Upload folder using huggingface_hub
Browse files- checkpoint-12573/config.json +134 -0
- checkpoint-12573/model.safetensors +3 -0
- checkpoint-12573/optimizer.pt +3 -0
- checkpoint-12573/preprocessor_config.json +22 -0
- checkpoint-12573/rng_state.pth +3 -0
- checkpoint-12573/scheduler.pt +3 -0
- checkpoint-12573/trainer_state.json +1060 -0
- checkpoint-12573/training_args.bin +3 -0
- model.safetensors +1 -1
- training_args.bin +1 -1
checkpoint-12573/config.json
ADDED
|
@@ -0,0 +1,134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-base-patch16-224-in21k",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"ViTForImageClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"encoder_stride": 16,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.0,
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "ace of clubs",
|
| 13 |
+
"1": "ace of diamonds",
|
| 14 |
+
"2": "ace of hearts",
|
| 15 |
+
"3": "ace of spades",
|
| 16 |
+
"4": "eight of clubs",
|
| 17 |
+
"5": "eight of diamonds",
|
| 18 |
+
"6": "eight of hearts",
|
| 19 |
+
"7": "eight of spades",
|
| 20 |
+
"8": "five of clubs",
|
| 21 |
+
"9": "five of diamonds",
|
| 22 |
+
"10": "five of hearts",
|
| 23 |
+
"11": "five of spades",
|
| 24 |
+
"12": "four of clubs",
|
| 25 |
+
"13": "four of diamonds",
|
| 26 |
+
"14": "four of hearts",
|
| 27 |
+
"15": "four of spades",
|
| 28 |
+
"16": "jack of clubs",
|
| 29 |
+
"17": "jack of diamonds",
|
| 30 |
+
"18": "jack of hearts",
|
| 31 |
+
"19": "jack of spades",
|
| 32 |
+
"20": "joker",
|
| 33 |
+
"21": "king of clubs",
|
| 34 |
+
"22": "king of diamonds",
|
| 35 |
+
"23": "king of hearts",
|
| 36 |
+
"24": "king of spades",
|
| 37 |
+
"25": "nine of clubs",
|
| 38 |
+
"26": "nine of diamonds",
|
| 39 |
+
"27": "nine of hearts",
|
| 40 |
+
"28": "nine of spades",
|
| 41 |
+
"29": "queen of clubs",
|
| 42 |
+
"30": "queen of diamonds",
|
| 43 |
+
"31": "queen of hearts",
|
| 44 |
+
"32": "queen of spades",
|
| 45 |
+
"33": "seven of clubs",
|
| 46 |
+
"34": "seven of diamonds",
|
| 47 |
+
"35": "seven of hearts",
|
| 48 |
+
"36": "seven of spades",
|
| 49 |
+
"37": "six of clubs",
|
| 50 |
+
"38": "six of diamonds",
|
| 51 |
+
"39": "six of hearts",
|
| 52 |
+
"40": "six of spades",
|
| 53 |
+
"41": "ten of clubs",
|
| 54 |
+
"42": "ten of diamonds",
|
| 55 |
+
"43": "ten of hearts",
|
| 56 |
+
"44": "ten of spades",
|
| 57 |
+
"45": "three of clubs",
|
| 58 |
+
"46": "three of diamonds",
|
| 59 |
+
"47": "three of hearts",
|
| 60 |
+
"48": "three of spades",
|
| 61 |
+
"49": "two of clubs",
|
| 62 |
+
"50": "two of diamonds",
|
| 63 |
+
"51": "two of hearts",
|
| 64 |
+
"52": "two of spades"
|
| 65 |
+
},
|
| 66 |
+
"image_size": 224,
|
| 67 |
+
"initializer_range": 0.02,
|
| 68 |
+
"intermediate_size": 3072,
|
| 69 |
+
"label2id": {
|
| 70 |
+
"ace of clubs": 0,
|
| 71 |
+
"ace of diamonds": 1,
|
| 72 |
+
"ace of hearts": 2,
|
| 73 |
+
"ace of spades": 3,
|
| 74 |
+
"eight of clubs": 4,
|
| 75 |
+
"eight of diamonds": 5,
|
| 76 |
+
"eight of hearts": 6,
|
| 77 |
+
"eight of spades": 7,
|
| 78 |
+
"five of clubs": 8,
|
| 79 |
+
"five of diamonds": 9,
|
| 80 |
+
"five of hearts": 10,
|
| 81 |
+
"five of spades": 11,
|
| 82 |
+
"four of clubs": 12,
|
| 83 |
+
"four of diamonds": 13,
|
| 84 |
+
"four of hearts": 14,
|
| 85 |
+
"four of spades": 15,
|
| 86 |
+
"jack of clubs": 16,
|
| 87 |
+
"jack of diamonds": 17,
|
| 88 |
+
"jack of hearts": 18,
|
| 89 |
+
"jack of spades": 19,
|
| 90 |
+
"joker": 20,
|
| 91 |
+
"king of clubs": 21,
|
| 92 |
+
"king of diamonds": 22,
|
| 93 |
+
"king of hearts": 23,
|
| 94 |
+
"king of spades": 24,
|
| 95 |
+
"nine of clubs": 25,
|
| 96 |
+
"nine of diamonds": 26,
|
| 97 |
+
"nine of hearts": 27,
|
| 98 |
+
"nine of spades": 28,
|
| 99 |
+
"queen of clubs": 29,
|
| 100 |
+
"queen of diamonds": 30,
|
| 101 |
+
"queen of hearts": 31,
|
| 102 |
+
"queen of spades": 32,
|
| 103 |
+
"seven of clubs": 33,
|
| 104 |
+
"seven of diamonds": 34,
|
| 105 |
+
"seven of hearts": 35,
|
| 106 |
+
"seven of spades": 36,
|
| 107 |
+
"six of clubs": 37,
|
| 108 |
+
"six of diamonds": 38,
|
| 109 |
+
"six of hearts": 39,
|
| 110 |
+
"six of spades": 40,
|
| 111 |
+
"ten of clubs": 41,
|
| 112 |
+
"ten of diamonds": 42,
|
| 113 |
+
"ten of hearts": 43,
|
| 114 |
+
"ten of spades": 44,
|
| 115 |
+
"three of clubs": 45,
|
| 116 |
+
"three of diamonds": 46,
|
| 117 |
+
"three of hearts": 47,
|
| 118 |
+
"three of spades": 48,
|
| 119 |
+
"two of clubs": 49,
|
| 120 |
+
"two of diamonds": 50,
|
| 121 |
+
"two of hearts": 51,
|
| 122 |
+
"two of spades": 52
|
| 123 |
+
},
|
| 124 |
+
"layer_norm_eps": 1e-12,
|
| 125 |
+
"model_type": "vit",
|
| 126 |
+
"num_attention_heads": 12,
|
| 127 |
+
"num_channels": 3,
|
| 128 |
+
"num_hidden_layers": 12,
|
| 129 |
+
"patch_size": 16,
|
| 130 |
+
"problem_type": "single_label_classification",
|
| 131 |
+
"qkv_bias": true,
|
| 132 |
+
"torch_dtype": "float32",
|
| 133 |
+
"transformers_version": "4.35.2"
|
| 134 |
+
}
|
checkpoint-12573/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e67fe27f4a357a2b8bce1a9e565ac9aa096eb5fed005739ce9dd866069f41146
|
| 3 |
+
size 343380860
|
checkpoint-12573/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d60df09a2bc821843db71c08038c9c7a7158ebc3ab78cdc726d3fdc81df26b80
|
| 3 |
+
size 686882181
|
checkpoint-12573/preprocessor_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_rescale": true,
|
| 4 |
+
"do_resize": true,
|
| 5 |
+
"image_mean": [
|
| 6 |
+
0.5,
|
| 7 |
+
0.5,
|
| 8 |
+
0.5
|
| 9 |
+
],
|
| 10 |
+
"image_processor_type": "ViTImageProcessor",
|
| 11 |
+
"image_std": [
|
| 12 |
+
0.5,
|
| 13 |
+
0.5,
|
| 14 |
+
0.5
|
| 15 |
+
],
|
| 16 |
+
"resample": 2,
|
| 17 |
+
"rescale_factor": 0.00392156862745098,
|
| 18 |
+
"size": {
|
| 19 |
+
"height": 224,
|
| 20 |
+
"width": 224
|
| 21 |
+
}
|
| 22 |
+
}
|
checkpoint-12573/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b08651b8766f027755554427495c802175585671567ae2d8e447cc15e645bcf3
|
| 3 |
+
size 14575
|
checkpoint-12573/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c79712c2aeb667d92ddfe915471307631ff08b6064be3501a25c1c88d1734e98
|
| 3 |
+
size 627
|
checkpoint-12573/trainer_state.json
ADDED
|
@@ -0,0 +1,1060 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 2.2068374156951904,
|
| 3 |
+
"best_model_checkpoint": "card_type_image_detection/checkpoint-12573",
|
| 4 |
+
"epoch": 99.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 12573,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_accuracy": 0.023209876543209877,
|
| 14 |
+
"eval_loss": 3.959347724914551,
|
| 15 |
+
"eval_runtime": 16.3056,
|
| 16 |
+
"eval_samples_per_second": 124.191,
|
| 17 |
+
"eval_steps_per_second": 3.925,
|
| 18 |
+
"step": 127
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 2.0,
|
| 22 |
+
"eval_accuracy": 0.047407407407407405,
|
| 23 |
+
"eval_loss": 3.927157163619995,
|
| 24 |
+
"eval_runtime": 16.3398,
|
| 25 |
+
"eval_samples_per_second": 123.93,
|
| 26 |
+
"eval_steps_per_second": 3.917,
|
| 27 |
+
"step": 254
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 3.0,
|
| 31 |
+
"eval_accuracy": 0.0819753086419753,
|
| 32 |
+
"eval_loss": 3.8877668380737305,
|
| 33 |
+
"eval_runtime": 16.2853,
|
| 34 |
+
"eval_samples_per_second": 124.345,
|
| 35 |
+
"eval_steps_per_second": 3.93,
|
| 36 |
+
"step": 381
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 3.94,
|
| 40 |
+
"learning_rate": 1.9288537549407114e-06,
|
| 41 |
+
"loss": 3.9022,
|
| 42 |
+
"step": 500
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 4.0,
|
| 46 |
+
"eval_accuracy": 0.09876543209876543,
|
| 47 |
+
"eval_loss": 3.8418056964874268,
|
| 48 |
+
"eval_runtime": 16.2864,
|
| 49 |
+
"eval_samples_per_second": 124.337,
|
| 50 |
+
"eval_steps_per_second": 3.93,
|
| 51 |
+
"step": 508
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 5.0,
|
| 55 |
+
"eval_accuracy": 0.12,
|
| 56 |
+
"eval_loss": 3.7911477088928223,
|
| 57 |
+
"eval_runtime": 16.2971,
|
| 58 |
+
"eval_samples_per_second": 124.255,
|
| 59 |
+
"eval_steps_per_second": 3.927,
|
| 60 |
+
"step": 635
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 6.0,
|
| 64 |
+
"eval_accuracy": 0.1382716049382716,
|
| 65 |
+
"eval_loss": 3.7379021644592285,
|
| 66 |
+
"eval_runtime": 16.4902,
|
| 67 |
+
"eval_samples_per_second": 122.8,
|
| 68 |
+
"eval_steps_per_second": 3.881,
|
| 69 |
+
"step": 762
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 7.0,
|
| 73 |
+
"eval_accuracy": 0.1580246913580247,
|
| 74 |
+
"eval_loss": 3.6840596199035645,
|
| 75 |
+
"eval_runtime": 16.2622,
|
| 76 |
+
"eval_samples_per_second": 124.522,
|
| 77 |
+
"eval_steps_per_second": 3.936,
|
| 78 |
+
"step": 889
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 7.87,
|
| 82 |
+
"learning_rate": 1.849802371541502e-06,
|
| 83 |
+
"loss": 3.6821,
|
| 84 |
+
"step": 1000
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 8.0,
|
| 88 |
+
"eval_accuracy": 0.17777777777777778,
|
| 89 |
+
"eval_loss": 3.6320221424102783,
|
| 90 |
+
"eval_runtime": 16.3988,
|
| 91 |
+
"eval_samples_per_second": 123.485,
|
| 92 |
+
"eval_steps_per_second": 3.903,
|
| 93 |
+
"step": 1016
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 9.0,
|
| 97 |
+
"eval_accuracy": 0.19654320987654322,
|
| 98 |
+
"eval_loss": 3.580777883529663,
|
| 99 |
+
"eval_runtime": 16.325,
|
| 100 |
+
"eval_samples_per_second": 124.043,
|
| 101 |
+
"eval_steps_per_second": 3.92,
|
| 102 |
+
"step": 1143
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 10.0,
|
| 106 |
+
"eval_accuracy": 0.21580246913580248,
|
| 107 |
+
"eval_loss": 3.5295674800872803,
|
| 108 |
+
"eval_runtime": 16.3625,
|
| 109 |
+
"eval_samples_per_second": 123.758,
|
| 110 |
+
"eval_steps_per_second": 3.911,
|
| 111 |
+
"step": 1270
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 11.0,
|
| 115 |
+
"eval_accuracy": 0.23703703703703705,
|
| 116 |
+
"eval_loss": 3.479665994644165,
|
| 117 |
+
"eval_runtime": 16.5252,
|
| 118 |
+
"eval_samples_per_second": 122.54,
|
| 119 |
+
"eval_steps_per_second": 3.873,
|
| 120 |
+
"step": 1397
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 11.81,
|
| 124 |
+
"learning_rate": 1.7707509881422924e-06,
|
| 125 |
+
"loss": 3.4599,
|
| 126 |
+
"step": 1500
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 12.0,
|
| 130 |
+
"eval_accuracy": 0.24592592592592594,
|
| 131 |
+
"eval_loss": 3.431598663330078,
|
| 132 |
+
"eval_runtime": 16.3048,
|
| 133 |
+
"eval_samples_per_second": 124.197,
|
| 134 |
+
"eval_steps_per_second": 3.925,
|
| 135 |
+
"step": 1524
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 13.0,
|
| 139 |
+
"eval_accuracy": 0.25728395061728393,
|
| 140 |
+
"eval_loss": 3.38520884513855,
|
| 141 |
+
"eval_runtime": 16.3716,
|
| 142 |
+
"eval_samples_per_second": 123.69,
|
| 143 |
+
"eval_steps_per_second": 3.909,
|
| 144 |
+
"step": 1651
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 14.0,
|
| 148 |
+
"eval_accuracy": 0.2730864197530864,
|
| 149 |
+
"eval_loss": 3.341092348098755,
|
| 150 |
+
"eval_runtime": 16.4208,
|
| 151 |
+
"eval_samples_per_second": 123.319,
|
| 152 |
+
"eval_steps_per_second": 3.897,
|
| 153 |
+
"step": 1778
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 15.0,
|
| 157 |
+
"eval_accuracy": 0.291358024691358,
|
| 158 |
+
"eval_loss": 3.299830198287964,
|
| 159 |
+
"eval_runtime": 16.349,
|
| 160 |
+
"eval_samples_per_second": 123.861,
|
| 161 |
+
"eval_steps_per_second": 3.915,
|
| 162 |
+
"step": 1905
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 15.75,
|
| 166 |
+
"learning_rate": 1.691699604743083e-06,
|
| 167 |
+
"loss": 3.2748,
|
| 168 |
+
"step": 2000
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 16.0,
|
| 172 |
+
"eval_accuracy": 0.305679012345679,
|
| 173 |
+
"eval_loss": 3.2592811584472656,
|
| 174 |
+
"eval_runtime": 16.2664,
|
| 175 |
+
"eval_samples_per_second": 124.49,
|
| 176 |
+
"eval_steps_per_second": 3.934,
|
| 177 |
+
"step": 2032
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 17.0,
|
| 181 |
+
"eval_accuracy": 0.32641975308641974,
|
| 182 |
+
"eval_loss": 3.2209341526031494,
|
| 183 |
+
"eval_runtime": 16.4355,
|
| 184 |
+
"eval_samples_per_second": 123.209,
|
| 185 |
+
"eval_steps_per_second": 3.894,
|
| 186 |
+
"step": 2159
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 18.0,
|
| 190 |
+
"eval_accuracy": 0.3437037037037037,
|
| 191 |
+
"eval_loss": 3.184448003768921,
|
| 192 |
+
"eval_runtime": 16.3387,
|
| 193 |
+
"eval_samples_per_second": 123.939,
|
| 194 |
+
"eval_steps_per_second": 3.917,
|
| 195 |
+
"step": 2286
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 19.0,
|
| 199 |
+
"eval_accuracy": 0.345679012345679,
|
| 200 |
+
"eval_loss": 3.1492130756378174,
|
| 201 |
+
"eval_runtime": 16.2457,
|
| 202 |
+
"eval_samples_per_second": 124.648,
|
| 203 |
+
"eval_steps_per_second": 3.939,
|
| 204 |
+
"step": 2413
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 19.69,
|
| 208 |
+
"learning_rate": 1.6126482213438735e-06,
|
| 209 |
+
"loss": 3.1183,
|
| 210 |
+
"step": 2500
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 20.0,
|
| 214 |
+
"eval_accuracy": 0.3634567901234568,
|
| 215 |
+
"eval_loss": 3.117238998413086,
|
| 216 |
+
"eval_runtime": 16.3142,
|
| 217 |
+
"eval_samples_per_second": 124.125,
|
| 218 |
+
"eval_steps_per_second": 3.923,
|
| 219 |
+
"step": 2540
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 21.0,
|
| 223 |
+
"eval_accuracy": 0.37382716049382714,
|
| 224 |
+
"eval_loss": 3.0856010913848877,
|
| 225 |
+
"eval_runtime": 16.4136,
|
| 226 |
+
"eval_samples_per_second": 123.374,
|
| 227 |
+
"eval_steps_per_second": 3.899,
|
| 228 |
+
"step": 2667
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 22.0,
|
| 232 |
+
"eval_accuracy": 0.38469135802469134,
|
| 233 |
+
"eval_loss": 3.0540201663970947,
|
| 234 |
+
"eval_runtime": 16.404,
|
| 235 |
+
"eval_samples_per_second": 123.446,
|
| 236 |
+
"eval_steps_per_second": 3.901,
|
| 237 |
+
"step": 2794
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 23.0,
|
| 241 |
+
"eval_accuracy": 0.3916049382716049,
|
| 242 |
+
"eval_loss": 3.025864362716675,
|
| 243 |
+
"eval_runtime": 16.3387,
|
| 244 |
+
"eval_samples_per_second": 123.939,
|
| 245 |
+
"eval_steps_per_second": 3.917,
|
| 246 |
+
"step": 2921
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 23.62,
|
| 250 |
+
"learning_rate": 1.533596837944664e-06,
|
| 251 |
+
"loss": 2.9883,
|
| 252 |
+
"step": 3000
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 24.0,
|
| 256 |
+
"eval_accuracy": 0.4014814814814815,
|
| 257 |
+
"eval_loss": 2.9977123737335205,
|
| 258 |
+
"eval_runtime": 16.5183,
|
| 259 |
+
"eval_samples_per_second": 122.592,
|
| 260 |
+
"eval_steps_per_second": 3.874,
|
| 261 |
+
"step": 3048
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 25.0,
|
| 265 |
+
"eval_accuracy": 0.4103703703703704,
|
| 266 |
+
"eval_loss": 2.970736503601074,
|
| 267 |
+
"eval_runtime": 16.4478,
|
| 268 |
+
"eval_samples_per_second": 123.117,
|
| 269 |
+
"eval_steps_per_second": 3.891,
|
| 270 |
+
"step": 3175
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 26.0,
|
| 274 |
+
"eval_accuracy": 0.42962962962962964,
|
| 275 |
+
"eval_loss": 2.9437484741210938,
|
| 276 |
+
"eval_runtime": 16.3805,
|
| 277 |
+
"eval_samples_per_second": 123.623,
|
| 278 |
+
"eval_steps_per_second": 3.907,
|
| 279 |
+
"step": 3302
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 27.0,
|
| 283 |
+
"eval_accuracy": 0.4365432098765432,
|
| 284 |
+
"eval_loss": 2.918938636779785,
|
| 285 |
+
"eval_runtime": 16.4584,
|
| 286 |
+
"eval_samples_per_second": 123.038,
|
| 287 |
+
"eval_steps_per_second": 3.889,
|
| 288 |
+
"step": 3429
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 27.56,
|
| 292 |
+
"learning_rate": 1.4545454545454544e-06,
|
| 293 |
+
"loss": 2.8743,
|
| 294 |
+
"step": 3500
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 28.0,
|
| 298 |
+
"eval_accuracy": 0.44592592592592595,
|
| 299 |
+
"eval_loss": 2.8959240913391113,
|
| 300 |
+
"eval_runtime": 16.8148,
|
| 301 |
+
"eval_samples_per_second": 120.43,
|
| 302 |
+
"eval_steps_per_second": 3.806,
|
| 303 |
+
"step": 3556
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 29.0,
|
| 307 |
+
"eval_accuracy": 0.4597530864197531,
|
| 308 |
+
"eval_loss": 2.8701860904693604,
|
| 309 |
+
"eval_runtime": 16.5161,
|
| 310 |
+
"eval_samples_per_second": 122.607,
|
| 311 |
+
"eval_steps_per_second": 3.875,
|
| 312 |
+
"step": 3683
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 30.0,
|
| 316 |
+
"eval_accuracy": 0.46469135802469136,
|
| 317 |
+
"eval_loss": 2.848033905029297,
|
| 318 |
+
"eval_runtime": 16.6005,
|
| 319 |
+
"eval_samples_per_second": 121.984,
|
| 320 |
+
"eval_steps_per_second": 3.855,
|
| 321 |
+
"step": 3810
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 31.0,
|
| 325 |
+
"eval_accuracy": 0.4730864197530864,
|
| 326 |
+
"eval_loss": 2.825526714324951,
|
| 327 |
+
"eval_runtime": 16.5596,
|
| 328 |
+
"eval_samples_per_second": 122.286,
|
| 329 |
+
"eval_steps_per_second": 3.865,
|
| 330 |
+
"step": 3937
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 31.5,
|
| 334 |
+
"learning_rate": 1.375494071146245e-06,
|
| 335 |
+
"loss": 2.773,
|
| 336 |
+
"step": 4000
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 32.0,
|
| 340 |
+
"eval_accuracy": 0.47802469135802467,
|
| 341 |
+
"eval_loss": 2.804414749145508,
|
| 342 |
+
"eval_runtime": 16.6816,
|
| 343 |
+
"eval_samples_per_second": 121.391,
|
| 344 |
+
"eval_steps_per_second": 3.837,
|
| 345 |
+
"step": 4064
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 33.0,
|
| 349 |
+
"eval_accuracy": 0.47555555555555556,
|
| 350 |
+
"eval_loss": 2.7853612899780273,
|
| 351 |
+
"eval_runtime": 16.7524,
|
| 352 |
+
"eval_samples_per_second": 120.878,
|
| 353 |
+
"eval_steps_per_second": 3.82,
|
| 354 |
+
"step": 4191
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 34.0,
|
| 358 |
+
"eval_accuracy": 0.4918518518518519,
|
| 359 |
+
"eval_loss": 2.7612335681915283,
|
| 360 |
+
"eval_runtime": 16.5289,
|
| 361 |
+
"eval_samples_per_second": 122.513,
|
| 362 |
+
"eval_steps_per_second": 3.872,
|
| 363 |
+
"step": 4318
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 35.0,
|
| 367 |
+
"eval_accuracy": 0.497283950617284,
|
| 368 |
+
"eval_loss": 2.741335391998291,
|
| 369 |
+
"eval_runtime": 16.6912,
|
| 370 |
+
"eval_samples_per_second": 121.321,
|
| 371 |
+
"eval_steps_per_second": 3.834,
|
| 372 |
+
"step": 4445
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 35.43,
|
| 376 |
+
"learning_rate": 1.2964426877470356e-06,
|
| 377 |
+
"loss": 2.6822,
|
| 378 |
+
"step": 4500
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 36.0,
|
| 382 |
+
"eval_accuracy": 0.49876543209876545,
|
| 383 |
+
"eval_loss": 2.724756956100464,
|
| 384 |
+
"eval_runtime": 16.517,
|
| 385 |
+
"eval_samples_per_second": 122.601,
|
| 386 |
+
"eval_steps_per_second": 3.875,
|
| 387 |
+
"step": 4572
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 37.0,
|
| 391 |
+
"eval_accuracy": 0.49777777777777776,
|
| 392 |
+
"eval_loss": 2.7084524631500244,
|
| 393 |
+
"eval_runtime": 16.6709,
|
| 394 |
+
"eval_samples_per_second": 121.469,
|
| 395 |
+
"eval_steps_per_second": 3.839,
|
| 396 |
+
"step": 4699
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 38.0,
|
| 400 |
+
"eval_accuracy": 0.5150617283950617,
|
| 401 |
+
"eval_loss": 2.6876704692840576,
|
| 402 |
+
"eval_runtime": 16.6152,
|
| 403 |
+
"eval_samples_per_second": 121.876,
|
| 404 |
+
"eval_steps_per_second": 3.852,
|
| 405 |
+
"step": 4826
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 39.0,
|
| 409 |
+
"eval_accuracy": 0.5106172839506172,
|
| 410 |
+
"eval_loss": 2.6705734729766846,
|
| 411 |
+
"eval_runtime": 16.5075,
|
| 412 |
+
"eval_samples_per_second": 122.671,
|
| 413 |
+
"eval_steps_per_second": 3.877,
|
| 414 |
+
"step": 4953
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 39.37,
|
| 418 |
+
"learning_rate": 1.217391304347826e-06,
|
| 419 |
+
"loss": 2.5988,
|
| 420 |
+
"step": 5000
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 40.0,
|
| 424 |
+
"eval_accuracy": 0.5116049382716049,
|
| 425 |
+
"eval_loss": 2.655836343765259,
|
| 426 |
+
"eval_runtime": 16.7707,
|
| 427 |
+
"eval_samples_per_second": 120.746,
|
| 428 |
+
"eval_steps_per_second": 3.816,
|
| 429 |
+
"step": 5080
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 41.0,
|
| 433 |
+
"eval_accuracy": 0.5125925925925926,
|
| 434 |
+
"eval_loss": 2.6394340991973877,
|
| 435 |
+
"eval_runtime": 16.7747,
|
| 436 |
+
"eval_samples_per_second": 120.718,
|
| 437 |
+
"eval_steps_per_second": 3.815,
|
| 438 |
+
"step": 5207
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 42.0,
|
| 442 |
+
"eval_accuracy": 0.5209876543209877,
|
| 443 |
+
"eval_loss": 2.623532295227051,
|
| 444 |
+
"eval_runtime": 16.7009,
|
| 445 |
+
"eval_samples_per_second": 121.251,
|
| 446 |
+
"eval_steps_per_second": 3.832,
|
| 447 |
+
"step": 5334
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 43.0,
|
| 451 |
+
"eval_accuracy": 0.5249382716049382,
|
| 452 |
+
"eval_loss": 2.6078622341156006,
|
| 453 |
+
"eval_runtime": 16.3629,
|
| 454 |
+
"eval_samples_per_second": 123.756,
|
| 455 |
+
"eval_steps_per_second": 3.911,
|
| 456 |
+
"step": 5461
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 43.31,
|
| 460 |
+
"learning_rate": 1.1383399209486167e-06,
|
| 461 |
+
"loss": 2.5306,
|
| 462 |
+
"step": 5500
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 44.0,
|
| 466 |
+
"eval_accuracy": 0.5293827160493827,
|
| 467 |
+
"eval_loss": 2.592928647994995,
|
| 468 |
+
"eval_runtime": 16.7766,
|
| 469 |
+
"eval_samples_per_second": 120.704,
|
| 470 |
+
"eval_steps_per_second": 3.815,
|
| 471 |
+
"step": 5588
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 45.0,
|
| 475 |
+
"eval_accuracy": 0.5338271604938272,
|
| 476 |
+
"eval_loss": 2.575329542160034,
|
| 477 |
+
"eval_runtime": 17.0334,
|
| 478 |
+
"eval_samples_per_second": 118.884,
|
| 479 |
+
"eval_steps_per_second": 3.757,
|
| 480 |
+
"step": 5715
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 46.0,
|
| 484 |
+
"eval_accuracy": 0.542716049382716,
|
| 485 |
+
"eval_loss": 2.5605058670043945,
|
| 486 |
+
"eval_runtime": 16.7153,
|
| 487 |
+
"eval_samples_per_second": 121.146,
|
| 488 |
+
"eval_steps_per_second": 3.829,
|
| 489 |
+
"step": 5842
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 47.0,
|
| 493 |
+
"eval_accuracy": 0.548641975308642,
|
| 494 |
+
"eval_loss": 2.548151969909668,
|
| 495 |
+
"eval_runtime": 16.3632,
|
| 496 |
+
"eval_samples_per_second": 123.754,
|
| 497 |
+
"eval_steps_per_second": 3.911,
|
| 498 |
+
"step": 5969
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 47.24,
|
| 502 |
+
"learning_rate": 1.0592885375494072e-06,
|
| 503 |
+
"loss": 2.4638,
|
| 504 |
+
"step": 6000
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 48.0,
|
| 508 |
+
"eval_accuracy": 0.5530864197530864,
|
| 509 |
+
"eval_loss": 2.5332412719726562,
|
| 510 |
+
"eval_runtime": 16.2257,
|
| 511 |
+
"eval_samples_per_second": 124.802,
|
| 512 |
+
"eval_steps_per_second": 3.944,
|
| 513 |
+
"step": 6096
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 49.0,
|
| 517 |
+
"eval_accuracy": 0.5535802469135802,
|
| 518 |
+
"eval_loss": 2.521514654159546,
|
| 519 |
+
"eval_runtime": 16.2855,
|
| 520 |
+
"eval_samples_per_second": 124.344,
|
| 521 |
+
"eval_steps_per_second": 3.93,
|
| 522 |
+
"step": 6223
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 50.0,
|
| 526 |
+
"eval_accuracy": 0.5619753086419753,
|
| 527 |
+
"eval_loss": 2.507723569869995,
|
| 528 |
+
"eval_runtime": 16.2945,
|
| 529 |
+
"eval_samples_per_second": 124.275,
|
| 530 |
+
"eval_steps_per_second": 3.928,
|
| 531 |
+
"step": 6350
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 51.0,
|
| 535 |
+
"eval_accuracy": 0.5595061728395062,
|
| 536 |
+
"eval_loss": 2.4946696758270264,
|
| 537 |
+
"eval_runtime": 16.2486,
|
| 538 |
+
"eval_samples_per_second": 124.626,
|
| 539 |
+
"eval_steps_per_second": 3.939,
|
| 540 |
+
"step": 6477
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 51.18,
|
| 544 |
+
"learning_rate": 9.802371541501976e-07,
|
| 545 |
+
"loss": 2.405,
|
| 546 |
+
"step": 6500
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 52.0,
|
| 550 |
+
"eval_accuracy": 0.5580246913580247,
|
| 551 |
+
"eval_loss": 2.4835174083709717,
|
| 552 |
+
"eval_runtime": 16.5375,
|
| 553 |
+
"eval_samples_per_second": 122.449,
|
| 554 |
+
"eval_steps_per_second": 3.87,
|
| 555 |
+
"step": 6604
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 53.0,
|
| 559 |
+
"eval_accuracy": 0.5585185185185185,
|
| 560 |
+
"eval_loss": 2.473555088043213,
|
| 561 |
+
"eval_runtime": 17.2437,
|
| 562 |
+
"eval_samples_per_second": 117.434,
|
| 563 |
+
"eval_steps_per_second": 3.711,
|
| 564 |
+
"step": 6731
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 54.0,
|
| 568 |
+
"eval_accuracy": 0.5644444444444444,
|
| 569 |
+
"eval_loss": 2.4599006175994873,
|
| 570 |
+
"eval_runtime": 17.2017,
|
| 571 |
+
"eval_samples_per_second": 117.721,
|
| 572 |
+
"eval_steps_per_second": 3.721,
|
| 573 |
+
"step": 6858
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 55.0,
|
| 577 |
+
"eval_accuracy": 0.5708641975308641,
|
| 578 |
+
"eval_loss": 2.444490432739258,
|
| 579 |
+
"eval_runtime": 16.4767,
|
| 580 |
+
"eval_samples_per_second": 122.901,
|
| 581 |
+
"eval_steps_per_second": 3.884,
|
| 582 |
+
"step": 6985
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 55.12,
|
| 586 |
+
"learning_rate": 9.011857707509881e-07,
|
| 587 |
+
"loss": 2.3499,
|
| 588 |
+
"step": 7000
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 56.0,
|
| 592 |
+
"eval_accuracy": 0.5718518518518518,
|
| 593 |
+
"eval_loss": 2.43546199798584,
|
| 594 |
+
"eval_runtime": 17.1098,
|
| 595 |
+
"eval_samples_per_second": 118.353,
|
| 596 |
+
"eval_steps_per_second": 3.741,
|
| 597 |
+
"step": 7112
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 57.0,
|
| 601 |
+
"eval_accuracy": 0.5738271604938272,
|
| 602 |
+
"eval_loss": 2.424128532409668,
|
| 603 |
+
"eval_runtime": 16.3356,
|
| 604 |
+
"eval_samples_per_second": 123.962,
|
| 605 |
+
"eval_steps_per_second": 3.918,
|
| 606 |
+
"step": 7239
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 58.0,
|
| 610 |
+
"eval_accuracy": 0.5679012345679012,
|
| 611 |
+
"eval_loss": 2.4162774085998535,
|
| 612 |
+
"eval_runtime": 16.3093,
|
| 613 |
+
"eval_samples_per_second": 124.162,
|
| 614 |
+
"eval_steps_per_second": 3.924,
|
| 615 |
+
"step": 7366
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 59.0,
|
| 619 |
+
"eval_accuracy": 0.5733333333333334,
|
| 620 |
+
"eval_loss": 2.404547929763794,
|
| 621 |
+
"eval_runtime": 16.3838,
|
| 622 |
+
"eval_samples_per_second": 123.598,
|
| 623 |
+
"eval_steps_per_second": 3.906,
|
| 624 |
+
"step": 7493
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 59.06,
|
| 628 |
+
"learning_rate": 8.221343873517787e-07,
|
| 629 |
+
"loss": 2.3047,
|
| 630 |
+
"step": 7500
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 60.0,
|
| 634 |
+
"eval_accuracy": 0.5767901234567901,
|
| 635 |
+
"eval_loss": 2.3955578804016113,
|
| 636 |
+
"eval_runtime": 16.4961,
|
| 637 |
+
"eval_samples_per_second": 122.756,
|
| 638 |
+
"eval_steps_per_second": 3.88,
|
| 639 |
+
"step": 7620
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 61.0,
|
| 643 |
+
"eval_accuracy": 0.5733333333333334,
|
| 644 |
+
"eval_loss": 2.3882274627685547,
|
| 645 |
+
"eval_runtime": 16.5941,
|
| 646 |
+
"eval_samples_per_second": 122.031,
|
| 647 |
+
"eval_steps_per_second": 3.857,
|
| 648 |
+
"step": 7747
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 62.0,
|
| 652 |
+
"eval_accuracy": 0.5758024691358025,
|
| 653 |
+
"eval_loss": 2.3768720626831055,
|
| 654 |
+
"eval_runtime": 16.3063,
|
| 655 |
+
"eval_samples_per_second": 124.185,
|
| 656 |
+
"eval_steps_per_second": 3.925,
|
| 657 |
+
"step": 7874
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 62.99,
|
| 661 |
+
"learning_rate": 7.430830039525692e-07,
|
| 662 |
+
"loss": 2.2616,
|
| 663 |
+
"step": 8000
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 63.0,
|
| 667 |
+
"eval_accuracy": 0.5802469135802469,
|
| 668 |
+
"eval_loss": 2.3649182319641113,
|
| 669 |
+
"eval_runtime": 16.4383,
|
| 670 |
+
"eval_samples_per_second": 123.188,
|
| 671 |
+
"eval_steps_per_second": 3.893,
|
| 672 |
+
"step": 8001
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 64.0,
|
| 676 |
+
"eval_accuracy": 0.5782716049382716,
|
| 677 |
+
"eval_loss": 2.3605239391326904,
|
| 678 |
+
"eval_runtime": 16.5048,
|
| 679 |
+
"eval_samples_per_second": 122.691,
|
| 680 |
+
"eval_steps_per_second": 3.878,
|
| 681 |
+
"step": 8128
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 65.0,
|
| 685 |
+
"eval_accuracy": 0.5881481481481482,
|
| 686 |
+
"eval_loss": 2.348599433898926,
|
| 687 |
+
"eval_runtime": 16.6986,
|
| 688 |
+
"eval_samples_per_second": 121.268,
|
| 689 |
+
"eval_steps_per_second": 3.833,
|
| 690 |
+
"step": 8255
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 66.0,
|
| 694 |
+
"eval_accuracy": 0.5891358024691358,
|
| 695 |
+
"eval_loss": 2.3403825759887695,
|
| 696 |
+
"eval_runtime": 16.6528,
|
| 697 |
+
"eval_samples_per_second": 121.601,
|
| 698 |
+
"eval_steps_per_second": 3.843,
|
| 699 |
+
"step": 8382
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 66.93,
|
| 703 |
+
"learning_rate": 6.640316205533597e-07,
|
| 704 |
+
"loss": 2.224,
|
| 705 |
+
"step": 8500
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 67.0,
|
| 709 |
+
"eval_accuracy": 0.5876543209876544,
|
| 710 |
+
"eval_loss": 2.333630323410034,
|
| 711 |
+
"eval_runtime": 16.323,
|
| 712 |
+
"eval_samples_per_second": 124.058,
|
| 713 |
+
"eval_steps_per_second": 3.921,
|
| 714 |
+
"step": 8509
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 68.0,
|
| 718 |
+
"eval_accuracy": 0.5881481481481482,
|
| 719 |
+
"eval_loss": 2.3267500400543213,
|
| 720 |
+
"eval_runtime": 16.3769,
|
| 721 |
+
"eval_samples_per_second": 123.649,
|
| 722 |
+
"eval_steps_per_second": 3.908,
|
| 723 |
+
"step": 8636
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 69.0,
|
| 727 |
+
"eval_accuracy": 0.5871604938271605,
|
| 728 |
+
"eval_loss": 2.3214409351348877,
|
| 729 |
+
"eval_runtime": 16.4152,
|
| 730 |
+
"eval_samples_per_second": 123.361,
|
| 731 |
+
"eval_steps_per_second": 3.899,
|
| 732 |
+
"step": 8763
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 70.0,
|
| 736 |
+
"eval_accuracy": 0.5920987654320987,
|
| 737 |
+
"eval_loss": 2.3111300468444824,
|
| 738 |
+
"eval_runtime": 16.7406,
|
| 739 |
+
"eval_samples_per_second": 120.963,
|
| 740 |
+
"eval_steps_per_second": 3.823,
|
| 741 |
+
"step": 8890
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 70.87,
|
| 745 |
+
"learning_rate": 5.849802371541502e-07,
|
| 746 |
+
"loss": 2.1885,
|
| 747 |
+
"step": 9000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 71.0,
|
| 751 |
+
"eval_accuracy": 0.5906172839506173,
|
| 752 |
+
"eval_loss": 2.304325580596924,
|
| 753 |
+
"eval_runtime": 16.5534,
|
| 754 |
+
"eval_samples_per_second": 122.331,
|
| 755 |
+
"eval_steps_per_second": 3.866,
|
| 756 |
+
"step": 9017
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 72.0,
|
| 760 |
+
"eval_accuracy": 0.5960493827160493,
|
| 761 |
+
"eval_loss": 2.297461986541748,
|
| 762 |
+
"eval_runtime": 16.4435,
|
| 763 |
+
"eval_samples_per_second": 123.149,
|
| 764 |
+
"eval_steps_per_second": 3.892,
|
| 765 |
+
"step": 9144
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 73.0,
|
| 769 |
+
"eval_accuracy": 0.5871604938271605,
|
| 770 |
+
"eval_loss": 2.296058177947998,
|
| 771 |
+
"eval_runtime": 16.3403,
|
| 772 |
+
"eval_samples_per_second": 123.927,
|
| 773 |
+
"eval_steps_per_second": 3.917,
|
| 774 |
+
"step": 9271
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 74.0,
|
| 778 |
+
"eval_accuracy": 0.5960493827160493,
|
| 779 |
+
"eval_loss": 2.2856781482696533,
|
| 780 |
+
"eval_runtime": 16.6488,
|
| 781 |
+
"eval_samples_per_second": 121.63,
|
| 782 |
+
"eval_steps_per_second": 3.844,
|
| 783 |
+
"step": 9398
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 74.8,
|
| 787 |
+
"learning_rate": 5.059288537549406e-07,
|
| 788 |
+
"loss": 2.1603,
|
| 789 |
+
"step": 9500
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 75.0,
|
| 793 |
+
"eval_accuracy": 0.5965432098765432,
|
| 794 |
+
"eval_loss": 2.2791759967803955,
|
| 795 |
+
"eval_runtime": 17.0015,
|
| 796 |
+
"eval_samples_per_second": 119.107,
|
| 797 |
+
"eval_steps_per_second": 3.764,
|
| 798 |
+
"step": 9525
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 76.0,
|
| 802 |
+
"eval_accuracy": 0.5955555555555555,
|
| 803 |
+
"eval_loss": 2.2750935554504395,
|
| 804 |
+
"eval_runtime": 16.6586,
|
| 805 |
+
"eval_samples_per_second": 121.559,
|
| 806 |
+
"eval_steps_per_second": 3.842,
|
| 807 |
+
"step": 9652
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 77.0,
|
| 811 |
+
"eval_accuracy": 0.5965432098765432,
|
| 812 |
+
"eval_loss": 2.268305540084839,
|
| 813 |
+
"eval_runtime": 16.6007,
|
| 814 |
+
"eval_samples_per_second": 121.983,
|
| 815 |
+
"eval_steps_per_second": 3.855,
|
| 816 |
+
"step": 9779
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 78.0,
|
| 820 |
+
"eval_accuracy": 0.6049382716049383,
|
| 821 |
+
"eval_loss": 2.2604947090148926,
|
| 822 |
+
"eval_runtime": 16.3684,
|
| 823 |
+
"eval_samples_per_second": 123.714,
|
| 824 |
+
"eval_steps_per_second": 3.91,
|
| 825 |
+
"step": 9906
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 78.74,
|
| 829 |
+
"learning_rate": 4.268774703557312e-07,
|
| 830 |
+
"loss": 2.1357,
|
| 831 |
+
"step": 10000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 79.0,
|
| 835 |
+
"eval_accuracy": 0.6064197530864197,
|
| 836 |
+
"eval_loss": 2.254981756210327,
|
| 837 |
+
"eval_runtime": 16.4537,
|
| 838 |
+
"eval_samples_per_second": 123.072,
|
| 839 |
+
"eval_steps_per_second": 3.89,
|
| 840 |
+
"step": 10033
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 80.0,
|
| 844 |
+
"eval_accuracy": 0.6034567901234568,
|
| 845 |
+
"eval_loss": 2.2524936199188232,
|
| 846 |
+
"eval_runtime": 16.3798,
|
| 847 |
+
"eval_samples_per_second": 123.628,
|
| 848 |
+
"eval_steps_per_second": 3.907,
|
| 849 |
+
"step": 10160
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 81.0,
|
| 853 |
+
"eval_accuracy": 0.6054320987654321,
|
| 854 |
+
"eval_loss": 2.2475554943084717,
|
| 855 |
+
"eval_runtime": 16.2339,
|
| 856 |
+
"eval_samples_per_second": 124.739,
|
| 857 |
+
"eval_steps_per_second": 3.942,
|
| 858 |
+
"step": 10287
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 82.0,
|
| 862 |
+
"eval_accuracy": 0.6004938271604938,
|
| 863 |
+
"eval_loss": 2.244779586791992,
|
| 864 |
+
"eval_runtime": 16.19,
|
| 865 |
+
"eval_samples_per_second": 125.077,
|
| 866 |
+
"eval_steps_per_second": 3.953,
|
| 867 |
+
"step": 10414
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 82.68,
|
| 871 |
+
"learning_rate": 3.478260869565217e-07,
|
| 872 |
+
"loss": 2.1153,
|
| 873 |
+
"step": 10500
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 83.0,
|
| 877 |
+
"eval_accuracy": 0.6009876543209877,
|
| 878 |
+
"eval_loss": 2.242079973220825,
|
| 879 |
+
"eval_runtime": 16.2447,
|
| 880 |
+
"eval_samples_per_second": 124.656,
|
| 881 |
+
"eval_steps_per_second": 3.94,
|
| 882 |
+
"step": 10541
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 84.0,
|
| 886 |
+
"eval_accuracy": 0.6009876543209877,
|
| 887 |
+
"eval_loss": 2.237128973007202,
|
| 888 |
+
"eval_runtime": 16.3552,
|
| 889 |
+
"eval_samples_per_second": 123.814,
|
| 890 |
+
"eval_steps_per_second": 3.913,
|
| 891 |
+
"step": 10668
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 85.0,
|
| 895 |
+
"eval_accuracy": 0.6034567901234568,
|
| 896 |
+
"eval_loss": 2.2331736087799072,
|
| 897 |
+
"eval_runtime": 16.4131,
|
| 898 |
+
"eval_samples_per_second": 123.377,
|
| 899 |
+
"eval_steps_per_second": 3.899,
|
| 900 |
+
"step": 10795
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 86.0,
|
| 904 |
+
"eval_accuracy": 0.6064197530864197,
|
| 905 |
+
"eval_loss": 2.230192184448242,
|
| 906 |
+
"eval_runtime": 16.263,
|
| 907 |
+
"eval_samples_per_second": 124.516,
|
| 908 |
+
"eval_steps_per_second": 3.935,
|
| 909 |
+
"step": 10922
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 86.61,
|
| 913 |
+
"learning_rate": 2.6877470355731227e-07,
|
| 914 |
+
"loss": 2.0969,
|
| 915 |
+
"step": 11000
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 87.0,
|
| 919 |
+
"eval_accuracy": 0.6083950617283951,
|
| 920 |
+
"eval_loss": 2.2269179821014404,
|
| 921 |
+
"eval_runtime": 16.4471,
|
| 922 |
+
"eval_samples_per_second": 123.122,
|
| 923 |
+
"eval_steps_per_second": 3.891,
|
| 924 |
+
"step": 11049
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 88.0,
|
| 928 |
+
"eval_accuracy": 0.6054320987654321,
|
| 929 |
+
"eval_loss": 2.224729537963867,
|
| 930 |
+
"eval_runtime": 16.5547,
|
| 931 |
+
"eval_samples_per_second": 122.322,
|
| 932 |
+
"eval_steps_per_second": 3.866,
|
| 933 |
+
"step": 11176
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 89.0,
|
| 937 |
+
"eval_accuracy": 0.6098765432098765,
|
| 938 |
+
"eval_loss": 2.2210581302642822,
|
| 939 |
+
"eval_runtime": 16.3636,
|
| 940 |
+
"eval_samples_per_second": 123.75,
|
| 941 |
+
"eval_steps_per_second": 3.911,
|
| 942 |
+
"step": 11303
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 90.0,
|
| 946 |
+
"eval_accuracy": 0.6049382716049383,
|
| 947 |
+
"eval_loss": 2.2199292182922363,
|
| 948 |
+
"eval_runtime": 16.5723,
|
| 949 |
+
"eval_samples_per_second": 122.192,
|
| 950 |
+
"eval_steps_per_second": 3.862,
|
| 951 |
+
"step": 11430
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 90.55,
|
| 955 |
+
"learning_rate": 1.8972332015810276e-07,
|
| 956 |
+
"loss": 2.0851,
|
| 957 |
+
"step": 11500
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 91.0,
|
| 961 |
+
"eval_accuracy": 0.6103703703703703,
|
| 962 |
+
"eval_loss": 2.2158267498016357,
|
| 963 |
+
"eval_runtime": 16.2141,
|
| 964 |
+
"eval_samples_per_second": 124.892,
|
| 965 |
+
"eval_steps_per_second": 3.947,
|
| 966 |
+
"step": 11557
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 92.0,
|
| 970 |
+
"eval_accuracy": 0.6079012345679012,
|
| 971 |
+
"eval_loss": 2.214700937271118,
|
| 972 |
+
"eval_runtime": 16.2859,
|
| 973 |
+
"eval_samples_per_second": 124.341,
|
| 974 |
+
"eval_steps_per_second": 3.93,
|
| 975 |
+
"step": 11684
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 93.0,
|
| 979 |
+
"eval_accuracy": 0.6064197530864197,
|
| 980 |
+
"eval_loss": 2.2131123542785645,
|
| 981 |
+
"eval_runtime": 16.337,
|
| 982 |
+
"eval_samples_per_second": 123.952,
|
| 983 |
+
"eval_steps_per_second": 3.917,
|
| 984 |
+
"step": 11811
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 94.0,
|
| 988 |
+
"eval_accuracy": 0.6098765432098765,
|
| 989 |
+
"eval_loss": 2.2114579677581787,
|
| 990 |
+
"eval_runtime": 16.393,
|
| 991 |
+
"eval_samples_per_second": 123.528,
|
| 992 |
+
"eval_steps_per_second": 3.904,
|
| 993 |
+
"step": 11938
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 94.49,
|
| 997 |
+
"learning_rate": 1.1067193675889327e-07,
|
| 998 |
+
"loss": 2.0754,
|
| 999 |
+
"step": 12000
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 95.0,
|
| 1003 |
+
"eval_accuracy": 0.6138271604938271,
|
| 1004 |
+
"eval_loss": 2.209113597869873,
|
| 1005 |
+
"eval_runtime": 16.4652,
|
| 1006 |
+
"eval_samples_per_second": 122.987,
|
| 1007 |
+
"eval_steps_per_second": 3.887,
|
| 1008 |
+
"step": 12065
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 96.0,
|
| 1012 |
+
"eval_accuracy": 0.6128395061728396,
|
| 1013 |
+
"eval_loss": 2.2085654735565186,
|
| 1014 |
+
"eval_runtime": 16.3862,
|
| 1015 |
+
"eval_samples_per_second": 123.579,
|
| 1016 |
+
"eval_steps_per_second": 3.906,
|
| 1017 |
+
"step": 12192
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 97.0,
|
| 1021 |
+
"eval_accuracy": 0.6123456790123457,
|
| 1022 |
+
"eval_loss": 2.207934617996216,
|
| 1023 |
+
"eval_runtime": 16.3381,
|
| 1024 |
+
"eval_samples_per_second": 123.943,
|
| 1025 |
+
"eval_steps_per_second": 3.917,
|
| 1026 |
+
"step": 12319
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 98.0,
|
| 1030 |
+
"eval_accuracy": 0.6128395061728396,
|
| 1031 |
+
"eval_loss": 2.206979513168335,
|
| 1032 |
+
"eval_runtime": 16.3485,
|
| 1033 |
+
"eval_samples_per_second": 123.865,
|
| 1034 |
+
"eval_steps_per_second": 3.915,
|
| 1035 |
+
"step": 12446
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 98.43,
|
| 1039 |
+
"learning_rate": 3.162055335968379e-08,
|
| 1040 |
+
"loss": 2.0667,
|
| 1041 |
+
"step": 12500
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 99.0,
|
| 1045 |
+
"eval_accuracy": 0.6133333333333333,
|
| 1046 |
+
"eval_loss": 2.2068374156951904,
|
| 1047 |
+
"eval_runtime": 16.2194,
|
| 1048 |
+
"eval_samples_per_second": 124.851,
|
| 1049 |
+
"eval_steps_per_second": 3.946,
|
| 1050 |
+
"step": 12573
|
| 1051 |
+
}
|
| 1052 |
+
],
|
| 1053 |
+
"logging_steps": 500,
|
| 1054 |
+
"max_steps": 12700,
|
| 1055 |
+
"num_train_epochs": 100,
|
| 1056 |
+
"save_steps": 500,
|
| 1057 |
+
"total_flos": 6.215388042176161e+19,
|
| 1058 |
+
"trial_name": null,
|
| 1059 |
+
"trial_params": null
|
| 1060 |
+
}
|
checkpoint-12573/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:228f29b6a2e9560079c65c3416fd8bcbe6376471ac08b36d341233494efae6e3
|
| 3 |
+
size 4091
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 343380860
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e67fe27f4a357a2b8bce1a9e565ac9aa096eb5fed005739ce9dd866069f41146
|
| 3 |
size 343380860
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4091
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:228f29b6a2e9560079c65c3416fd8bcbe6376471ac08b36d341233494efae6e3
|
| 3 |
size 4091
|