lombardata commited on
Commit
4834c01
1 Parent(s): 840b521

Evaluation on the test set completed on 2024_11_03.

Browse files
README.md ADDED
@@ -0,0 +1,151 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: facebook/dinov2-large
4
+ tags:
5
+ - generated_from_trainer
6
+ metrics:
7
+ - accuracy
8
+ model-index:
9
+ - name: drone-DinoVdeau-produttoria_binary-binary-large-2024_11_03-batch-size64_freeze
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ # drone-DinoVdeau-produttoria_binary-binary-large-2024_11_03-batch-size64_freeze
17
+
18
+ This model is a fine-tuned version of [facebook/dinov2-large](https://huggingface.co/facebook/dinov2-large) on the None dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 0.2854
21
+ - F1 Micro: 0.8468
22
+ - F1 Macro: 0.6351
23
+ - Accuracy: 0.2786
24
+ - Learning Rate: 0.0000
25
+
26
+ ## Model description
27
+
28
+ More information needed
29
+
30
+ ## Intended uses & limitations
31
+
32
+ More information needed
33
+
34
+ ## Training and evaluation data
35
+
36
+ More information needed
37
+
38
+ ## Training procedure
39
+
40
+ ### Training hyperparameters
41
+
42
+ The following hyperparameters were used during training:
43
+ - learning_rate: 0.001
44
+ - train_batch_size: 64
45
+ - eval_batch_size: 64
46
+ - seed: 42
47
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
48
+ - lr_scheduler_type: linear
49
+ - num_epochs: 150
50
+ - mixed_precision_training: Native AMP
51
+
52
+ ### Training results
53
+
54
+ | Training Loss | Epoch | Step | Validation Loss | F1 Micro | F1 Macro | Accuracy | Rate |
55
+ |:-------------:|:-----:|:-----:|:---------------:|:--------:|:--------:|:--------:|:------:|
56
+ | No log | 1.0 | 181 | 0.3236 | 0.8262 | 0.5774 | 0.2630 | 0.001 |
57
+ | No log | 2.0 | 362 | 0.3146 | 0.8379 | 0.6199 | 0.2412 | 0.001 |
58
+ | 0.3995 | 3.0 | 543 | 0.3090 | 0.8398 | 0.6044 | 0.2555 | 0.001 |
59
+ | 0.3995 | 4.0 | 724 | 0.3074 | 0.8349 | 0.6003 | 0.2562 | 0.001 |
60
+ | 0.3995 | 5.0 | 905 | 0.3039 | 0.8406 | 0.6248 | 0.2516 | 0.001 |
61
+ | 0.3299 | 6.0 | 1086 | 0.3060 | 0.8420 | 0.6225 | 0.2596 | 0.001 |
62
+ | 0.3299 | 7.0 | 1267 | 0.3014 | 0.8387 | 0.5955 | 0.2820 | 0.001 |
63
+ | 0.3299 | 8.0 | 1448 | 0.3013 | 0.8391 | 0.5975 | 0.2703 | 0.001 |
64
+ | 0.3216 | 9.0 | 1629 | 0.3010 | 0.8407 | 0.5974 | 0.2841 | 0.001 |
65
+ | 0.3216 | 10.0 | 1810 | 0.3007 | 0.8376 | 0.5938 | 0.2711 | 0.001 |
66
+ | 0.3216 | 11.0 | 1991 | 0.3036 | 0.8349 | 0.5762 | 0.2773 | 0.001 |
67
+ | 0.3167 | 12.0 | 2172 | 0.3013 | 0.8385 | 0.6115 | 0.2674 | 0.001 |
68
+ | 0.3167 | 13.0 | 2353 | 0.2978 | 0.8421 | 0.6146 | 0.2648 | 0.001 |
69
+ | 0.315 | 14.0 | 2534 | 0.2977 | 0.8400 | 0.6059 | 0.2734 | 0.001 |
70
+ | 0.315 | 15.0 | 2715 | 0.2981 | 0.8434 | 0.6075 | 0.2666 | 0.001 |
71
+ | 0.315 | 16.0 | 2896 | 0.2974 | 0.8394 | 0.5933 | 0.2747 | 0.001 |
72
+ | 0.3147 | 17.0 | 3077 | 0.2984 | 0.8438 | 0.6147 | 0.2664 | 0.001 |
73
+ | 0.3147 | 18.0 | 3258 | 0.3023 | 0.8356 | 0.5804 | 0.2763 | 0.001 |
74
+ | 0.3147 | 19.0 | 3439 | 0.2985 | 0.8424 | 0.6159 | 0.2739 | 0.001 |
75
+ | 0.3122 | 20.0 | 3620 | 0.2968 | 0.8412 | 0.5984 | 0.2807 | 0.001 |
76
+ | 0.3122 | 21.0 | 3801 | 0.3005 | 0.8419 | 0.6060 | 0.2703 | 0.001 |
77
+ | 0.3122 | 22.0 | 3982 | 0.2982 | 0.8375 | 0.5804 | 0.2747 | 0.001 |
78
+ | 0.3149 | 23.0 | 4163 | 0.2939 | 0.8436 | 0.6152 | 0.2781 | 0.001 |
79
+ | 0.3149 | 24.0 | 4344 | 0.2948 | 0.8453 | 0.6229 | 0.2760 | 0.001 |
80
+ | 0.3118 | 25.0 | 4525 | 0.2968 | 0.8427 | 0.6103 | 0.2737 | 0.001 |
81
+ | 0.3118 | 26.0 | 4706 | 0.2956 | 0.8421 | 0.6045 | 0.2755 | 0.001 |
82
+ | 0.3118 | 27.0 | 4887 | 0.2959 | 0.8438 | 0.6115 | 0.2765 | 0.001 |
83
+ | 0.3126 | 28.0 | 5068 | 0.2955 | 0.8447 | 0.6191 | 0.2693 | 0.001 |
84
+ | 0.3126 | 29.0 | 5249 | 0.3011 | 0.8438 | 0.6216 | 0.2664 | 0.001 |
85
+ | 0.3126 | 30.0 | 5430 | 0.2921 | 0.8437 | 0.6025 | 0.2810 | 0.0001 |
86
+ | 0.3093 | 31.0 | 5611 | 0.2904 | 0.8439 | 0.6072 | 0.2812 | 0.0001 |
87
+ | 0.3093 | 32.0 | 5792 | 0.2903 | 0.8437 | 0.6112 | 0.2810 | 0.0001 |
88
+ | 0.3093 | 33.0 | 5973 | 0.2889 | 0.8462 | 0.6202 | 0.2854 | 0.0001 |
89
+ | 0.3049 | 34.0 | 6154 | 0.2896 | 0.8446 | 0.6151 | 0.2862 | 0.0001 |
90
+ | 0.3049 | 35.0 | 6335 | 0.2887 | 0.8449 | 0.6112 | 0.2867 | 0.0001 |
91
+ | 0.3012 | 36.0 | 6516 | 0.2889 | 0.8447 | 0.6120 | 0.2836 | 0.0001 |
92
+ | 0.3012 | 37.0 | 6697 | 0.2883 | 0.8476 | 0.6256 | 0.2867 | 0.0001 |
93
+ | 0.3012 | 38.0 | 6878 | 0.2905 | 0.8453 | 0.6057 | 0.2825 | 0.0001 |
94
+ | 0.299 | 39.0 | 7059 | 0.2878 | 0.8471 | 0.6254 | 0.2854 | 0.0001 |
95
+ | 0.299 | 40.0 | 7240 | 0.2886 | 0.8468 | 0.6223 | 0.2810 | 0.0001 |
96
+ | 0.299 | 41.0 | 7421 | 0.2877 | 0.8473 | 0.6261 | 0.2843 | 0.0001 |
97
+ | 0.2989 | 42.0 | 7602 | 0.2878 | 0.8477 | 0.6199 | 0.2856 | 0.0001 |
98
+ | 0.2989 | 43.0 | 7783 | 0.2872 | 0.8479 | 0.6288 | 0.2830 | 0.0001 |
99
+ | 0.2989 | 44.0 | 7964 | 0.2868 | 0.8464 | 0.6190 | 0.2841 | 0.0001 |
100
+ | 0.2983 | 45.0 | 8145 | 0.2870 | 0.8463 | 0.6236 | 0.2838 | 0.0001 |
101
+ | 0.2983 | 46.0 | 8326 | 0.2868 | 0.8460 | 0.6151 | 0.2825 | 0.0001 |
102
+ | 0.298 | 47.0 | 8507 | 0.2872 | 0.8462 | 0.6211 | 0.2846 | 0.0001 |
103
+ | 0.298 | 48.0 | 8688 | 0.2866 | 0.8467 | 0.6231 | 0.2836 | 0.0001 |
104
+ | 0.298 | 49.0 | 8869 | 0.2863 | 0.8460 | 0.6161 | 0.2859 | 0.0001 |
105
+ | 0.2965 | 50.0 | 9050 | 0.2864 | 0.8483 | 0.6255 | 0.2846 | 0.0001 |
106
+ | 0.2965 | 51.0 | 9231 | 0.2891 | 0.8486 | 0.6278 | 0.2849 | 0.0001 |
107
+ | 0.2965 | 52.0 | 9412 | 0.2856 | 0.8464 | 0.6255 | 0.2851 | 0.0001 |
108
+ | 0.2956 | 53.0 | 9593 | 0.2872 | 0.8490 | 0.6458 | 0.2789 | 0.0001 |
109
+ | 0.2956 | 54.0 | 9774 | 0.2856 | 0.8477 | 0.6244 | 0.2903 | 0.0001 |
110
+ | 0.2956 | 55.0 | 9955 | 0.2857 | 0.8475 | 0.6340 | 0.2846 | 0.0001 |
111
+ | 0.2958 | 56.0 | 10136 | 0.2862 | 0.8466 | 0.6241 | 0.2867 | 0.0001 |
112
+ | 0.2958 | 57.0 | 10317 | 0.2871 | 0.8454 | 0.6249 | 0.2862 | 0.0001 |
113
+ | 0.2958 | 58.0 | 10498 | 0.2858 | 0.8492 | 0.6334 | 0.2812 | 0.0001 |
114
+ | 0.2954 | 59.0 | 10679 | 0.2862 | 0.8468 | 0.6178 | 0.2888 | 1e-05 |
115
+ | 0.2954 | 60.0 | 10860 | 0.2847 | 0.8485 | 0.6276 | 0.2854 | 1e-05 |
116
+ | 0.2923 | 61.0 | 11041 | 0.2849 | 0.8480 | 0.6224 | 0.2830 | 1e-05 |
117
+ | 0.2923 | 62.0 | 11222 | 0.2855 | 0.8469 | 0.6248 | 0.2843 | 1e-05 |
118
+ | 0.2923 | 63.0 | 11403 | 0.2849 | 0.8489 | 0.6275 | 0.2828 | 1e-05 |
119
+ | 0.2918 | 64.0 | 11584 | 0.2846 | 0.8475 | 0.6371 | 0.2823 | 1e-05 |
120
+ | 0.2918 | 65.0 | 11765 | 0.2860 | 0.8468 | 0.6241 | 0.2869 | 1e-05 |
121
+ | 0.2918 | 66.0 | 11946 | 0.2847 | 0.8481 | 0.6347 | 0.2841 | 1e-05 |
122
+ | 0.2906 | 67.0 | 12127 | 0.2853 | 0.8488 | 0.6287 | 0.2854 | 1e-05 |
123
+ | 0.2906 | 68.0 | 12308 | 0.2853 | 0.8480 | 0.6321 | 0.2867 | 1e-05 |
124
+ | 0.2906 | 69.0 | 12489 | 0.2848 | 0.8477 | 0.6397 | 0.2836 | 1e-05 |
125
+ | 0.2918 | 70.0 | 12670 | 0.2853 | 0.8492 | 0.6381 | 0.2823 | 1e-05 |
126
+ | 0.2918 | 71.0 | 12851 | 0.2851 | 0.8476 | 0.6325 | 0.2882 | 0.0000 |
127
+ | 0.2918 | 72.0 | 13032 | 0.2845 | 0.8474 | 0.6236 | 0.2849 | 0.0000 |
128
+ | 0.2918 | 73.0 | 13213 | 0.2845 | 0.8476 | 0.6333 | 0.2812 | 0.0000 |
129
+ | 0.2918 | 74.0 | 13394 | 0.2845 | 0.8466 | 0.6300 | 0.2828 | 0.0000 |
130
+ | 0.2913 | 75.0 | 13575 | 0.2851 | 0.8474 | 0.6235 | 0.2820 | 0.0000 |
131
+ | 0.2913 | 76.0 | 13756 | 0.2860 | 0.8473 | 0.6186 | 0.2880 | 0.0000 |
132
+ | 0.2913 | 77.0 | 13937 | 0.2858 | 0.8459 | 0.6173 | 0.2856 | 0.0000 |
133
+ | 0.2913 | 78.0 | 14118 | 0.2844 | 0.8481 | 0.6326 | 0.2843 | 0.0000 |
134
+ | 0.2913 | 79.0 | 14299 | 0.2871 | 0.8472 | 0.6179 | 0.2875 | 0.0000 |
135
+ | 0.2913 | 80.0 | 14480 | 0.2848 | 0.8477 | 0.6287 | 0.2838 | 0.0000 |
136
+ | 0.2915 | 81.0 | 14661 | 0.2848 | 0.8490 | 0.6305 | 0.2854 | 0.0000 |
137
+ | 0.2915 | 82.0 | 14842 | 0.2851 | 0.8480 | 0.6394 | 0.2859 | 0.0000 |
138
+ | 0.2913 | 83.0 | 15023 | 0.2846 | 0.8488 | 0.6255 | 0.2856 | 0.0000 |
139
+ | 0.2913 | 84.0 | 15204 | 0.2857 | 0.8482 | 0.6458 | 0.2833 | 0.0000 |
140
+ | 0.2913 | 85.0 | 15385 | 0.2855 | 0.8488 | 0.6340 | 0.2812 | 0.0000 |
141
+ | 0.2922 | 86.0 | 15566 | 0.2849 | 0.8480 | 0.6363 | 0.2859 | 0.0000 |
142
+ | 0.2922 | 87.0 | 15747 | 0.2845 | 0.8474 | 0.6328 | 0.2851 | 0.0000 |
143
+ | 0.2922 | 88.0 | 15928 | 0.2854 | 0.8478 | 0.6371 | 0.2812 | 0.0000 |
144
+
145
+
146
+ ### Framework versions
147
+
148
+ - Transformers 4.41.0
149
+ - Pytorch 2.5.0+cu124
150
+ - Datasets 3.0.2
151
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 88.0,
3
+ "eval_accuracy": 0.2786458333333333,
4
+ "eval_f1_macro": 0.6350734566442581,
5
+ "eval_f1_micro": 0.8467861336600807,
6
+ "eval_loss": 0.28541260957717896,
7
+ "eval_runtime": 76.7857,
8
+ "eval_samples_per_second": 50.009,
9
+ "eval_steps_per_second": 0.781,
10
+ "learning_rate": 1.0000000000000002e-07,
11
+ "total_flos": 1.500719176717825e+20,
12
+ "train_loss": 0.3047179739897961,
13
+ "train_runtime": 30841.8451,
14
+ "train_samples_per_second": 56.038,
15
+ "train_steps_per_second": 0.88
16
+ }
logs/events.out.tfevents.1730642710.datavisu2 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3fa5bbf2f190fb02623956ae5a2ee043ab21ec69339490ed418ebf0293c564d9
3
- size 54523
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1153ede07b604838b374d477545613ec02b5fef26fc96b8b0af1ba3fabb69be
3
+ size 55905
logs/events.out.tfevents.1730673734.datavisu2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d9f10a9c09da967e73fe949bac0d9f1f288eca3b0e5fa21454b3fe17c038fda
3
+ size 40
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:14fd3ab29f6baf6bf175d13021f237caacac090f24f91ffd71f1afcc5747ce0e
3
  size 1222958756
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8ec6d0a15d562cc3f4d35744dd773595c117cff73cc265a352ca85718ba80955
3
  size 1222958756
test_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 88.0,
3
+ "eval_accuracy": 0.2786458333333333,
4
+ "eval_f1_macro": 0.6350734566442581,
5
+ "eval_f1_micro": 0.8467861336600807,
6
+ "eval_loss": 0.28541260957717896,
7
+ "eval_runtime": 76.7857,
8
+ "eval_samples_per_second": 50.009,
9
+ "eval_steps_per_second": 0.781,
10
+ "learning_rate": 1.0000000000000002e-07
11
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 88.0,
3
+ "learning_rate": 1.0000000000000002e-07,
4
+ "total_flos": 1.500719176717825e+20,
5
+ "train_loss": 0.3047179739897961,
6
+ "train_runtime": 30841.8451,
7
+ "train_samples_per_second": 56.038,
8
+ "train_steps_per_second": 0.88
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,1325 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.28438833355903625,
3
+ "best_model_checkpoint": "/home/datawork-iot-nos/Seatizen/models/multilabel/drone/drone-DinoVdeau-produttoria_binary-binary-large-2024_11_03-batch-size64_freeze/checkpoint-14118",
4
+ "epoch": 88.0,
5
+ "eval_steps": 500,
6
+ "global_step": 15928,
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": 0.2630072840790843,
14
+ "eval_f1_macro": 0.5774239185038708,
15
+ "eval_f1_micro": 0.8262109753225342,
16
+ "eval_loss": 0.3235681354999542,
17
+ "eval_runtime": 79.5924,
18
+ "eval_samples_per_second": 48.296,
19
+ "eval_steps_per_second": 0.766,
20
+ "learning_rate": 0.001,
21
+ "step": 181
22
+ },
23
+ {
24
+ "epoch": 2.0,
25
+ "eval_accuracy": 0.24115504682622269,
26
+ "eval_f1_macro": 0.6199165901601139,
27
+ "eval_f1_micro": 0.8378565084377776,
28
+ "eval_loss": 0.3146470785140991,
29
+ "eval_runtime": 78.3589,
30
+ "eval_samples_per_second": 49.056,
31
+ "eval_steps_per_second": 0.778,
32
+ "learning_rate": 0.001,
33
+ "step": 362
34
+ },
35
+ {
36
+ "epoch": 2.7624309392265194,
37
+ "grad_norm": 0.39133042097091675,
38
+ "learning_rate": 0.001,
39
+ "loss": 0.3995,
40
+ "step": 500
41
+ },
42
+ {
43
+ "epoch": 3.0,
44
+ "eval_accuracy": 0.2554630593132154,
45
+ "eval_f1_macro": 0.6043570009634397,
46
+ "eval_f1_micro": 0.8398465111582348,
47
+ "eval_loss": 0.3090434670448303,
48
+ "eval_runtime": 77.1271,
49
+ "eval_samples_per_second": 49.84,
50
+ "eval_steps_per_second": 0.791,
51
+ "learning_rate": 0.001,
52
+ "step": 543
53
+ },
54
+ {
55
+ "epoch": 4.0,
56
+ "eval_accuracy": 0.25624349635796045,
57
+ "eval_f1_macro": 0.600278483167516,
58
+ "eval_f1_micro": 0.8348980169243037,
59
+ "eval_loss": 0.30735355615615845,
60
+ "eval_runtime": 77.9527,
61
+ "eval_samples_per_second": 49.312,
62
+ "eval_steps_per_second": 0.783,
63
+ "learning_rate": 0.001,
64
+ "step": 724
65
+ },
66
+ {
67
+ "epoch": 5.0,
68
+ "eval_accuracy": 0.2515608740894901,
69
+ "eval_f1_macro": 0.6247746971203368,
70
+ "eval_f1_micro": 0.8405948994360434,
71
+ "eval_loss": 0.30385810136795044,
72
+ "eval_runtime": 79.3747,
73
+ "eval_samples_per_second": 48.429,
74
+ "eval_steps_per_second": 0.769,
75
+ "learning_rate": 0.001,
76
+ "step": 905
77
+ },
78
+ {
79
+ "epoch": 5.524861878453039,
80
+ "grad_norm": 0.30673518776893616,
81
+ "learning_rate": 0.001,
82
+ "loss": 0.3299,
83
+ "step": 1000
84
+ },
85
+ {
86
+ "epoch": 6.0,
87
+ "eval_accuracy": 0.2596253902185224,
88
+ "eval_f1_macro": 0.6225111439021958,
89
+ "eval_f1_micro": 0.841987466427932,
90
+ "eval_loss": 0.3059956729412079,
91
+ "eval_runtime": 78.3475,
92
+ "eval_samples_per_second": 49.063,
93
+ "eval_steps_per_second": 0.779,
94
+ "learning_rate": 0.001,
95
+ "step": 1086
96
+ },
97
+ {
98
+ "epoch": 7.0,
99
+ "eval_accuracy": 0.28199791883454733,
100
+ "eval_f1_macro": 0.5954695621655504,
101
+ "eval_f1_micro": 0.8387498056289846,
102
+ "eval_loss": 0.3013758361339569,
103
+ "eval_runtime": 77.8519,
104
+ "eval_samples_per_second": 49.376,
105
+ "eval_steps_per_second": 0.784,
106
+ "learning_rate": 0.001,
107
+ "step": 1267
108
+ },
109
+ {
110
+ "epoch": 8.0,
111
+ "eval_accuracy": 0.2702913631633715,
112
+ "eval_f1_macro": 0.5974832028652961,
113
+ "eval_f1_micro": 0.8390550208451284,
114
+ "eval_loss": 0.30131709575653076,
115
+ "eval_runtime": 76.3356,
116
+ "eval_samples_per_second": 50.357,
117
+ "eval_steps_per_second": 0.799,
118
+ "learning_rate": 0.001,
119
+ "step": 1448
120
+ },
121
+ {
122
+ "epoch": 8.287292817679559,
123
+ "grad_norm": 0.2473623901605606,
124
+ "learning_rate": 0.001,
125
+ "loss": 0.3216,
126
+ "step": 1500
127
+ },
128
+ {
129
+ "epoch": 9.0,
130
+ "eval_accuracy": 0.28407908428720086,
131
+ "eval_f1_macro": 0.5974259992816957,
132
+ "eval_f1_micro": 0.8406665130922214,
133
+ "eval_loss": 0.30098479986190796,
134
+ "eval_runtime": 76.6263,
135
+ "eval_samples_per_second": 50.166,
136
+ "eval_steps_per_second": 0.796,
137
+ "learning_rate": 0.001,
138
+ "step": 1629
139
+ },
140
+ {
141
+ "epoch": 10.0,
142
+ "eval_accuracy": 0.27107180020811655,
143
+ "eval_f1_macro": 0.5937940362628795,
144
+ "eval_f1_micro": 0.8376187886791475,
145
+ "eval_loss": 0.30072343349456787,
146
+ "eval_runtime": 78.1999,
147
+ "eval_samples_per_second": 49.156,
148
+ "eval_steps_per_second": 0.78,
149
+ "learning_rate": 0.001,
150
+ "step": 1810
151
+ },
152
+ {
153
+ "epoch": 11.0,
154
+ "eval_accuracy": 0.277315296566077,
155
+ "eval_f1_macro": 0.5761905737205768,
156
+ "eval_f1_micro": 0.8348592565387339,
157
+ "eval_loss": 0.3035621643066406,
158
+ "eval_runtime": 77.9385,
159
+ "eval_samples_per_second": 49.321,
160
+ "eval_steps_per_second": 0.783,
161
+ "learning_rate": 0.001,
162
+ "step": 1991
163
+ },
164
+ {
165
+ "epoch": 11.049723756906078,
166
+ "grad_norm": 0.24565815925598145,
167
+ "learning_rate": 0.001,
168
+ "loss": 0.3167,
169
+ "step": 2000
170
+ },
171
+ {
172
+ "epoch": 12.0,
173
+ "eval_accuracy": 0.26742976066597296,
174
+ "eval_f1_macro": 0.6114755503631268,
175
+ "eval_f1_micro": 0.838466245156027,
176
+ "eval_loss": 0.3012838363647461,
177
+ "eval_runtime": 77.6902,
178
+ "eval_samples_per_second": 49.479,
179
+ "eval_steps_per_second": 0.785,
180
+ "learning_rate": 0.001,
181
+ "step": 2172
182
+ },
183
+ {
184
+ "epoch": 13.0,
185
+ "eval_accuracy": 0.2648283038501561,
186
+ "eval_f1_macro": 0.6145726431106396,
187
+ "eval_f1_micro": 0.8421213122252433,
188
+ "eval_loss": 0.29778778553009033,
189
+ "eval_runtime": 77.1253,
190
+ "eval_samples_per_second": 49.841,
191
+ "eval_steps_per_second": 0.791,
192
+ "learning_rate": 0.001,
193
+ "step": 2353
194
+ },
195
+ {
196
+ "epoch": 13.812154696132596,
197
+ "grad_norm": 0.2421799898147583,
198
+ "learning_rate": 0.001,
199
+ "loss": 0.315,
200
+ "step": 2500
201
+ },
202
+ {
203
+ "epoch": 14.0,
204
+ "eval_accuracy": 0.27341311134235174,
205
+ "eval_f1_macro": 0.605884177295118,
206
+ "eval_f1_micro": 0.8399742101869762,
207
+ "eval_loss": 0.29774588346481323,
208
+ "eval_runtime": 76.619,
209
+ "eval_samples_per_second": 50.17,
210
+ "eval_steps_per_second": 0.796,
211
+ "learning_rate": 0.001,
212
+ "step": 2534
213
+ },
214
+ {
215
+ "epoch": 15.0,
216
+ "eval_accuracy": 0.2666493236212279,
217
+ "eval_f1_macro": 0.6074624445346274,
218
+ "eval_f1_micro": 0.8433503513117323,
219
+ "eval_loss": 0.29809942841529846,
220
+ "eval_runtime": 77.2846,
221
+ "eval_samples_per_second": 49.738,
222
+ "eval_steps_per_second": 0.789,
223
+ "learning_rate": 0.001,
224
+ "step": 2715
225
+ },
226
+ {
227
+ "epoch": 16.0,
228
+ "eval_accuracy": 0.27471383975026015,
229
+ "eval_f1_macro": 0.5932952143692389,
230
+ "eval_f1_micro": 0.8394100355835181,
231
+ "eval_loss": 0.29744812846183777,
232
+ "eval_runtime": 77.1329,
233
+ "eval_samples_per_second": 49.836,
234
+ "eval_steps_per_second": 0.791,
235
+ "learning_rate": 0.001,
236
+ "step": 2896
237
+ },
238
+ {
239
+ "epoch": 16.574585635359117,
240
+ "grad_norm": 0.2012377828359604,
241
+ "learning_rate": 0.001,
242
+ "loss": 0.3147,
243
+ "step": 3000
244
+ },
245
+ {
246
+ "epoch": 17.0,
247
+ "eval_accuracy": 0.2663891779396462,
248
+ "eval_f1_macro": 0.6146867059353278,
249
+ "eval_f1_micro": 0.8437578624264077,
250
+ "eval_loss": 0.2983638644218445,
251
+ "eval_runtime": 77.7728,
252
+ "eval_samples_per_second": 49.426,
253
+ "eval_steps_per_second": 0.784,
254
+ "learning_rate": 0.001,
255
+ "step": 3077
256
+ },
257
+ {
258
+ "epoch": 18.0,
259
+ "eval_accuracy": 0.2762747138397503,
260
+ "eval_f1_macro": 0.5803903225868541,
261
+ "eval_f1_micro": 0.8356339535005088,
262
+ "eval_loss": 0.3023049235343933,
263
+ "eval_runtime": 77.2379,
264
+ "eval_samples_per_second": 49.768,
265
+ "eval_steps_per_second": 0.79,
266
+ "learning_rate": 0.001,
267
+ "step": 3258
268
+ },
269
+ {
270
+ "epoch": 19.0,
271
+ "eval_accuracy": 0.2739334027055151,
272
+ "eval_f1_macro": 0.6158875389283108,
273
+ "eval_f1_micro": 0.8423529411764706,
274
+ "eval_loss": 0.2984697222709656,
275
+ "eval_runtime": 76.7734,
276
+ "eval_samples_per_second": 50.069,
277
+ "eval_steps_per_second": 0.795,
278
+ "learning_rate": 0.001,
279
+ "step": 3439
280
+ },
281
+ {
282
+ "epoch": 19.337016574585636,
283
+ "grad_norm": 0.20086592435836792,
284
+ "learning_rate": 0.001,
285
+ "loss": 0.3122,
286
+ "step": 3500
287
+ },
288
+ {
289
+ "epoch": 20.0,
290
+ "eval_accuracy": 0.28069719042663893,
291
+ "eval_f1_macro": 0.5984147849283556,
292
+ "eval_f1_micro": 0.8411767731317183,
293
+ "eval_loss": 0.29680272936820984,
294
+ "eval_runtime": 77.2197,
295
+ "eval_samples_per_second": 49.78,
296
+ "eval_steps_per_second": 0.79,
297
+ "learning_rate": 0.001,
298
+ "step": 3620
299
+ },
300
+ {
301
+ "epoch": 21.0,
302
+ "eval_accuracy": 0.2702913631633715,
303
+ "eval_f1_macro": 0.6060492619397649,
304
+ "eval_f1_micro": 0.8418969323285377,
305
+ "eval_loss": 0.30051520466804504,
306
+ "eval_runtime": 76.7663,
307
+ "eval_samples_per_second": 50.074,
308
+ "eval_steps_per_second": 0.795,
309
+ "learning_rate": 0.001,
310
+ "step": 3801
311
+ },
312
+ {
313
+ "epoch": 22.0,
314
+ "eval_accuracy": 0.27471383975026015,
315
+ "eval_f1_macro": 0.580353532272699,
316
+ "eval_f1_micro": 0.8374817746302854,
317
+ "eval_loss": 0.29818177223205566,
318
+ "eval_runtime": 77.0688,
319
+ "eval_samples_per_second": 49.877,
320
+ "eval_steps_per_second": 0.792,
321
+ "learning_rate": 0.001,
322
+ "step": 3982
323
+ },
324
+ {
325
+ "epoch": 22.099447513812155,
326
+ "grad_norm": 0.21317744255065918,
327
+ "learning_rate": 0.001,
328
+ "loss": 0.3149,
329
+ "step": 4000
330
+ },
331
+ {
332
+ "epoch": 23.0,
333
+ "eval_accuracy": 0.27809573361082207,
334
+ "eval_f1_macro": 0.615237110287355,
335
+ "eval_f1_micro": 0.8436262061960386,
336
+ "eval_loss": 0.29393449425697327,
337
+ "eval_runtime": 76.8217,
338
+ "eval_samples_per_second": 50.038,
339
+ "eval_steps_per_second": 0.794,
340
+ "learning_rate": 0.001,
341
+ "step": 4163
342
+ },
343
+ {
344
+ "epoch": 24.0,
345
+ "eval_accuracy": 0.27601456815816855,
346
+ "eval_f1_macro": 0.6228721497006335,
347
+ "eval_f1_micro": 0.8453232862164007,
348
+ "eval_loss": 0.2948347330093384,
349
+ "eval_runtime": 77.3517,
350
+ "eval_samples_per_second": 49.695,
351
+ "eval_steps_per_second": 0.789,
352
+ "learning_rate": 0.001,
353
+ "step": 4344
354
+ },
355
+ {
356
+ "epoch": 24.861878453038674,
357
+ "grad_norm": 0.17206734418869019,
358
+ "learning_rate": 0.001,
359
+ "loss": 0.3118,
360
+ "step": 4500
361
+ },
362
+ {
363
+ "epoch": 25.0,
364
+ "eval_accuracy": 0.2736732570239334,
365
+ "eval_f1_macro": 0.610255370235793,
366
+ "eval_f1_micro": 0.8427456149244652,
367
+ "eval_loss": 0.29676035046577454,
368
+ "eval_runtime": 77.1567,
369
+ "eval_samples_per_second": 49.821,
370
+ "eval_steps_per_second": 0.791,
371
+ "learning_rate": 0.001,
372
+ "step": 4525
373
+ },
374
+ {
375
+ "epoch": 26.0,
376
+ "eval_accuracy": 0.2754942767950052,
377
+ "eval_f1_macro": 0.6045462014226007,
378
+ "eval_f1_micro": 0.8420542140997499,
379
+ "eval_loss": 0.2955995500087738,
380
+ "eval_runtime": 77.6213,
381
+ "eval_samples_per_second": 49.522,
382
+ "eval_steps_per_second": 0.786,
383
+ "learning_rate": 0.001,
384
+ "step": 4706
385
+ },
386
+ {
387
+ "epoch": 27.0,
388
+ "eval_accuracy": 0.27653485952133194,
389
+ "eval_f1_macro": 0.6115221375683754,
390
+ "eval_f1_micro": 0.8437684356323902,
391
+ "eval_loss": 0.29585039615631104,
392
+ "eval_runtime": 76.6184,
393
+ "eval_samples_per_second": 50.171,
394
+ "eval_steps_per_second": 0.796,
395
+ "learning_rate": 0.001,
396
+ "step": 4887
397
+ },
398
+ {
399
+ "epoch": 27.624309392265193,
400
+ "grad_norm": 0.1697782576084137,
401
+ "learning_rate": 0.001,
402
+ "loss": 0.3126,
403
+ "step": 5000
404
+ },
405
+ {
406
+ "epoch": 28.0,
407
+ "eval_accuracy": 0.26925078043704476,
408
+ "eval_f1_macro": 0.6191186747828321,
409
+ "eval_f1_micro": 0.8446938104986479,
410
+ "eval_loss": 0.295540988445282,
411
+ "eval_runtime": 76.9629,
412
+ "eval_samples_per_second": 49.946,
413
+ "eval_steps_per_second": 0.793,
414
+ "learning_rate": 0.001,
415
+ "step": 5068
416
+ },
417
+ {
418
+ "epoch": 29.0,
419
+ "eval_accuracy": 0.2663891779396462,
420
+ "eval_f1_macro": 0.6215750043898619,
421
+ "eval_f1_micro": 0.8437664387164651,
422
+ "eval_loss": 0.3010655343532562,
423
+ "eval_runtime": 77.7511,
424
+ "eval_samples_per_second": 49.44,
425
+ "eval_steps_per_second": 0.785,
426
+ "learning_rate": 0.001,
427
+ "step": 5249
428
+ },
429
+ {
430
+ "epoch": 30.0,
431
+ "eval_accuracy": 0.2809573361082206,
432
+ "eval_f1_macro": 0.6025311078598518,
433
+ "eval_f1_micro": 0.8437435686355217,
434
+ "eval_loss": 0.29214760661125183,
435
+ "eval_runtime": 76.285,
436
+ "eval_samples_per_second": 50.39,
437
+ "eval_steps_per_second": 0.8,
438
+ "learning_rate": 0.0001,
439
+ "step": 5430
440
+ },
441
+ {
442
+ "epoch": 30.386740331491712,
443
+ "grad_norm": 0.15570667386054993,
444
+ "learning_rate": 0.0001,
445
+ "loss": 0.3093,
446
+ "step": 5500
447
+ },
448
+ {
449
+ "epoch": 31.0,
450
+ "eval_accuracy": 0.28121748178980227,
451
+ "eval_f1_macro": 0.6071651131848005,
452
+ "eval_f1_micro": 0.8439103638567266,
453
+ "eval_loss": 0.29040178656578064,
454
+ "eval_runtime": 80.1919,
455
+ "eval_samples_per_second": 47.935,
456
+ "eval_steps_per_second": 0.761,
457
+ "learning_rate": 0.0001,
458
+ "step": 5611
459
+ },
460
+ {
461
+ "epoch": 32.0,
462
+ "eval_accuracy": 0.2809573361082206,
463
+ "eval_f1_macro": 0.6111569473926136,
464
+ "eval_f1_micro": 0.8437194965322373,
465
+ "eval_loss": 0.29034462571144104,
466
+ "eval_runtime": 76.5731,
467
+ "eval_samples_per_second": 50.2,
468
+ "eval_steps_per_second": 0.797,
469
+ "learning_rate": 0.0001,
470
+ "step": 5792
471
+ },
472
+ {
473
+ "epoch": 33.0,
474
+ "eval_accuracy": 0.28537981269510926,
475
+ "eval_f1_macro": 0.6202495870793918,
476
+ "eval_f1_micro": 0.8461617038663874,
477
+ "eval_loss": 0.2888760268688202,
478
+ "eval_runtime": 78.8612,
479
+ "eval_samples_per_second": 48.744,
480
+ "eval_steps_per_second": 0.774,
481
+ "learning_rate": 0.0001,
482
+ "step": 5973
483
+ },
484
+ {
485
+ "epoch": 33.149171270718234,
486
+ "grad_norm": 0.16711881756782532,
487
+ "learning_rate": 0.0001,
488
+ "loss": 0.3049,
489
+ "step": 6000
490
+ },
491
+ {
492
+ "epoch": 34.0,
493
+ "eval_accuracy": 0.2861602497398543,
494
+ "eval_f1_macro": 0.6150504150317478,
495
+ "eval_f1_micro": 0.8446023671361742,
496
+ "eval_loss": 0.28964364528656006,
497
+ "eval_runtime": 78.2568,
498
+ "eval_samples_per_second": 49.12,
499
+ "eval_steps_per_second": 0.779,
500
+ "learning_rate": 0.0001,
501
+ "step": 6154
502
+ },
503
+ {
504
+ "epoch": 35.0,
505
+ "eval_accuracy": 0.2866805411030177,
506
+ "eval_f1_macro": 0.611180048847438,
507
+ "eval_f1_micro": 0.8449244728566273,
508
+ "eval_loss": 0.28874215483665466,
509
+ "eval_runtime": 78.411,
510
+ "eval_samples_per_second": 49.024,
511
+ "eval_steps_per_second": 0.778,
512
+ "learning_rate": 0.0001,
513
+ "step": 6335
514
+ },
515
+ {
516
+ "epoch": 35.91160220994475,
517
+ "grad_norm": 0.1489323228597641,
518
+ "learning_rate": 0.0001,
519
+ "loss": 0.3012,
520
+ "step": 6500
521
+ },
522
+ {
523
+ "epoch": 36.0,
524
+ "eval_accuracy": 0.28355879292403746,
525
+ "eval_f1_macro": 0.6119874534823754,
526
+ "eval_f1_micro": 0.8447173058645225,
527
+ "eval_loss": 0.2888963222503662,
528
+ "eval_runtime": 80.9886,
529
+ "eval_samples_per_second": 47.463,
530
+ "eval_steps_per_second": 0.753,
531
+ "learning_rate": 0.0001,
532
+ "step": 6516
533
+ },
534
+ {
535
+ "epoch": 37.0,
536
+ "eval_accuracy": 0.2866805411030177,
537
+ "eval_f1_macro": 0.6255767175486281,
538
+ "eval_f1_micro": 0.8475834540970686,
539
+ "eval_loss": 0.288282573223114,
540
+ "eval_runtime": 78.5643,
541
+ "eval_samples_per_second": 48.928,
542
+ "eval_steps_per_second": 0.776,
543
+ "learning_rate": 0.0001,
544
+ "step": 6697
545
+ },
546
+ {
547
+ "epoch": 38.0,
548
+ "eval_accuracy": 0.28251821019771073,
549
+ "eval_f1_macro": 0.6057239934398935,
550
+ "eval_f1_micro": 0.8452536426724028,
551
+ "eval_loss": 0.29050976037979126,
552
+ "eval_runtime": 79.802,
553
+ "eval_samples_per_second": 48.169,
554
+ "eval_steps_per_second": 0.764,
555
+ "learning_rate": 0.0001,
556
+ "step": 6878
557
+ },
558
+ {
559
+ "epoch": 38.67403314917127,
560
+ "grad_norm": 0.1844823658466339,
561
+ "learning_rate": 0.0001,
562
+ "loss": 0.299,
563
+ "step": 7000
564
+ },
565
+ {
566
+ "epoch": 39.0,
567
+ "eval_accuracy": 0.28537981269510926,
568
+ "eval_f1_macro": 0.625366961909805,
569
+ "eval_f1_micro": 0.8470600182796791,
570
+ "eval_loss": 0.28778275847435,
571
+ "eval_runtime": 78.7507,
572
+ "eval_samples_per_second": 48.812,
573
+ "eval_steps_per_second": 0.775,
574
+ "learning_rate": 0.0001,
575
+ "step": 7059
576
+ },
577
+ {
578
+ "epoch": 40.0,
579
+ "eval_accuracy": 0.2809573361082206,
580
+ "eval_f1_macro": 0.622337777946806,
581
+ "eval_f1_micro": 0.8468000302716884,
582
+ "eval_loss": 0.2885717749595642,
583
+ "eval_runtime": 79.0959,
584
+ "eval_samples_per_second": 48.599,
585
+ "eval_steps_per_second": 0.771,
586
+ "learning_rate": 0.0001,
587
+ "step": 7240
588
+ },
589
+ {
590
+ "epoch": 41.0,
591
+ "eval_accuracy": 0.2843392299687825,
592
+ "eval_f1_macro": 0.6260539681026288,
593
+ "eval_f1_micro": 0.847323400258903,
594
+ "eval_loss": 0.28773826360702515,
595
+ "eval_runtime": 81.7413,
596
+ "eval_samples_per_second": 47.026,
597
+ "eval_steps_per_second": 0.746,
598
+ "learning_rate": 0.0001,
599
+ "step": 7421
600
+ },
601
+ {
602
+ "epoch": 41.43646408839779,
603
+ "grad_norm": 0.16540081799030304,
604
+ "learning_rate": 0.0001,
605
+ "loss": 0.2989,
606
+ "step": 7500
607
+ },
608
+ {
609
+ "epoch": 42.0,
610
+ "eval_accuracy": 0.28563995837669093,
611
+ "eval_f1_macro": 0.6199392946357273,
612
+ "eval_f1_micro": 0.8476613005450627,
613
+ "eval_loss": 0.28776827454566956,
614
+ "eval_runtime": 78.9339,
615
+ "eval_samples_per_second": 48.699,
616
+ "eval_steps_per_second": 0.773,
617
+ "learning_rate": 0.0001,
618
+ "step": 7602
619
+ },
620
+ {
621
+ "epoch": 43.0,
622
+ "eval_accuracy": 0.28303850156087407,
623
+ "eval_f1_macro": 0.6287571427217789,
624
+ "eval_f1_micro": 0.8479237095716232,
625
+ "eval_loss": 0.28717148303985596,
626
+ "eval_runtime": 77.7099,
627
+ "eval_samples_per_second": 49.466,
628
+ "eval_steps_per_second": 0.785,
629
+ "learning_rate": 0.0001,
630
+ "step": 7783
631
+ },
632
+ {
633
+ "epoch": 44.0,
634
+ "eval_accuracy": 0.28407908428720086,
635
+ "eval_f1_macro": 0.6189979239207937,
636
+ "eval_f1_micro": 0.8463665693654939,
637
+ "eval_loss": 0.28678667545318604,
638
+ "eval_runtime": 78.2343,
639
+ "eval_samples_per_second": 49.134,
640
+ "eval_steps_per_second": 0.78,
641
+ "learning_rate": 0.0001,
642
+ "step": 7964
643
+ },
644
+ {
645
+ "epoch": 44.19889502762431,
646
+ "grad_norm": 0.17522749304771423,
647
+ "learning_rate": 0.0001,
648
+ "loss": 0.2983,
649
+ "step": 8000
650
+ },
651
+ {
652
+ "epoch": 45.0,
653
+ "eval_accuracy": 0.28381893860561913,
654
+ "eval_f1_macro": 0.6235508782461164,
655
+ "eval_f1_micro": 0.8462928555066304,
656
+ "eval_loss": 0.28698909282684326,
657
+ "eval_runtime": 78.0653,
658
+ "eval_samples_per_second": 49.241,
659
+ "eval_steps_per_second": 0.781,
660
+ "learning_rate": 0.0001,
661
+ "step": 8145
662
+ },
663
+ {
664
+ "epoch": 46.0,
665
+ "eval_accuracy": 0.28251821019771073,
666
+ "eval_f1_macro": 0.6151318511304835,
667
+ "eval_f1_micro": 0.8459846547314578,
668
+ "eval_loss": 0.2868472635746002,
669
+ "eval_runtime": 77.6178,
670
+ "eval_samples_per_second": 49.525,
671
+ "eval_steps_per_second": 0.786,
672
+ "learning_rate": 0.0001,
673
+ "step": 8326
674
+ },
675
+ {
676
+ "epoch": 46.96132596685083,
677
+ "grad_norm": 0.20419611036777496,
678
+ "learning_rate": 0.0001,
679
+ "loss": 0.298,
680
+ "step": 8500
681
+ },
682
+ {
683
+ "epoch": 47.0,
684
+ "eval_accuracy": 0.2845993756503642,
685
+ "eval_f1_macro": 0.6211457155619424,
686
+ "eval_f1_micro": 0.8462129359348595,
687
+ "eval_loss": 0.28715068101882935,
688
+ "eval_runtime": 77.3289,
689
+ "eval_samples_per_second": 49.71,
690
+ "eval_steps_per_second": 0.789,
691
+ "learning_rate": 0.0001,
692
+ "step": 8507
693
+ },
694
+ {
695
+ "epoch": 48.0,
696
+ "eval_accuracy": 0.28355879292403746,
697
+ "eval_f1_macro": 0.6231150403485404,
698
+ "eval_f1_micro": 0.8466852933705867,
699
+ "eval_loss": 0.28661593794822693,
700
+ "eval_runtime": 76.7641,
701
+ "eval_samples_per_second": 50.075,
702
+ "eval_steps_per_second": 0.795,
703
+ "learning_rate": 0.0001,
704
+ "step": 8688
705
+ },
706
+ {
707
+ "epoch": 49.0,
708
+ "eval_accuracy": 0.28590010405827265,
709
+ "eval_f1_macro": 0.616055362439494,
710
+ "eval_f1_micro": 0.8460415439387342,
711
+ "eval_loss": 0.28633347153663635,
712
+ "eval_runtime": 76.5745,
713
+ "eval_samples_per_second": 50.199,
714
+ "eval_steps_per_second": 0.797,
715
+ "learning_rate": 0.0001,
716
+ "step": 8869
717
+ },
718
+ {
719
+ "epoch": 49.72375690607735,
720
+ "grad_norm": 0.2103131115436554,
721
+ "learning_rate": 0.0001,
722
+ "loss": 0.2965,
723
+ "step": 9000
724
+ },
725
+ {
726
+ "epoch": 50.0,
727
+ "eval_accuracy": 0.2845993756503642,
728
+ "eval_f1_macro": 0.625458075101288,
729
+ "eval_f1_micro": 0.8482882700250868,
730
+ "eval_loss": 0.28642749786376953,
731
+ "eval_runtime": 76.3371,
732
+ "eval_samples_per_second": 50.356,
733
+ "eval_steps_per_second": 0.799,
734
+ "learning_rate": 0.0001,
735
+ "step": 9050
736
+ },
737
+ {
738
+ "epoch": 51.0,
739
+ "eval_accuracy": 0.28485952133194586,
740
+ "eval_f1_macro": 0.6278100779578839,
741
+ "eval_f1_micro": 0.848592785832539,
742
+ "eval_loss": 0.2890762686729431,
743
+ "eval_runtime": 77.0258,
744
+ "eval_samples_per_second": 49.905,
745
+ "eval_steps_per_second": 0.792,
746
+ "learning_rate": 0.0001,
747
+ "step": 9231
748
+ },
749
+ {
750
+ "epoch": 52.0,
751
+ "eval_accuracy": 0.2851196670135276,
752
+ "eval_f1_macro": 0.6255462096645672,
753
+ "eval_f1_micro": 0.8464228285561143,
754
+ "eval_loss": 0.2855978012084961,
755
+ "eval_runtime": 76.6781,
756
+ "eval_samples_per_second": 50.132,
757
+ "eval_steps_per_second": 0.796,
758
+ "learning_rate": 0.0001,
759
+ "step": 9412
760
+ },
761
+ {
762
+ "epoch": 52.48618784530387,
763
+ "grad_norm": 0.24192312359809875,
764
+ "learning_rate": 0.0001,
765
+ "loss": 0.2956,
766
+ "step": 9500
767
+ },
768
+ {
769
+ "epoch": 53.0,
770
+ "eval_accuracy": 0.27887617065556713,
771
+ "eval_f1_macro": 0.6457587856102145,
772
+ "eval_f1_micro": 0.8489991514001897,
773
+ "eval_loss": 0.2872205674648285,
774
+ "eval_runtime": 76.6479,
775
+ "eval_samples_per_second": 50.151,
776
+ "eval_steps_per_second": 0.796,
777
+ "learning_rate": 0.0001,
778
+ "step": 9593
779
+ },
780
+ {
781
+ "epoch": 54.0,
782
+ "eval_accuracy": 0.2903225806451613,
783
+ "eval_f1_macro": 0.6243869856844756,
784
+ "eval_f1_micro": 0.8476844874709444,
785
+ "eval_loss": 0.2855803072452545,
786
+ "eval_runtime": 77.582,
787
+ "eval_samples_per_second": 49.548,
788
+ "eval_steps_per_second": 0.786,
789
+ "learning_rate": 0.0001,
790
+ "step": 9774
791
+ },
792
+ {
793
+ "epoch": 55.0,
794
+ "eval_accuracy": 0.2845993756503642,
795
+ "eval_f1_macro": 0.6339630509281279,
796
+ "eval_f1_micro": 0.8475136716266056,
797
+ "eval_loss": 0.28568968176841736,
798
+ "eval_runtime": 77.102,
799
+ "eval_samples_per_second": 49.856,
800
+ "eval_steps_per_second": 0.791,
801
+ "learning_rate": 0.0001,
802
+ "step": 9955
803
+ },
804
+ {
805
+ "epoch": 55.248618784530386,
806
+ "grad_norm": 0.21083500981330872,
807
+ "learning_rate": 0.0001,
808
+ "loss": 0.2958,
809
+ "step": 10000
810
+ },
811
+ {
812
+ "epoch": 56.0,
813
+ "eval_accuracy": 0.2866805411030177,
814
+ "eval_f1_macro": 0.6241465491773776,
815
+ "eval_f1_micro": 0.8465597622829039,
816
+ "eval_loss": 0.28617897629737854,
817
+ "eval_runtime": 76.1445,
818
+ "eval_samples_per_second": 50.483,
819
+ "eval_steps_per_second": 0.801,
820
+ "learning_rate": 0.0001,
821
+ "step": 10136
822
+ },
823
+ {
824
+ "epoch": 57.0,
825
+ "eval_accuracy": 0.2861602497398543,
826
+ "eval_f1_macro": 0.6249269702519318,
827
+ "eval_f1_micro": 0.845436853426201,
828
+ "eval_loss": 0.2870914936065674,
829
+ "eval_runtime": 77.4556,
830
+ "eval_samples_per_second": 49.628,
831
+ "eval_steps_per_second": 0.788,
832
+ "learning_rate": 0.0001,
833
+ "step": 10317
834
+ },
835
+ {
836
+ "epoch": 58.0,
837
+ "eval_accuracy": 0.28121748178980227,
838
+ "eval_f1_macro": 0.6333866717026029,
839
+ "eval_f1_micro": 0.8491941382702348,
840
+ "eval_loss": 0.2857914865016937,
841
+ "eval_runtime": 77.2551,
842
+ "eval_samples_per_second": 49.757,
843
+ "eval_steps_per_second": 0.79,
844
+ "learning_rate": 0.0001,
845
+ "step": 10498
846
+ },
847
+ {
848
+ "epoch": 58.011049723756905,
849
+ "grad_norm": 0.22250542044639587,
850
+ "learning_rate": 1e-05,
851
+ "loss": 0.2954,
852
+ "step": 10500
853
+ },
854
+ {
855
+ "epoch": 59.0,
856
+ "eval_accuracy": 0.2887617065556712,
857
+ "eval_f1_macro": 0.6178461796051926,
858
+ "eval_f1_micro": 0.8468232576049287,
859
+ "eval_loss": 0.28617140650749207,
860
+ "eval_runtime": 76.6548,
861
+ "eval_samples_per_second": 50.147,
862
+ "eval_steps_per_second": 0.796,
863
+ "learning_rate": 1e-05,
864
+ "step": 10679
865
+ },
866
+ {
867
+ "epoch": 60.0,
868
+ "eval_accuracy": 0.28537981269510926,
869
+ "eval_f1_macro": 0.6275748058546806,
870
+ "eval_f1_micro": 0.8485033598045205,
871
+ "eval_loss": 0.2846605181694031,
872
+ "eval_runtime": 76.2984,
873
+ "eval_samples_per_second": 50.381,
874
+ "eval_steps_per_second": 0.799,
875
+ "learning_rate": 1e-05,
876
+ "step": 10860
877
+ },
878
+ {
879
+ "epoch": 60.773480662983424,
880
+ "grad_norm": 0.25101110339164734,
881
+ "learning_rate": 1e-05,
882
+ "loss": 0.2923,
883
+ "step": 11000
884
+ },
885
+ {
886
+ "epoch": 61.0,
887
+ "eval_accuracy": 0.28303850156087407,
888
+ "eval_f1_macro": 0.6223888517425455,
889
+ "eval_f1_micro": 0.8479865171982329,
890
+ "eval_loss": 0.2848633825778961,
891
+ "eval_runtime": 76.6322,
892
+ "eval_samples_per_second": 50.162,
893
+ "eval_steps_per_second": 0.796,
894
+ "learning_rate": 1e-05,
895
+ "step": 11041
896
+ },
897
+ {
898
+ "epoch": 62.0,
899
+ "eval_accuracy": 0.2843392299687825,
900
+ "eval_f1_macro": 0.6247632003821695,
901
+ "eval_f1_micro": 0.8469200122586577,
902
+ "eval_loss": 0.28548601269721985,
903
+ "eval_runtime": 77.5636,
904
+ "eval_samples_per_second": 49.559,
905
+ "eval_steps_per_second": 0.786,
906
+ "learning_rate": 1e-05,
907
+ "step": 11222
908
+ },
909
+ {
910
+ "epoch": 63.0,
911
+ "eval_accuracy": 0.2827783558792924,
912
+ "eval_f1_macro": 0.6274806463168713,
913
+ "eval_f1_micro": 0.8488979777323336,
914
+ "eval_loss": 0.28493326902389526,
915
+ "eval_runtime": 77.2321,
916
+ "eval_samples_per_second": 49.772,
917
+ "eval_steps_per_second": 0.79,
918
+ "learning_rate": 1e-05,
919
+ "step": 11403
920
+ },
921
+ {
922
+ "epoch": 63.53591160220994,
923
+ "grad_norm": 0.23796355724334717,
924
+ "learning_rate": 1e-05,
925
+ "loss": 0.2918,
926
+ "step": 11500
927
+ },
928
+ {
929
+ "epoch": 64.0,
930
+ "eval_accuracy": 0.28225806451612906,
931
+ "eval_f1_macro": 0.6370787064578803,
932
+ "eval_f1_micro": 0.8475187206498287,
933
+ "eval_loss": 0.28459736704826355,
934
+ "eval_runtime": 77.0797,
935
+ "eval_samples_per_second": 49.87,
936
+ "eval_steps_per_second": 0.791,
937
+ "learning_rate": 1e-05,
938
+ "step": 11584
939
+ },
940
+ {
941
+ "epoch": 65.0,
942
+ "eval_accuracy": 0.2869406867845994,
943
+ "eval_f1_macro": 0.6240984315849201,
944
+ "eval_f1_micro": 0.8467700785794469,
945
+ "eval_loss": 0.2860054671764374,
946
+ "eval_runtime": 76.4904,
947
+ "eval_samples_per_second": 50.255,
948
+ "eval_steps_per_second": 0.797,
949
+ "learning_rate": 1e-05,
950
+ "step": 11765
951
+ },
952
+ {
953
+ "epoch": 66.0,
954
+ "eval_accuracy": 0.28407908428720086,
955
+ "eval_f1_macro": 0.6346693986906206,
956
+ "eval_f1_micro": 0.8481340441736481,
957
+ "eval_loss": 0.2847185730934143,
958
+ "eval_runtime": 77.2653,
959
+ "eval_samples_per_second": 49.751,
960
+ "eval_steps_per_second": 0.789,
961
+ "learning_rate": 1e-05,
962
+ "step": 11946
963
+ },
964
+ {
965
+ "epoch": 66.29834254143647,
966
+ "grad_norm": 0.25470152497291565,
967
+ "learning_rate": 1e-05,
968
+ "loss": 0.2906,
969
+ "step": 12000
970
+ },
971
+ {
972
+ "epoch": 67.0,
973
+ "eval_accuracy": 0.28537981269510926,
974
+ "eval_f1_macro": 0.6287121285420982,
975
+ "eval_f1_micro": 0.8487528745798691,
976
+ "eval_loss": 0.28529325127601624,
977
+ "eval_runtime": 79.9065,
978
+ "eval_samples_per_second": 48.106,
979
+ "eval_steps_per_second": 0.763,
980
+ "learning_rate": 1e-05,
981
+ "step": 12127
982
+ },
983
+ {
984
+ "epoch": 68.0,
985
+ "eval_accuracy": 0.2866805411030177,
986
+ "eval_f1_macro": 0.6321379394582358,
987
+ "eval_f1_micro": 0.8480251642525557,
988
+ "eval_loss": 0.2852926254272461,
989
+ "eval_runtime": 78.4728,
990
+ "eval_samples_per_second": 48.985,
991
+ "eval_steps_per_second": 0.777,
992
+ "learning_rate": 1e-05,
993
+ "step": 12308
994
+ },
995
+ {
996
+ "epoch": 69.0,
997
+ "eval_accuracy": 0.28355879292403746,
998
+ "eval_f1_macro": 0.6397237492354447,
999
+ "eval_f1_micro": 0.847692190707931,
1000
+ "eval_loss": 0.284834623336792,
1001
+ "eval_runtime": 77.7721,
1002
+ "eval_samples_per_second": 49.426,
1003
+ "eval_steps_per_second": 0.784,
1004
+ "learning_rate": 1e-05,
1005
+ "step": 12489
1006
+ },
1007
+ {
1008
+ "epoch": 69.06077348066299,
1009
+ "grad_norm": 0.19653503596782684,
1010
+ "learning_rate": 1e-05,
1011
+ "loss": 0.2918,
1012
+ "step": 12500
1013
+ },
1014
+ {
1015
+ "epoch": 70.0,
1016
+ "eval_accuracy": 0.28225806451612906,
1017
+ "eval_f1_macro": 0.6381143671040704,
1018
+ "eval_f1_micro": 0.8492167101827677,
1019
+ "eval_loss": 0.28527727723121643,
1020
+ "eval_runtime": 76.6607,
1021
+ "eval_samples_per_second": 50.143,
1022
+ "eval_steps_per_second": 0.796,
1023
+ "learning_rate": 1e-05,
1024
+ "step": 12670
1025
+ },
1026
+ {
1027
+ "epoch": 71.0,
1028
+ "eval_accuracy": 0.2882414151925078,
1029
+ "eval_f1_macro": 0.6325489300082728,
1030
+ "eval_f1_micro": 0.8475971370143149,
1031
+ "eval_loss": 0.28507113456726074,
1032
+ "eval_runtime": 76.9731,
1033
+ "eval_samples_per_second": 49.94,
1034
+ "eval_steps_per_second": 0.792,
1035
+ "learning_rate": 1.0000000000000002e-06,
1036
+ "step": 12851
1037
+ },
1038
+ {
1039
+ "epoch": 71.8232044198895,
1040
+ "grad_norm": 0.19946995377540588,
1041
+ "learning_rate": 1.0000000000000002e-06,
1042
+ "loss": 0.2918,
1043
+ "step": 13000
1044
+ },
1045
+ {
1046
+ "epoch": 72.0,
1047
+ "eval_accuracy": 0.28485952133194586,
1048
+ "eval_f1_macro": 0.6236352127811986,
1049
+ "eval_f1_micro": 0.8474255781269963,
1050
+ "eval_loss": 0.28452861309051514,
1051
+ "eval_runtime": 79.7463,
1052
+ "eval_samples_per_second": 48.203,
1053
+ "eval_steps_per_second": 0.765,
1054
+ "learning_rate": 1.0000000000000002e-06,
1055
+ "step": 13032
1056
+ },
1057
+ {
1058
+ "epoch": 73.0,
1059
+ "eval_accuracy": 0.28121748178980227,
1060
+ "eval_f1_macro": 0.6333277250193455,
1061
+ "eval_f1_micro": 0.847641772858811,
1062
+ "eval_loss": 0.28448227047920227,
1063
+ "eval_runtime": 76.9059,
1064
+ "eval_samples_per_second": 49.983,
1065
+ "eval_steps_per_second": 0.793,
1066
+ "learning_rate": 1.0000000000000002e-06,
1067
+ "step": 13213
1068
+ },
1069
+ {
1070
+ "epoch": 74.0,
1071
+ "eval_accuracy": 0.2827783558792924,
1072
+ "eval_f1_macro": 0.6300187593616763,
1073
+ "eval_f1_micro": 0.8465770953294945,
1074
+ "eval_loss": 0.28447526693344116,
1075
+ "eval_runtime": 77.6657,
1076
+ "eval_samples_per_second": 49.494,
1077
+ "eval_steps_per_second": 0.785,
1078
+ "learning_rate": 1.0000000000000002e-06,
1079
+ "step": 13394
1080
+ },
1081
+ {
1082
+ "epoch": 74.58563535911603,
1083
+ "grad_norm": 0.251558780670166,
1084
+ "learning_rate": 1.0000000000000002e-06,
1085
+ "loss": 0.2913,
1086
+ "step": 13500
1087
+ },
1088
+ {
1089
+ "epoch": 75.0,
1090
+ "eval_accuracy": 0.28199791883454733,
1091
+ "eval_f1_macro": 0.6235297745568456,
1092
+ "eval_f1_micro": 0.8473772748126625,
1093
+ "eval_loss": 0.2851284146308899,
1094
+ "eval_runtime": 77.914,
1095
+ "eval_samples_per_second": 49.336,
1096
+ "eval_steps_per_second": 0.783,
1097
+ "learning_rate": 1.0000000000000002e-06,
1098
+ "step": 13575
1099
+ },
1100
+ {
1101
+ "epoch": 76.0,
1102
+ "eval_accuracy": 0.2879812695109261,
1103
+ "eval_f1_macro": 0.6186062513830065,
1104
+ "eval_f1_micro": 0.847320835674516,
1105
+ "eval_loss": 0.2859683036804199,
1106
+ "eval_runtime": 77.7414,
1107
+ "eval_samples_per_second": 49.446,
1108
+ "eval_steps_per_second": 0.785,
1109
+ "learning_rate": 1.0000000000000002e-06,
1110
+ "step": 13756
1111
+ },
1112
+ {
1113
+ "epoch": 77.0,
1114
+ "eval_accuracy": 0.28563995837669093,
1115
+ "eval_f1_macro": 0.6172786558676017,
1116
+ "eval_f1_micro": 0.8459046737621472,
1117
+ "eval_loss": 0.2858298718929291,
1118
+ "eval_runtime": 79.1015,
1119
+ "eval_samples_per_second": 48.596,
1120
+ "eval_steps_per_second": 0.771,
1121
+ "learning_rate": 1.0000000000000002e-06,
1122
+ "step": 13937
1123
+ },
1124
+ {
1125
+ "epoch": 77.34806629834254,
1126
+ "grad_norm": 0.22088366746902466,
1127
+ "learning_rate": 1.0000000000000002e-06,
1128
+ "loss": 0.2913,
1129
+ "step": 14000
1130
+ },
1131
+ {
1132
+ "epoch": 78.0,
1133
+ "eval_accuracy": 0.2843392299687825,
1134
+ "eval_f1_macro": 0.6325947858436887,
1135
+ "eval_f1_micro": 0.8480547459130655,
1136
+ "eval_loss": 0.28438833355903625,
1137
+ "eval_runtime": 77.2562,
1138
+ "eval_samples_per_second": 49.757,
1139
+ "eval_steps_per_second": 0.79,
1140
+ "learning_rate": 1.0000000000000002e-06,
1141
+ "step": 14118
1142
+ },
1143
+ {
1144
+ "epoch": 79.0,
1145
+ "eval_accuracy": 0.2874609781477627,
1146
+ "eval_f1_macro": 0.617917490234713,
1147
+ "eval_f1_micro": 0.8472353346431579,
1148
+ "eval_loss": 0.2870919704437256,
1149
+ "eval_runtime": 76.5647,
1150
+ "eval_samples_per_second": 50.206,
1151
+ "eval_steps_per_second": 0.797,
1152
+ "learning_rate": 1.0000000000000002e-06,
1153
+ "step": 14299
1154
+ },
1155
+ {
1156
+ "epoch": 80.0,
1157
+ "eval_accuracy": 0.28381893860561913,
1158
+ "eval_f1_macro": 0.6286567457369128,
1159
+ "eval_f1_micro": 0.8477330616403465,
1160
+ "eval_loss": 0.28482332825660706,
1161
+ "eval_runtime": 76.83,
1162
+ "eval_samples_per_second": 50.033,
1163
+ "eval_steps_per_second": 0.794,
1164
+ "learning_rate": 1.0000000000000002e-06,
1165
+ "step": 14480
1166
+ },
1167
+ {
1168
+ "epoch": 80.11049723756906,
1169
+ "grad_norm": 0.21530944108963013,
1170
+ "learning_rate": 1.0000000000000002e-06,
1171
+ "loss": 0.2915,
1172
+ "step": 14500
1173
+ },
1174
+ {
1175
+ "epoch": 81.0,
1176
+ "eval_accuracy": 0.28537981269510926,
1177
+ "eval_f1_macro": 0.6304525529970205,
1178
+ "eval_f1_micro": 0.8489678202792957,
1179
+ "eval_loss": 0.2847617268562317,
1180
+ "eval_runtime": 77.26,
1181
+ "eval_samples_per_second": 49.754,
1182
+ "eval_steps_per_second": 0.79,
1183
+ "learning_rate": 1.0000000000000002e-06,
1184
+ "step": 14661
1185
+ },
1186
+ {
1187
+ "epoch": 82.0,
1188
+ "eval_accuracy": 0.28590010405827265,
1189
+ "eval_f1_macro": 0.6394217270135759,
1190
+ "eval_f1_micro": 0.8480416961845967,
1191
+ "eval_loss": 0.28511229157447815,
1192
+ "eval_runtime": 78.9926,
1193
+ "eval_samples_per_second": 48.663,
1194
+ "eval_steps_per_second": 0.772,
1195
+ "learning_rate": 1.0000000000000002e-06,
1196
+ "step": 14842
1197
+ },
1198
+ {
1199
+ "epoch": 82.87292817679558,
1200
+ "grad_norm": 0.2371624857187271,
1201
+ "learning_rate": 1.0000000000000002e-06,
1202
+ "loss": 0.2913,
1203
+ "step": 15000
1204
+ },
1205
+ {
1206
+ "epoch": 83.0,
1207
+ "eval_accuracy": 0.28563995837669093,
1208
+ "eval_f1_macro": 0.6255055774993536,
1209
+ "eval_f1_micro": 0.8488055562622434,
1210
+ "eval_loss": 0.284644216299057,
1211
+ "eval_runtime": 76.4754,
1212
+ "eval_samples_per_second": 50.265,
1213
+ "eval_steps_per_second": 0.798,
1214
+ "learning_rate": 1.0000000000000002e-06,
1215
+ "step": 15023
1216
+ },
1217
+ {
1218
+ "epoch": 84.0,
1219
+ "eval_accuracy": 0.2832986472424558,
1220
+ "eval_f1_macro": 0.6457553263622914,
1221
+ "eval_f1_micro": 0.848188643119867,
1222
+ "eval_loss": 0.2857225835323334,
1223
+ "eval_runtime": 77.2675,
1224
+ "eval_samples_per_second": 49.749,
1225
+ "eval_steps_per_second": 0.789,
1226
+ "learning_rate": 1.0000000000000002e-06,
1227
+ "step": 15204
1228
+ },
1229
+ {
1230
+ "epoch": 85.0,
1231
+ "eval_accuracy": 0.28121748178980227,
1232
+ "eval_f1_macro": 0.6339586571635658,
1233
+ "eval_f1_micro": 0.848818698673405,
1234
+ "eval_loss": 0.28550758957862854,
1235
+ "eval_runtime": 77.3218,
1236
+ "eval_samples_per_second": 49.714,
1237
+ "eval_steps_per_second": 0.789,
1238
+ "learning_rate": 1.0000000000000002e-07,
1239
+ "step": 15385
1240
+ },
1241
+ {
1242
+ "epoch": 85.6353591160221,
1243
+ "grad_norm": 0.22222235798835754,
1244
+ "learning_rate": 1.0000000000000002e-07,
1245
+ "loss": 0.2922,
1246
+ "step": 15500
1247
+ },
1248
+ {
1249
+ "epoch": 86.0,
1250
+ "eval_accuracy": 0.28590010405827265,
1251
+ "eval_f1_macro": 0.6362631688004041,
1252
+ "eval_f1_micro": 0.8479890588592848,
1253
+ "eval_loss": 0.284895658493042,
1254
+ "eval_runtime": 76.6317,
1255
+ "eval_samples_per_second": 50.162,
1256
+ "eval_steps_per_second": 0.796,
1257
+ "learning_rate": 1.0000000000000002e-07,
1258
+ "step": 15566
1259
+ },
1260
+ {
1261
+ "epoch": 87.0,
1262
+ "eval_accuracy": 0.2851196670135276,
1263
+ "eval_f1_macro": 0.6327749126527296,
1264
+ "eval_f1_micro": 0.8473590201582036,
1265
+ "eval_loss": 0.2845035493373871,
1266
+ "eval_runtime": 77.1171,
1267
+ "eval_samples_per_second": 49.846,
1268
+ "eval_steps_per_second": 0.791,
1269
+ "learning_rate": 1.0000000000000002e-07,
1270
+ "step": 15747
1271
+ },
1272
+ {
1273
+ "epoch": 88.0,
1274
+ "eval_accuracy": 0.28121748178980227,
1275
+ "eval_f1_macro": 0.6370893160624239,
1276
+ "eval_f1_micro": 0.8477551536613127,
1277
+ "eval_loss": 0.28541097044944763,
1278
+ "eval_runtime": 76.6873,
1279
+ "eval_samples_per_second": 50.126,
1280
+ "eval_steps_per_second": 0.795,
1281
+ "learning_rate": 1.0000000000000002e-07,
1282
+ "step": 15928
1283
+ },
1284
+ {
1285
+ "epoch": 88.0,
1286
+ "learning_rate": 1.0000000000000002e-07,
1287
+ "step": 15928,
1288
+ "total_flos": 1.500719176717825e+20,
1289
+ "train_loss": 0.3047179739897961,
1290
+ "train_runtime": 30841.8451,
1291
+ "train_samples_per_second": 56.038,
1292
+ "train_steps_per_second": 0.88
1293
+ }
1294
+ ],
1295
+ "logging_steps": 500,
1296
+ "max_steps": 27150,
1297
+ "num_input_tokens_seen": 0,
1298
+ "num_train_epochs": 150,
1299
+ "save_steps": 500,
1300
+ "stateful_callbacks": {
1301
+ "EarlyStoppingCallback": {
1302
+ "args": {
1303
+ "early_stopping_patience": 10,
1304
+ "early_stopping_threshold": 0.0
1305
+ },
1306
+ "attributes": {
1307
+ "early_stopping_patience_counter": 0
1308
+ }
1309
+ },
1310
+ "TrainerControl": {
1311
+ "args": {
1312
+ "should_epoch_stop": false,
1313
+ "should_evaluate": false,
1314
+ "should_log": false,
1315
+ "should_save": true,
1316
+ "should_training_stop": true
1317
+ },
1318
+ "attributes": {}
1319
+ }
1320
+ },
1321
+ "total_flos": 1.500719176717825e+20,
1322
+ "train_batch_size": 64,
1323
+ "trial_name": null,
1324
+ "trial_params": null
1325
+ }