtime": 4.2208,
|
| 482 |
+
"eval_samples_per_second": 5.212,
|
| 483 |
+
"eval_steps_per_second": 0.711,
|
| 484 |
+
"step": 3400
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 67.31,
|
| 488 |
+
"learning_rate": 7.762820512820513e-07,
|
| 489 |
+
"loss": 0.9368,
|
| 490 |
+
"step": 3500
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 67.31,
|
| 494 |
+
"eval_loss": 1.2771656513214111,
|
| 495 |
+
"eval_runtime": 4.2547,
|
| 496 |
+
"eval_samples_per_second": 5.171,
|
| 497 |
+
"eval_steps_per_second": 0.705,
|
| 498 |
+
"step": 3500
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 69.23,
|
| 502 |
+
"learning_rate": 7.698717948717949e-07,
|
| 503 |
+
"loss": 1.0294,
|
| 504 |
+
"step": 3600
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 69.23,
|
| 508 |
+
"eval_loss": 1.2606873512268066,
|
| 509 |
+
"eval_runtime": 4.2609,
|
| 510 |
+
"eval_samples_per_second": 5.163,
|
| 511 |
+
"eval_steps_per_second": 0.704,
|
| 512 |
+
"step": 3600
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 71.15,
|
| 516 |
+
"learning_rate": 7.634615384615384e-07,
|
| 517 |
+
"loss": 0.9596,
|
| 518 |
+
"step": 3700
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 71.15,
|
| 522 |
+
"eval_loss": 1.2618690729141235,
|
| 523 |
+
"eval_runtime": 4.2669,
|
| 524 |
+
"eval_samples_per_second": 5.156,
|
| 525 |
+
"eval_steps_per_second": 0.703,
|
| 526 |
+
"step": 3700
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 73.08,
|
| 530 |
+
"learning_rate": 7.57051282051282e-07,
|
| 531 |
+
"loss": 0.9926,
|
| 532 |
+
"step": 3800
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 73.08,
|
| 536 |
+
"eval_loss": 1.2698349952697754,
|
| 537 |
+
"eval_runtime": 4.2523,
|
| 538 |
+
"eval_samples_per_second": 5.174,
|
| 539 |
+
"eval_steps_per_second": 0.705,
|
| 540 |
+
"step": 3800
|
| 541 |
+
}
|
| 542 |
+
],
|
| 543 |
+
"max_steps": 15600,
|
| 544 |
+
"num_train_epochs": 300,
|
| 545 |
+
"total_flos": 9.161067103204147e+18,
|
| 546 |
+
"trial_name": null,
|
| 547 |
+
"trial_params": null
|
| 548 |
+
}
|
drawing/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:31eb086f4f700a46b8a588c8dae7b104f2f5d21d8ee9b18e7c0c532a52b308a1
|
| 3 |
+
size 3899
|
edge_tip/config.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "hustvl/yolos-base",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"YolosForObjectDetection"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"auxiliary_loss": false,
|
| 8 |
+
"bbox_cost": 5,
|
| 9 |
+
"bbox_loss_coefficient": 5,
|
| 10 |
+
"class_cost": 1,
|
| 11 |
+
"eos_coefficient": 0.1,
|
| 12 |
+
"giou_cost": 2,
|
| 13 |
+
"giou_loss_coefficient": 2,
|
| 14 |
+
"hidden_act": "gelu",
|
| 15 |
+
"hidden_dropout_prob": 0.0,
|
| 16 |
+
"hidden_size": 768,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "drawing"
|
| 19 |
+
},
|
| 20 |
+
"image_size": [
|
| 21 |
+
800,
|
| 22 |
+
1344
|
| 23 |
+
],
|
| 24 |
+
"initializer_range": 0.02,
|
| 25 |
+
"intermediate_size": 3072,
|
| 26 |
+
"label2id": {
|
| 27 |
+
"drawing": 0
|
| 28 |
+
},
|
| 29 |
+
"layer_norm_eps": 1e-12,
|
| 30 |
+
"model_type": "yolos",
|
| 31 |
+
"num_attention_heads": 12,
|
| 32 |
+
"num_channels": 3,
|
| 33 |
+
"num_detection_tokens": 100,
|
| 34 |
+
"num_hidden_layers": 12,
|
| 35 |
+
"patch_size": 16,
|
| 36 |
+
"qkv_bias": true,
|
| 37 |
+
"torch_dtype": "float32",
|
| 38 |
+
"transformers_version": "4.31.0",
|
| 39 |
+
"use_mid_position_embeddings": true
|
| 40 |
+
}
|
edge_tip/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12caec9d47cb1b7d6acda5916bb61a9fd86eb7ec4cbc173ca17cd4df84f11890
|
| 3 |
+
size 1021957701
|
edge_tip/preprocessor_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_pad": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"format": "coco_detection",
|
| 7 |
+
"image_mean": [
|
| 8 |
+
0.485,
|
| 9 |
+
0.456,
|
| 10 |
+
0.406
|
| 11 |
+
],
|
| 12 |
+
"image_processor_type": "YolosImageProcessor",
|
| 13 |
+
"image_std": [
|
| 14 |
+
0.229,
|
| 15 |
+
0.224,
|
| 16 |
+
0.225
|
| 17 |
+
],
|
| 18 |
+
"resample": 2,
|
| 19 |
+
"rescale_factor": 0.00392156862745098,
|
| 20 |
+
"size": {
|
| 21 |
+
"longest_edge": 1333,
|
| 22 |
+
"shortest_edge": 1024
|
| 23 |
+
}
|
| 24 |
+
}
|
edge_tip/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a057eb25e621eaf8e2fdb8fe87a553e3c6c452e871a4815344cef60018d547f
|
| 3 |
+
size 510988573
|
edge_tip/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:25bdd974f302d237677de2dd8a6748a9f463165ab8304abf1ef5916554d08cbf
|
| 3 |
+
size 14575
|
edge_tip/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e371dc7e728d370f4debbddc234e2afd33fa943d72ea521094475520f8d718c9
|
| 3 |
+
size 627
|
edge_tip/trainer_state.json
ADDED
|
@@ -0,0 +1,184 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 23.076923076923077,
|
| 5 |
+
"global_step": 1200,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.92,
|
| 12 |
+
"learning_rate": 3.7200000000000004e-06,
|
| 13 |
+
"loss": 3.4957,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.92,
|
| 18 |
+
"eval_loss": 3.337008237838745,
|
| 19 |
+
"eval_runtime": 4.5047,
|
| 20 |
+
"eval_samples_per_second": 4.662,
|
| 21 |
+
"eval_steps_per_second": 0.666,
|
| 22 |
+
"step": 100
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 3.85,
|
| 26 |
+
"learning_rate": 7.72e-06,
|
| 27 |
+
"loss": 2.8817,
|
| 28 |
+
"step": 200
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 3.85,
|
| 32 |
+
"eval_loss": 2.994506597518921,
|
| 33 |
+
"eval_runtime": 4.0853,
|
| 34 |
+
"eval_samples_per_second": 5.14,
|
| 35 |
+
"eval_steps_per_second": 0.734,
|
| 36 |
+
"step": 200
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 5.77,
|
| 40 |
+
"learning_rate": 1.172e-05,
|
| 41 |
+
"loss": 2.5363,
|
| 42 |
+
"step": 300
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 5.77,
|
| 46 |
+
"eval_loss": 2.85880970954895,
|
| 47 |
+
"eval_runtime": 4.1442,
|
| 48 |
+
"eval_samples_per_second": 5.067,
|
| 49 |
+
"eval_steps_per_second": 0.724,
|
| 50 |
+
"step": 300
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 7.69,
|
| 54 |
+
"learning_rate": 1.5720000000000002e-05,
|
| 55 |
+
"loss": 2.3781,
|
| 56 |
+
"step": 400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 7.69,
|
| 60 |
+
"eval_loss": 2.8444266319274902,
|
| 61 |
+
"eval_runtime": 4.0477,
|
| 62 |
+
"eval_samples_per_second": 5.188,
|
| 63 |
+
"eval_steps_per_second": 0.741,
|
| 64 |
+
"step": 400
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 9.62,
|
| 68 |
+
"learning_rate": 1.968e-05,
|
| 69 |
+
"loss": 2.2276,
|
| 70 |
+
"step": 500
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 9.62,
|
| 74 |
+
"eval_loss": 2.846156120300293,
|
| 75 |
+
"eval_runtime": 4.0932,
|
| 76 |
+
"eval_samples_per_second": 5.13,
|
| 77 |
+
"eval_steps_per_second": 0.733,
|
| 78 |
+
"step": 500
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 11.54,
|
| 82 |
+
"learning_rate": 1.8835443037974686e-05,
|
| 83 |
+
"loss": 2.249,
|
| 84 |
+
"step": 600
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 11.54,
|
| 88 |
+
"eval_loss": 2.6710026264190674,
|
| 89 |
+
"eval_runtime": 4.1454,
|
| 90 |
+
"eval_samples_per_second": 5.066,
|
| 91 |
+
"eval_steps_per_second": 0.724,
|
| 92 |
+
"step": 600
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 13.46,
|
| 96 |
+
"learning_rate": 1.7569620253164558e-05,
|
| 97 |
+
"loss": 2.0842,
|
| 98 |
+
"step": 700
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 13.46,
|
| 102 |
+
"eval_loss": 2.591163396835327,
|
| 103 |
+
"eval_runtime": 4.2833,
|
| 104 |
+
"eval_samples_per_second": 4.903,
|
| 105 |
+
"eval_steps_per_second": 0.7,
|
| 106 |
+
"step": 700
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 15.38,
|
| 110 |
+
"learning_rate": 1.6303797468354434e-05,
|
| 111 |
+
"loss": 1.9348,
|
| 112 |
+
"step": 800
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 15.38,
|
| 116 |
+
"eval_loss": 2.744408369064331,
|
| 117 |
+
"eval_runtime": 4.1103,
|
| 118 |
+
"eval_samples_per_second": 5.109,
|
| 119 |
+
"eval_steps_per_second": 0.73,
|
| 120 |
+
"step": 800
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 17.31,
|
| 124 |
+
"learning_rate": 1.5037974683544306e-05,
|
| 125 |
+
"loss": 1.9074,
|
| 126 |
+
"step": 900
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 17.31,
|
| 130 |
+
"eval_loss": 2.6080548763275146,
|
| 131 |
+
"eval_runtime": 4.1034,
|
| 132 |
+
"eval_samples_per_second": 5.118,
|
| 133 |
+
"eval_steps_per_second": 0.731,
|
| 134 |
+
"step": 900
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 19.23,
|
| 138 |
+
"learning_rate": 1.377215189873418e-05,
|
| 139 |
+
"loss": 1.8806,
|
| 140 |
+
"step": 1000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 19.23,
|
| 144 |
+
"eval_loss": 2.558474540710449,
|
| 145 |
+
"eval_runtime": 4.3476,
|
| 146 |
+
"eval_samples_per_second": 4.83,
|
| 147 |
+
"eval_steps_per_second": 0.69,
|
| 148 |
+
"step": 1000
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 21.15,
|
| 152 |
+
"learning_rate": 1.2506329113924051e-05,
|
| 153 |
+
"loss": 1.8524,
|
| 154 |
+
"step": 1100
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 21.15,
|
| 158 |
+
"eval_loss": 2.5464015007019043,
|
| 159 |
+
"eval_runtime": 4.1253,
|
| 160 |
+
"eval_samples_per_second": 5.09,
|
| 161 |
+
"eval_steps_per_second": 0.727,
|
| 162 |
+
"step": 1100
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 23.08,
|
| 166 |
+
"learning_rate": 1.1240506329113925e-05,
|
| 167 |
+
"loss": 1.6774,
|
| 168 |
+
"step": 1200
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 23.08,
|
| 172 |
+
"eval_loss": 2.465409755706787,
|
| 173 |
+
"eval_runtime": 4.0985,
|
| 174 |
+
"eval_samples_per_second": 5.124,
|
| 175 |
+
"eval_steps_per_second": 0.732,
|
| 176 |
+
"step": 1200
|
| 177 |
+
}
|
| 178 |
+
],
|
| 179 |
+
"max_steps": 2080,
|
| 180 |
+
"num_train_epochs": 40,
|
| 181 |
+
"total_flos": 2.892968558906573e+18,
|
| 182 |
+
"trial_name": null,
|
| 183 |
+
"trial_params": null
|
| 184 |
+
}
|
edge_tip/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:57012aeac3225927f86e8111c78ca6c8cd3304f5c764b7ab5d948b63cb1bac01
|
| 3 |
+
size 3899
|
node/config.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "hustvl/yolos-base",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"YolosForObjectDetection"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"auxiliary_loss": false,
|
| 8 |
+
"bbox_cost": 5,
|
| 9 |
+
"bbox_loss_coefficient": 5,
|
| 10 |
+
"class_cost": 1,
|
| 11 |
+
"eos_coefficient": 0.1,
|
| 12 |
+
"giou_cost": 2,
|
| 13 |
+
"giou_loss_coefficient": 2,
|
| 14 |
+
"hidden_act": "gelu",
|
| 15 |
+
"hidden_dropout_prob": 0.0,
|
| 16 |
+
"hidden_size": 768,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "drawing"
|
| 19 |
+
},
|
| 20 |
+
"image_size": [
|
| 21 |
+
800,
|
| 22 |
+
1344
|
| 23 |
+
],
|
| 24 |
+
"initializer_range": 0.02,
|
| 25 |
+
"intermediate_size": 3072,
|
| 26 |
+
"label2id": {
|
| 27 |
+
"drawing": 0
|
| 28 |
+
},
|
| 29 |
+
"layer_norm_eps": 1e-12,
|
| 30 |
+
"model_type": "yolos",
|
| 31 |
+
"num_attention_heads": 12,
|
| 32 |
+
"num_channels": 3,
|
| 33 |
+
"num_detection_tokens": 100,
|
| 34 |
+
"num_hidden_layers": 12,
|
| 35 |
+
"patch_size": 16,
|
| 36 |
+
"qkv_bias": true,
|
| 37 |
+
"torch_dtype": "float32",
|
| 38 |
+
"transformers_version": "4.31.0",
|
| 39 |
+
"use_mid_position_embeddings": true
|
| 40 |
+
}
|
node/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ee7e5a1d31cf4a84089f4b2b51513ee2653bf4bb6642aefc55847afeb391245d
|
| 3 |
+
size 1021957701
|
node/preprocessor_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_pad": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"format": "coco_detection",
|
| 7 |
+
"image_mean": [
|
| 8 |
+
0.485,
|
| 9 |
+
0.456,
|
| 10 |
+
0.406
|
| 11 |
+
],
|
| 12 |
+
"image_processor_type": "YolosImageProcessor",
|
| 13 |
+
"image_std": [
|
| 14 |
+
0.229,
|
| 15 |
+
0.224,
|
| 16 |
+
0.225
|
| 17 |
+
],
|
| 18 |
+
"resample": 2,
|
| 19 |
+
"rescale_factor": 0.00392156862745098,
|
| 20 |
+
"size": {
|
| 21 |
+
"longest_edge": 1333,
|
| 22 |
+
"shortest_edge": 1024
|
| 23 |
+
}
|
| 24 |
+
}
|
node/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ea725b1303c57f5a30a6c30134b2b9f3170149c1638eff11fe44538af3fb734b
|
| 3 |
+
size 510988573
|
node/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b4fe081ffc9f09975a7fd61c6896e2d4d4f22ca17130b7c39dec2b3f771b138b
|
| 3 |
+
size 14575
|
node/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d6eadcbe850cf9fe8f32b73ae3df5ed0bab20b2b8c3fc55e473da44d056d58f
|
| 3 |
+
size 627
|
node/trainer_state.json
ADDED
|
@@ -0,0 +1,240 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 39.02439024390244,
|
| 5 |
+
"global_step": 1600,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 2.44,
|
| 12 |
+
"learning_rate": 3.7600000000000004e-06,
|
| 13 |
+
"loss": 3.0404,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 2.44,
|
| 18 |
+
"eval_loss": 2.7579736709594727,
|
| 19 |
+
"eval_runtime": 4.5474,
|
| 20 |
+
"eval_samples_per_second": 4.618,
|
| 21 |
+
"eval_steps_per_second": 0.66,
|
| 22 |
+
"step": 100
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 4.88,
|
| 26 |
+
"learning_rate": 7.76e-06,
|
| 27 |
+
"loss": 2.3488,
|
| 28 |
+
"step": 200
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 4.88,
|
| 32 |
+
"eval_loss": 2.4349780082702637,
|
| 33 |
+
"eval_runtime": 4.0455,
|
| 34 |
+
"eval_samples_per_second": 5.191,
|
| 35 |
+
"eval_steps_per_second": 0.742,
|
| 36 |
+
"step": 200
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 7.32,
|
| 40 |
+
"learning_rate": 1.172e-05,
|
| 41 |
+
"loss": 1.9928,
|
| 42 |
+
"step": 300
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 7.32,
|
| 46 |
+
"eval_loss": 2.113464832305908,
|
| 47 |
+
"eval_runtime": 4.0677,
|
| 48 |
+
"eval_samples_per_second": 5.163,
|
| 49 |
+
"eval_steps_per_second": 0.738,
|
| 50 |
+
"step": 300
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 9.76,
|
| 54 |
+
"learning_rate": 1.5720000000000002e-05,
|
| 55 |
+
"loss": 1.7819,
|
| 56 |
+
"step": 400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 9.76,
|
| 60 |
+
"eval_loss": 1.968690276145935,
|
| 61 |
+
"eval_runtime": 4.0436,
|
| 62 |
+
"eval_samples_per_second": 5.193,
|
| 63 |
+
"eval_steps_per_second": 0.742,
|
| 64 |
+
"step": 400
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 12.2,
|
| 68 |
+
"learning_rate": 1.972e-05,
|
| 69 |
+
"loss": 1.5877,
|
| 70 |
+
"step": 500
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 12.2,
|
| 74 |
+
"eval_loss": 1.7520062923431396,
|
| 75 |
+
"eval_runtime": 4.029,
|
| 76 |
+
"eval_samples_per_second": 5.212,
|
| 77 |
+
"eval_steps_per_second": 0.745,
|
| 78 |
+
"step": 500
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 14.63,
|
| 82 |
+
"learning_rate": 1.836842105263158e-05,
|
| 83 |
+
"loss": 1.4275,
|
| 84 |
+
"step": 600
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 14.63,
|
| 88 |
+
"eval_loss": 1.7901742458343506,
|
| 89 |
+
"eval_runtime": 4.0456,
|
| 90 |
+
"eval_samples_per_second": 5.191,
|
| 91 |
+
"eval_steps_per_second": 0.742,
|
| 92 |
+
"step": 600
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 17.07,
|
| 96 |
+
"learning_rate": 1.66140350877193e-05,
|
| 97 |
+
"loss": 1.1999,
|
| 98 |
+
"step": 700
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 17.07,
|
| 102 |
+
"eval_loss": 1.8930068016052246,
|
| 103 |
+
"eval_runtime": 4.0412,
|
| 104 |
+
"eval_samples_per_second": 5.196,
|
| 105 |
+
"eval_steps_per_second": 0.742,
|
| 106 |
+
"step": 700
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 19.51,
|
| 110 |
+
"learning_rate": 1.4859649122807018e-05,
|
| 111 |
+
"loss": 1.2049,
|
| 112 |
+
"step": 800
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 19.51,
|
| 116 |
+
"eval_loss": 1.6747853755950928,
|
| 117 |
+
"eval_runtime": 4.0641,
|
| 118 |
+
"eval_samples_per_second": 5.167,
|
| 119 |
+
"eval_steps_per_second": 0.738,
|
| 120 |
+
"step": 800
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 21.95,
|
| 124 |
+
"learning_rate": 1.310526315789474e-05,
|
| 125 |
+
"loss": 1.0865,
|
| 126 |
+
"step": 900
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 21.95,
|
| 130 |
+
"eval_loss": 1.642054796218872,
|
| 131 |
+
"eval_runtime": 4.0327,
|
| 132 |
+
"eval_samples_per_second": 5.207,
|
| 133 |
+
"eval_steps_per_second": 0.744,
|
| 134 |
+
"step": 900
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 24.39,
|
| 138 |
+
"learning_rate": 1.136842105263158e-05,
|
| 139 |
+
"loss": 1.0559,
|
| 140 |
+
"step": 1000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 24.39,
|
| 144 |
+
"eval_loss": 1.6494040489196777,
|
| 145 |
+
"eval_runtime": 4.1243,
|
| 146 |
+
"eval_samples_per_second": 5.092,
|
| 147 |
+
"eval_steps_per_second": 0.727,
|
| 148 |
+
"step": 1000
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 26.83,
|
| 152 |
+
"learning_rate": 9.614035087719298e-06,
|
| 153 |
+
"loss": 0.9572,
|
| 154 |
+
"step": 1100
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 26.83,
|
| 158 |
+
"eval_loss": 1.6545538902282715,
|
| 159 |
+
"eval_runtime": 4.1781,
|
| 160 |
+
"eval_samples_per_second": 5.026,
|
| 161 |
+
"eval_steps_per_second": 0.718,
|
| 162 |
+
"step": 1100
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 29.27,
|
| 166 |
+
"learning_rate": 7.859649122807018e-06,
|
| 167 |
+
"loss": 0.91,
|
| 168 |
+
"step": 1200
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 29.27,
|
| 172 |
+
"eval_loss": 1.637965440750122,
|
| 173 |
+
"eval_runtime": 4.203,
|
| 174 |
+
"eval_samples_per_second": 4.996,
|
| 175 |
+
"eval_steps_per_second": 0.714,
|
| 176 |
+
"step": 1200
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 31.71,
|
| 180 |
+
"learning_rate": 6.105263157894738e-06,
|
| 181 |
+
"loss": 0.8774,
|
| 182 |
+
"step": 1300
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 31.71,
|
| 186 |
+
"eval_loss": 1.5634162425994873,
|
| 187 |
+
"eval_runtime": 4.1567,
|
| 188 |
+
"eval_samples_per_second": 5.052,
|
| 189 |
+
"eval_steps_per_second": 0.722,
|
| 190 |
+
"step": 1300
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 34.15,
|
| 194 |
+
"learning_rate": 4.350877192982457e-06,
|
| 195 |
+
"loss": 0.8501,
|
| 196 |
+
"step": 1400
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 34.15,
|
| 200 |
+
"eval_loss": 1.613369345664978,
|
| 201 |
+
"eval_runtime": 4.1441,
|
| 202 |
+
"eval_samples_per_second": 5.067,
|
| 203 |
+
"eval_steps_per_second": 0.724,
|
| 204 |
+
"step": 1400
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 36.59,
|
| 208 |
+
"learning_rate": 2.5964912280701754e-06,
|
| 209 |
+
"loss": 0.8121,
|
| 210 |
+
"step": 1500
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 36.59,
|
| 214 |
+
"eval_loss": 1.5846848487854004,
|
| 215 |
+
"eval_runtime": 4.1418,
|
| 216 |
+
"eval_samples_per_second": 5.07,
|
| 217 |
+
"eval_steps_per_second": 0.724,
|
| 218 |
+
"step": 1500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 39.02,
|
| 222 |
+
"learning_rate": 8.421052631578948e-07,
|
| 223 |
+
"loss": 0.8692,
|
| 224 |
+
"step": 1600
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 39.02,
|
| 228 |
+
"eval_loss": 1.591667890548706,
|
| 229 |
+
"eval_runtime": 4.1152,
|
| 230 |
+
"eval_samples_per_second": 5.103,
|
| 231 |
+
"eval_steps_per_second": 0.729,
|
| 232 |
+
"step": 1600
|
| 233 |
+
}
|
| 234 |
+
],
|
| 235 |
+
"max_steps": 1640,
|
| 236 |
+
"num_train_epochs": 40,
|
| 237 |
+
"total_flos": 3.8572914118754304e+18,
|
| 238 |
+
"trial_name": null,
|
| 239 |
+
"trial_params": null
|
| 240 |
+
}
|
node/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e229a160272e6cf6fa41a627ed2551f4b782be702ca87f0606dc18beb65dccf
|
| 3 |
+
size 3899
|