Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- cross_cell_type_generization/L1/B cells/fold0/all_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold0/checkpoint-1309/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold0/checkpoint-1309/trainer_state.json +398 -0
- cross_cell_type_generization/L1/B cells/fold0/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold0/eval_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold1/all_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold1/checkpoint-1496/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold1/checkpoint-1496/trainer_state.json +450 -0
- cross_cell_type_generization/L1/B cells/fold1/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold1/eval_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold2/all_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold2/checkpoint-1122/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold2/checkpoint-1122/trainer_state.json +340 -0
- cross_cell_type_generization/L1/B cells/fold2/eval_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold3/all_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold3/checkpoint-1309/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold3/checkpoint-1309/trainer_state.json +398 -0
- cross_cell_type_generization/L1/B cells/fold3/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold3/eval_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold4/all_results.json +8 -0
- cross_cell_type_generization/L1/B cells/fold4/checkpoint-1309/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold4/checkpoint-1309/trainer_state.json +398 -0
- cross_cell_type_generization/L1/B cells/fold4/config.json +25 -0
- cross_cell_type_generization/L1/B cells/fold4/eval_results.json +8 -0
- cross_cell_type_generization/L1/Epitheliums/fold1/all_results.json +8 -0
- cross_cell_type_generization/L1/Epitheliums/fold1/checkpoint-1496/config.json +25 -0
- cross_cell_type_generization/L1/Epitheliums/fold1/config.json +25 -0
- cross_cell_type_generization/L1/Epitheliums/fold1/eval_results.json +8 -0
- cross_cell_type_generization/L6/.DS_Store +0 -0
- cross_cell_type_generization/L6/NKT/.DS_Store +0 -0
- cross_cell_type_generization/L6/NKT/fold0/.DS_Store +0 -0
- cross_cell_type_generization/L6/NKT/fold0/all_results.json +8 -0
- cross_cell_type_generization/L6/NKT/fold0/checkpoint-1683/config.json +25 -0
- cross_cell_type_generization/L6/NKT/fold0/checkpoint-1683/trainer_state.json +508 -0
- cross_cell_type_generization/L6/NKT/fold0/config.json +25 -0
- cross_cell_type_generization/L6/NKT/fold0/eval_results.json +8 -0
- cross_cell_type_generization/L6/NKT/fold1/.DS_Store +0 -0
- cross_cell_type_generization/L6/NKT/fold1/all_results.json +8 -0
- cross_cell_type_generization/L6/NKT/fold1/checkpoint-1870/config.json +25 -0
- cross_cell_type_generization/L6/NKT/fold1/checkpoint-1870/trainer_state.json +560 -0
- cross_cell_type_generization/L6/NKT/fold1/config.json +25 -0
- cross_cell_type_generization/L6/NKT/fold1/eval_results.json +8 -0
- cross_cell_type_generization/L6/NKT/fold2/.DS_Store +0 -0
- cross_cell_type_generization/L6/NKT/fold2/all_results.json +8 -0
- cross_cell_type_generization/L6/NKT/fold2/checkpoint-1496/config.json +25 -0
- cross_cell_type_generization/L6/NKT/fold2/checkpoint-1496/trainer_state.json +450 -0
- cross_cell_type_generization/L6/NKT/fold2/config.json +25 -0
- cross_cell_type_generization/L6/NKT/fold2/eval_results.json +8 -0
- cross_cell_type_generization/L6/NKT/fold3/.DS_Store +0 -0
- cross_cell_type_generization/L6/NKT/fold3/all_results.json +8 -0
cross_cell_type_generization/L1/B cells/fold0/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9,
|
| 3 |
+
"test_loss": 0.25478145480155945,
|
| 4 |
+
"test_macro_f1": 0.8771714868012422,
|
| 5 |
+
"test_runtime": 2.097,
|
| 6 |
+
"test_samples_per_second": 429.179,
|
| 7 |
+
"test_steps_per_second": 4.292
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold0/checkpoint-1309/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold0/checkpoint-1309/trainer_state.json
ADDED
|
@@ -0,0 +1,398 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.4302835166454315,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L1/B cells/fold0/checkpoint-935",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"global_step": 1309,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.5e-06,
|
| 13 |
+
"loss": 0.6964,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6941,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 7.5e-06,
|
| 25 |
+
"loss": 0.6922,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1e-05,
|
| 31 |
+
"loss": 0.6918,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6926,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 1.5e-05,
|
| 43 |
+
"loss": 0.6918,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.94,
|
| 48 |
+
"learning_rate": 1.75e-05,
|
| 49 |
+
"loss": 0.693,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5176397771817619,
|
| 55 |
+
"eval_loss": 0.6918941140174866,
|
| 56 |
+
"eval_macro_f1": 0.5155677016806794,
|
| 57 |
+
"eval_runtime": 11.5871,
|
| 58 |
+
"eval_samples_per_second": 418.309,
|
| 59 |
+
"eval_steps_per_second": 4.056,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6929,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6907,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6889,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6869,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6884,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6855,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6822,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5743759026201775,
|
| 107 |
+
"eval_loss": 0.6831711530685425,
|
| 108 |
+
"eval_macro_f1": 0.5743595969251305,
|
| 109 |
+
"eval_runtime": 12.0925,
|
| 110 |
+
"eval_samples_per_second": 400.825,
|
| 111 |
+
"eval_steps_per_second": 3.887,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6859,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6775,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6739,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6557,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5907,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5263,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.4987,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4886,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7604703940581803,
|
| 165 |
+
"eval_loss": 0.4802941679954529,
|
| 166 |
+
"eval_macro_f1": 0.7602220164383996,
|
| 167 |
+
"eval_runtime": 11.1597,
|
| 168 |
+
"eval_samples_per_second": 434.332,
|
| 169 |
+
"eval_steps_per_second": 4.212,
|
| 170 |
+
"step": 561
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.07,
|
| 174 |
+
"learning_rate": 4.957627118644068e-05,
|
| 175 |
+
"loss": 0.4343,
|
| 176 |
+
"step": 575
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.21,
|
| 180 |
+
"learning_rate": 4.9435028248587575e-05,
|
| 181 |
+
"loss": 0.4086,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.34,
|
| 186 |
+
"learning_rate": 4.929378531073446e-05,
|
| 187 |
+
"loss": 0.402,
|
| 188 |
+
"step": 625
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.48,
|
| 192 |
+
"learning_rate": 4.915254237288136e-05,
|
| 193 |
+
"loss": 0.4011,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.411,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.4226,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3854,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7804827728491851,
|
| 217 |
+
"eval_loss": 0.4387362599372864,
|
| 218 |
+
"eval_macro_f1": 0.7766738608531174,
|
| 219 |
+
"eval_runtime": 11.1917,
|
| 220 |
+
"eval_samples_per_second": 433.089,
|
| 221 |
+
"eval_steps_per_second": 4.2,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3963,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3448,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3083,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3138,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3217,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3232,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3425,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3285,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7881163606354447,
|
| 275 |
+
"eval_loss": 0.4302835166454315,
|
| 276 |
+
"eval_macro_f1": 0.7880586242377647,
|
| 277 |
+
"eval_runtime": 10.8108,
|
| 278 |
+
"eval_samples_per_second": 448.348,
|
| 279 |
+
"eval_steps_per_second": 4.348,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.2966,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.2981,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2619,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2922,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2767,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.2833,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2785,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.7959562616051166,
|
| 327 |
+
"eval_loss": 0.4455493986606598,
|
| 328 |
+
"eval_macro_f1": 0.7944126109828238,
|
| 329 |
+
"eval_runtime": 11.5547,
|
| 330 |
+
"eval_samples_per_second": 419.482,
|
| 331 |
+
"eval_steps_per_second": 4.068,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2769,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2363,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.2484,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.235,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2514,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.255,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2451,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.2316,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.7926552506705179,
|
| 385 |
+
"eval_loss": 0.47305065393447876,
|
| 386 |
+
"eval_macro_f1": 0.792609488371301,
|
| 387 |
+
"eval_runtime": 11.0801,
|
| 388 |
+
"eval_samples_per_second": 437.45,
|
| 389 |
+
"eval_steps_per_second": 4.242,
|
| 390 |
+
"step": 1309
|
| 391 |
+
}
|
| 392 |
+
],
|
| 393 |
+
"max_steps": 9350,
|
| 394 |
+
"num_train_epochs": 50,
|
| 395 |
+
"total_flos": 990266429472768.0,
|
| 396 |
+
"trial_name": null,
|
| 397 |
+
"trial_params": null
|
| 398 |
+
}
|
cross_cell_type_generization/L1/B cells/fold0/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold0/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9,
|
| 3 |
+
"test_loss": 0.25478145480155945,
|
| 4 |
+
"test_macro_f1": 0.8771714868012422,
|
| 5 |
+
"test_runtime": 2.097,
|
| 6 |
+
"test_samples_per_second": 429.179,
|
| 7 |
+
"test_steps_per_second": 4.292
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold1/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9188888888888889,
|
| 3 |
+
"test_loss": 0.23316439986228943,
|
| 4 |
+
"test_macro_f1": 0.8977780336352279,
|
| 5 |
+
"test_runtime": 2.0883,
|
| 6 |
+
"test_samples_per_second": 430.98,
|
| 7 |
+
"test_steps_per_second": 4.31
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold1/checkpoint-1496/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold1/checkpoint-1496/trainer_state.json
ADDED
|
@@ -0,0 +1,450 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.43855828046798706,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L1/B cells/fold1/checkpoint-1122",
|
| 4 |
+
"epoch": 8.0,
|
| 5 |
+
"global_step": 1496,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.5e-06,
|
| 13 |
+
"loss": 0.6952,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6939,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 7.5e-06,
|
| 25 |
+
"loss": 0.6926,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1e-05,
|
| 31 |
+
"loss": 0.6922,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6918,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 1.5e-05,
|
| 43 |
+
"loss": 0.6926,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.94,
|
| 48 |
+
"learning_rate": 1.75e-05,
|
| 49 |
+
"loss": 0.69,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5078415187783739,
|
| 55 |
+
"eval_loss": 0.6918497085571289,
|
| 56 |
+
"eval_macro_f1": 0.38482986882942877,
|
| 57 |
+
"eval_runtime": 10.8702,
|
| 58 |
+
"eval_samples_per_second": 445.807,
|
| 59 |
+
"eval_steps_per_second": 4.324,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6933,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6901,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6874,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6886,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6866,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.687,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6855,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5602558811390838,
|
| 107 |
+
"eval_loss": 0.6841353178024292,
|
| 108 |
+
"eval_macro_f1": 0.5544113010481543,
|
| 109 |
+
"eval_runtime": 11.2058,
|
| 110 |
+
"eval_samples_per_second": 432.456,
|
| 111 |
+
"eval_steps_per_second": 4.194,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6814,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6752,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6728,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6528,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.6025,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5249,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.5112,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4875,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7614527445315724,
|
| 165 |
+
"eval_loss": 0.49462753534317017,
|
| 166 |
+
"eval_macro_f1": 0.7564871967009394,
|
| 167 |
+
"eval_runtime": 11.4201,
|
| 168 |
+
"eval_samples_per_second": 424.338,
|
| 169 |
+
"eval_steps_per_second": 4.116,
|
| 170 |
+
"step": 561
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.07,
|
| 174 |
+
"learning_rate": 4.957627118644068e-05,
|
| 175 |
+
"loss": 0.4376,
|
| 176 |
+
"step": 575
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.21,
|
| 180 |
+
"learning_rate": 4.9435028248587575e-05,
|
| 181 |
+
"loss": 0.4059,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.34,
|
| 186 |
+
"learning_rate": 4.929378531073446e-05,
|
| 187 |
+
"loss": 0.3936,
|
| 188 |
+
"step": 625
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.48,
|
| 192 |
+
"learning_rate": 4.915254237288136e-05,
|
| 193 |
+
"loss": 0.4002,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.4084,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.4079,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3865,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7754849360297152,
|
| 217 |
+
"eval_loss": 0.4549817442893982,
|
| 218 |
+
"eval_macro_f1": 0.7750462498378696,
|
| 219 |
+
"eval_runtime": 11.3477,
|
| 220 |
+
"eval_samples_per_second": 427.046,
|
| 221 |
+
"eval_steps_per_second": 4.142,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3783,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3371,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3454,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3255,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3107,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3326,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3362,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3207,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7903425505571605,
|
| 275 |
+
"eval_loss": 0.4410189390182495,
|
| 276 |
+
"eval_macro_f1": 0.7903282651359017,
|
| 277 |
+
"eval_runtime": 11.6496,
|
| 278 |
+
"eval_samples_per_second": 415.98,
|
| 279 |
+
"eval_steps_per_second": 4.034,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.2995,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.2774,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2809,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2688,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2678,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.267,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2821,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.8025175402393727,
|
| 327 |
+
"eval_loss": 0.43855828046798706,
|
| 328 |
+
"eval_macro_f1": 0.8010585423300466,
|
| 329 |
+
"eval_runtime": 11.0599,
|
| 330 |
+
"eval_samples_per_second": 438.161,
|
| 331 |
+
"eval_steps_per_second": 4.25,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2839,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2361,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.231,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2287,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2386,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.2397,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2319,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.2378,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8021048287247214,
|
| 385 |
+
"eval_loss": 0.4693628251552582,
|
| 386 |
+
"eval_macro_f1": 0.8013383852471122,
|
| 387 |
+
"eval_runtime": 11.2153,
|
| 388 |
+
"eval_samples_per_second": 432.089,
|
| 389 |
+
"eval_steps_per_second": 4.191,
|
| 390 |
+
"step": 1309
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.09,
|
| 394 |
+
"learning_rate": 4.533898305084746e-05,
|
| 395 |
+
"loss": 0.2167,
|
| 396 |
+
"step": 1325
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.22,
|
| 400 |
+
"learning_rate": 4.519774011299435e-05,
|
| 401 |
+
"loss": 0.2069,
|
| 402 |
+
"step": 1350
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.35,
|
| 406 |
+
"learning_rate": 4.505649717514124e-05,
|
| 407 |
+
"loss": 0.1866,
|
| 408 |
+
"step": 1375
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.49,
|
| 412 |
+
"learning_rate": 4.491525423728814e-05,
|
| 413 |
+
"loss": 0.1691,
|
| 414 |
+
"step": 1400
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.62,
|
| 418 |
+
"learning_rate": 4.477401129943503e-05,
|
| 419 |
+
"loss": 0.1878,
|
| 420 |
+
"step": 1425
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.75,
|
| 424 |
+
"learning_rate": 4.463276836158192e-05,
|
| 425 |
+
"loss": 0.1892,
|
| 426 |
+
"step": 1450
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.89,
|
| 430 |
+
"learning_rate": 4.4491525423728816e-05,
|
| 431 |
+
"loss": 0.2054,
|
| 432 |
+
"step": 1475
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 8.0,
|
| 436 |
+
"eval_accuracy": 0.8018984729673958,
|
| 437 |
+
"eval_loss": 0.4967544674873352,
|
| 438 |
+
"eval_macro_f1": 0.8017084877441987,
|
| 439 |
+
"eval_runtime": 10.9594,
|
| 440 |
+
"eval_samples_per_second": 442.178,
|
| 441 |
+
"eval_steps_per_second": 4.289,
|
| 442 |
+
"step": 1496
|
| 443 |
+
}
|
| 444 |
+
],
|
| 445 |
+
"max_steps": 9350,
|
| 446 |
+
"num_train_epochs": 50,
|
| 447 |
+
"total_flos": 1131791447162880.0,
|
| 448 |
+
"trial_name": null,
|
| 449 |
+
"trial_params": null
|
| 450 |
+
}
|
cross_cell_type_generization/L1/B cells/fold1/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold1/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9188888888888889,
|
| 3 |
+
"test_loss": 0.23316439986228943,
|
| 4 |
+
"test_macro_f1": 0.8977780336352279,
|
| 5 |
+
"test_runtime": 2.0883,
|
| 6 |
+
"test_samples_per_second": 430.98,
|
| 7 |
+
"test_steps_per_second": 4.31
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold2/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.88,
|
| 3 |
+
"test_loss": 0.3085847496986389,
|
| 4 |
+
"test_macro_f1": 0.8440748440748441,
|
| 5 |
+
"test_runtime": 2.0326,
|
| 6 |
+
"test_samples_per_second": 442.792,
|
| 7 |
+
"test_steps_per_second": 4.428
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold2/checkpoint-1122/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold2/checkpoint-1122/trainer_state.json
ADDED
|
@@ -0,0 +1,340 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.41548681259155273,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L1/B cells/fold2/checkpoint-748",
|
| 4 |
+
"epoch": 6.0,
|
| 5 |
+
"global_step": 1122,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.5e-06,
|
| 13 |
+
"loss": 0.6936,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6927,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 7.5e-06,
|
| 25 |
+
"loss": 0.6929,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1e-05,
|
| 31 |
+
"loss": 0.6917,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6915,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 1.5e-05,
|
| 43 |
+
"loss": 0.6912,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.94,
|
| 48 |
+
"learning_rate": 1.75e-05,
|
| 49 |
+
"loss": 0.6887,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5196037969459348,
|
| 55 |
+
"eval_loss": 0.6914771795272827,
|
| 56 |
+
"eval_macro_f1": 0.3655374347652674,
|
| 57 |
+
"eval_runtime": 11.0306,
|
| 58 |
+
"eval_samples_per_second": 439.323,
|
| 59 |
+
"eval_steps_per_second": 4.261,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6922,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6892,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6888,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6862,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6879,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6859,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6846,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5334296326867519,
|
| 107 |
+
"eval_loss": 0.6868591904640198,
|
| 108 |
+
"eval_macro_f1": 0.5119338235225336,
|
| 109 |
+
"eval_runtime": 11.2171,
|
| 110 |
+
"eval_samples_per_second": 432.018,
|
| 111 |
+
"eval_steps_per_second": 4.19,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6826,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6743,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6688,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6458,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.574,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5137,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.4851,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4636,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.773421378456459,
|
| 165 |
+
"eval_loss": 0.45748546719551086,
|
| 166 |
+
"eval_macro_f1": 0.771673459509422,
|
| 167 |
+
"eval_runtime": 11.6042,
|
| 168 |
+
"eval_samples_per_second": 417.609,
|
| 169 |
+
"eval_steps_per_second": 4.05,
|
| 170 |
+
"step": 561
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.07,
|
| 174 |
+
"learning_rate": 4.957627118644068e-05,
|
| 175 |
+
"loss": 0.4393,
|
| 176 |
+
"step": 575
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.21,
|
| 180 |
+
"learning_rate": 4.9435028248587575e-05,
|
| 181 |
+
"loss": 0.4209,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.34,
|
| 186 |
+
"learning_rate": 4.929378531073446e-05,
|
| 187 |
+
"loss": 0.3985,
|
| 188 |
+
"step": 625
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.48,
|
| 192 |
+
"learning_rate": 4.915254237288136e-05,
|
| 193 |
+
"loss": 0.3722,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.38,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.3841,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3904,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7948823772183244,
|
| 217 |
+
"eval_loss": 0.41548681259155273,
|
| 218 |
+
"eval_macro_f1": 0.7916890174572287,
|
| 219 |
+
"eval_runtime": 10.8837,
|
| 220 |
+
"eval_samples_per_second": 445.254,
|
| 221 |
+
"eval_steps_per_second": 4.318,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3807,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3181,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3138,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3275,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.318,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3021,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3126,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3294,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7880726372265786,
|
| 275 |
+
"eval_loss": 0.42716553807258606,
|
| 276 |
+
"eval_macro_f1": 0.788036813173498,
|
| 277 |
+
"eval_runtime": 10.978,
|
| 278 |
+
"eval_samples_per_second": 441.427,
|
| 279 |
+
"eval_steps_per_second": 4.281,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.2965,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.2776,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2743,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2679,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2632,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.2771,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2703,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.7992158481221626,
|
| 327 |
+
"eval_loss": 0.43729907274246216,
|
| 328 |
+
"eval_macro_f1": 0.7992148135751788,
|
| 329 |
+
"eval_runtime": 11.7995,
|
| 330 |
+
"eval_samples_per_second": 410.695,
|
| 331 |
+
"eval_steps_per_second": 3.983,
|
| 332 |
+
"step": 1122
|
| 333 |
+
}
|
| 334 |
+
],
|
| 335 |
+
"max_steps": 9350,
|
| 336 |
+
"num_train_epochs": 50,
|
| 337 |
+
"total_flos": 848843585372160.0,
|
| 338 |
+
"trial_name": null,
|
| 339 |
+
"trial_params": null
|
| 340 |
+
}
|
cross_cell_type_generization/L1/B cells/fold2/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.88,
|
| 3 |
+
"test_loss": 0.3085847496986389,
|
| 4 |
+
"test_macro_f1": 0.8440748440748441,
|
| 5 |
+
"test_runtime": 2.0326,
|
| 6 |
+
"test_samples_per_second": 442.792,
|
| 7 |
+
"test_steps_per_second": 4.428
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold3/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9155555555555556,
|
| 3 |
+
"test_loss": 0.26062139868736267,
|
| 4 |
+
"test_macro_f1": 0.8972355769230769,
|
| 5 |
+
"test_runtime": 2.2582,
|
| 6 |
+
"test_samples_per_second": 398.546,
|
| 7 |
+
"test_steps_per_second": 3.985
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold3/checkpoint-1309/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold3/checkpoint-1309/trainer_state.json
ADDED
|
@@ -0,0 +1,398 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.4226238429546356,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L1/B cells/fold3/checkpoint-935",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"global_step": 1309,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.5e-06,
|
| 13 |
+
"loss": 0.6975,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6966,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 7.5e-06,
|
| 25 |
+
"loss": 0.6943,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1e-05,
|
| 31 |
+
"loss": 0.6909,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.693,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 1.5e-05,
|
| 43 |
+
"loss": 0.6925,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.94,
|
| 48 |
+
"learning_rate": 1.75e-05,
|
| 49 |
+
"loss": 0.6932,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5154766817994222,
|
| 55 |
+
"eval_loss": 0.691068708896637,
|
| 56 |
+
"eval_macro_f1": 0.4829185650936648,
|
| 57 |
+
"eval_runtime": 12.2694,
|
| 58 |
+
"eval_samples_per_second": 394.968,
|
| 59 |
+
"eval_steps_per_second": 3.831,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.689,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6908,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6886,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6884,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6875,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6887,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6864,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5435410647957079,
|
| 107 |
+
"eval_loss": 0.6844078302383423,
|
| 108 |
+
"eval_macro_f1": 0.43449410435016383,
|
| 109 |
+
"eval_runtime": 11.3083,
|
| 110 |
+
"eval_samples_per_second": 428.535,
|
| 111 |
+
"eval_steps_per_second": 4.156,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6836,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6815,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6778,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6637,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.6206,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.565,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.5312,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4995,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7602146099876187,
|
| 165 |
+
"eval_loss": 0.4883873462677002,
|
| 166 |
+
"eval_macro_f1": 0.7547402082097463,
|
| 167 |
+
"eval_runtime": 11.9956,
|
| 168 |
+
"eval_samples_per_second": 403.983,
|
| 169 |
+
"eval_steps_per_second": 3.918,
|
| 170 |
+
"step": 561
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.07,
|
| 174 |
+
"learning_rate": 4.957627118644068e-05,
|
| 175 |
+
"loss": 0.4676,
|
| 176 |
+
"step": 575
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.21,
|
| 180 |
+
"learning_rate": 4.9435028248587575e-05,
|
| 181 |
+
"loss": 0.4293,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.34,
|
| 186 |
+
"learning_rate": 4.929378531073446e-05,
|
| 187 |
+
"loss": 0.4152,
|
| 188 |
+
"step": 625
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.48,
|
| 192 |
+
"learning_rate": 4.915254237288136e-05,
|
| 193 |
+
"loss": 0.4157,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.4014,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.4173,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.4106,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7963268675196038,
|
| 217 |
+
"eval_loss": 0.428460955619812,
|
| 218 |
+
"eval_macro_f1": 0.7931237082942566,
|
| 219 |
+
"eval_runtime": 11.505,
|
| 220 |
+
"eval_samples_per_second": 421.207,
|
| 221 |
+
"eval_steps_per_second": 4.085,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3815,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3347,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3282,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3406,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3185,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3311,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3294,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3394,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.797358646306232,
|
| 275 |
+
"eval_loss": 0.4226238429546356,
|
| 276 |
+
"eval_macro_f1": 0.79729180116589,
|
| 277 |
+
"eval_runtime": 11.1135,
|
| 278 |
+
"eval_samples_per_second": 436.045,
|
| 279 |
+
"eval_steps_per_second": 4.229,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.2855,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.2783,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2821,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2609,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2822,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.273,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2868,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.8052001650846059,
|
| 327 |
+
"eval_loss": 0.4286108911037445,
|
| 328 |
+
"eval_macro_f1": 0.8033503768417375,
|
| 329 |
+
"eval_runtime": 11.4324,
|
| 330 |
+
"eval_samples_per_second": 423.881,
|
| 331 |
+
"eval_steps_per_second": 4.111,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2971,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2251,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.2308,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2524,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2353,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.2587,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2358,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.231,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8037556747833264,
|
| 385 |
+
"eval_loss": 0.4610762596130371,
|
| 386 |
+
"eval_macro_f1": 0.8037225017799383,
|
| 387 |
+
"eval_runtime": 12.7165,
|
| 388 |
+
"eval_samples_per_second": 381.079,
|
| 389 |
+
"eval_steps_per_second": 3.696,
|
| 390 |
+
"step": 1309
|
| 391 |
+
}
|
| 392 |
+
],
|
| 393 |
+
"max_steps": 9350,
|
| 394 |
+
"num_train_epochs": 50,
|
| 395 |
+
"total_flos": 990317516267520.0,
|
| 396 |
+
"trial_name": null,
|
| 397 |
+
"trial_params": null
|
| 398 |
+
}
|
cross_cell_type_generization/L1/B cells/fold3/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold3/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9155555555555556,
|
| 3 |
+
"test_loss": 0.26062139868736267,
|
| 4 |
+
"test_macro_f1": 0.8972355769230769,
|
| 5 |
+
"test_runtime": 2.2582,
|
| 6 |
+
"test_samples_per_second": 398.546,
|
| 7 |
+
"test_steps_per_second": 3.985
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold4/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9022222222222223,
|
| 3 |
+
"test_loss": 0.24500779807567596,
|
| 4 |
+
"test_macro_f1": 0.879330099217474,
|
| 5 |
+
"test_runtime": 2.1229,
|
| 6 |
+
"test_samples_per_second": 423.954,
|
| 7 |
+
"test_steps_per_second": 4.24
|
| 8 |
+
}
|
cross_cell_type_generization/L1/B cells/fold4/checkpoint-1309/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold4/checkpoint-1309/trainer_state.json
ADDED
|
@@ -0,0 +1,398 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.41602012515068054,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L1/B cells/fold4/checkpoint-935",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"global_step": 1309,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.5e-06,
|
| 13 |
+
"loss": 0.6966,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6947,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 7.5e-06,
|
| 25 |
+
"loss": 0.6926,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1e-05,
|
| 31 |
+
"loss": 0.6934,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6932,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 1.5e-05,
|
| 43 |
+
"loss": 0.6921,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.94,
|
| 48 |
+
"learning_rate": 1.75e-05,
|
| 49 |
+
"loss": 0.691,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5196037969459348,
|
| 55 |
+
"eval_loss": 0.6908240914344788,
|
| 56 |
+
"eval_macro_f1": 0.43146980142166214,
|
| 57 |
+
"eval_runtime": 12.083,
|
| 58 |
+
"eval_samples_per_second": 401.06,
|
| 59 |
+
"eval_steps_per_second": 3.89,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6905,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6897,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6886,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6881,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6855,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6853,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6851,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5536524969046637,
|
| 107 |
+
"eval_loss": 0.6839553713798523,
|
| 108 |
+
"eval_macro_f1": 0.5497177718485124,
|
| 109 |
+
"eval_runtime": 12.1721,
|
| 110 |
+
"eval_samples_per_second": 398.123,
|
| 111 |
+
"eval_steps_per_second": 3.861,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6851,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6776,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6691,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6484,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5782,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5506,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.511,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4786,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7593891869583161,
|
| 165 |
+
"eval_loss": 0.47733813524246216,
|
| 166 |
+
"eval_macro_f1": 0.756202927064996,
|
| 167 |
+
"eval_runtime": 11.4938,
|
| 168 |
+
"eval_samples_per_second": 421.617,
|
| 169 |
+
"eval_steps_per_second": 4.089,
|
| 170 |
+
"step": 561
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.07,
|
| 174 |
+
"learning_rate": 4.957627118644068e-05,
|
| 175 |
+
"loss": 0.4523,
|
| 176 |
+
"step": 575
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.21,
|
| 180 |
+
"learning_rate": 4.9435028248587575e-05,
|
| 181 |
+
"loss": 0.4184,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.34,
|
| 186 |
+
"learning_rate": 4.929378531073446e-05,
|
| 187 |
+
"loss": 0.414,
|
| 188 |
+
"step": 625
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.48,
|
| 192 |
+
"learning_rate": 4.915254237288136e-05,
|
| 193 |
+
"loss": 0.4262,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.4006,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.3851,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3987,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7903425505571605,
|
| 217 |
+
"eval_loss": 0.4233468472957611,
|
| 218 |
+
"eval_macro_f1": 0.7890316273278981,
|
| 219 |
+
"eval_runtime": 11.5586,
|
| 220 |
+
"eval_samples_per_second": 419.254,
|
| 221 |
+
"eval_steps_per_second": 4.066,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3708,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3086,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3368,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3348,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3229,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3289,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.318,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.331,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7981840693355344,
|
| 275 |
+
"eval_loss": 0.41602012515068054,
|
| 276 |
+
"eval_macro_f1": 0.7981113046198609,
|
| 277 |
+
"eval_runtime": 11.2785,
|
| 278 |
+
"eval_samples_per_second": 429.665,
|
| 279 |
+
"eval_steps_per_second": 4.167,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.3089,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.2626,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2881,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2881,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2805,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.2881,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2641,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.7994222038794883,
|
| 327 |
+
"eval_loss": 0.43916967511177063,
|
| 328 |
+
"eval_macro_f1": 0.7994098697093706,
|
| 329 |
+
"eval_runtime": 11.6288,
|
| 330 |
+
"eval_samples_per_second": 416.724,
|
| 331 |
+
"eval_steps_per_second": 4.042,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2693,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2494,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.2127,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2361,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2393,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.265,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.238,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.2111,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.7988031366075113,
|
| 385 |
+
"eval_loss": 0.47231557965278625,
|
| 386 |
+
"eval_macro_f1": 0.7987872940426756,
|
| 387 |
+
"eval_runtime": 11.4803,
|
| 388 |
+
"eval_samples_per_second": 422.115,
|
| 389 |
+
"eval_steps_per_second": 4.094,
|
| 390 |
+
"step": 1309
|
| 391 |
+
}
|
| 392 |
+
],
|
| 393 |
+
"max_steps": 9350,
|
| 394 |
+
"num_train_epochs": 50,
|
| 395 |
+
"total_flos": 990317516267520.0,
|
| 396 |
+
"trial_name": null,
|
| 397 |
+
"trial_params": null
|
| 398 |
+
}
|
cross_cell_type_generization/L1/B cells/fold4/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/B cells/fold4/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9022222222222223,
|
| 3 |
+
"test_loss": 0.24500779807567596,
|
| 4 |
+
"test_macro_f1": 0.879330099217474,
|
| 5 |
+
"test_runtime": 2.1229,
|
| 6 |
+
"test_samples_per_second": 423.954,
|
| 7 |
+
"test_steps_per_second": 4.24
|
| 8 |
+
}
|
cross_cell_type_generization/L1/Epitheliums/fold1/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7411111111111112,
|
| 3 |
+
"test_loss": 0.5591853260993958,
|
| 4 |
+
"test_macro_f1": 0.6959067041188183,
|
| 5 |
+
"test_runtime": 2.1685,
|
| 6 |
+
"test_samples_per_second": 415.029,
|
| 7 |
+
"test_steps_per_second": 4.15
|
| 8 |
+
}
|
cross_cell_type_generization/L1/Epitheliums/fold1/checkpoint-1496/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/Epitheliums/fold1/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 1,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L1/Epitheliums/fold1/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7411111111111112,
|
| 3 |
+
"test_loss": 0.5591853260993958,
|
| 4 |
+
"test_macro_f1": 0.6959067041188183,
|
| 5 |
+
"test_runtime": 2.1685,
|
| 6 |
+
"test_samples_per_second": 415.029,
|
| 7 |
+
"test_steps_per_second": 4.15
|
| 8 |
+
}
|
cross_cell_type_generization/L6/.DS_Store
ADDED
|
Binary file (26.6 kB). View file
|
|
|
cross_cell_type_generization/L6/NKT/.DS_Store
ADDED
|
Binary file (6.15 kB). View file
|
|
|
cross_cell_type_generization/L6/NKT/fold0/.DS_Store
ADDED
|
Binary file (6.15 kB). View file
|
|
|
cross_cell_type_generization/L6/NKT/fold0/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.96,
|
| 3 |
+
"test_loss": 0.1425999402999878,
|
| 4 |
+
"test_macro_f1": 0.9564185753716528,
|
| 5 |
+
"test_runtime": 3.1636,
|
| 6 |
+
"test_samples_per_second": 284.49,
|
| 7 |
+
"test_steps_per_second": 2.845
|
| 8 |
+
}
|
cross_cell_type_generization/L6/NKT/fold0/checkpoint-1683/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 6,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L6/NKT/fold0/checkpoint-1683/trainer_state.json
ADDED
|
@@ -0,0 +1,508 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.417743980884552,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L6/NKT/fold0/checkpoint-1309",
|
| 4 |
+
"epoch": 9.0,
|
| 5 |
+
"global_step": 1683,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.5e-06,
|
| 13 |
+
"loss": 0.694,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6928,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 7.5e-06,
|
| 25 |
+
"loss": 0.6911,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1e-05,
|
| 31 |
+
"loss": 0.6923,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6924,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 1.5e-05,
|
| 43 |
+
"loss": 0.6917,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.94,
|
| 48 |
+
"learning_rate": 1.75e-05,
|
| 49 |
+
"loss": 0.6913,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5395089746234785,
|
| 55 |
+
"eval_loss": 0.6894506216049194,
|
| 56 |
+
"eval_macro_f1": 0.525770112056468,
|
| 57 |
+
"eval_runtime": 16.5426,
|
| 58 |
+
"eval_samples_per_second": 293.001,
|
| 59 |
+
"eval_steps_per_second": 2.841,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6927,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6894,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6873,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6859,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6835,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6786,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6748,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.6063544460491025,
|
| 107 |
+
"eval_loss": 0.6603953838348389,
|
| 108 |
+
"eval_macro_f1": 0.6057031030160396,
|
| 109 |
+
"eval_runtime": 17.079,
|
| 110 |
+
"eval_samples_per_second": 283.799,
|
| 111 |
+
"eval_steps_per_second": 2.752,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6734,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6421,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6333,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6114,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5595,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5133,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.5002,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4708,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7711986795956262,
|
| 165 |
+
"eval_loss": 0.47740572690963745,
|
| 166 |
+
"eval_macro_f1": 0.7649547260041831,
|
| 167 |
+
"eval_runtime": 17.4796,
|
| 168 |
+
"eval_samples_per_second": 277.295,
|
| 169 |
+
"eval_steps_per_second": 2.689,
|
| 170 |
+
"step": 561
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.07,
|
| 174 |
+
"learning_rate": 4.957627118644068e-05,
|
| 175 |
+
"loss": 0.439,
|
| 176 |
+
"step": 575
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.21,
|
| 180 |
+
"learning_rate": 4.9435028248587575e-05,
|
| 181 |
+
"loss": 0.4296,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.34,
|
| 186 |
+
"learning_rate": 4.929378531073446e-05,
|
| 187 |
+
"loss": 0.4112,
|
| 188 |
+
"step": 625
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.48,
|
| 192 |
+
"learning_rate": 4.915254237288136e-05,
|
| 193 |
+
"loss": 0.4166,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.4142,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.4433,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3856,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7808953992160099,
|
| 217 |
+
"eval_loss": 0.4686849117279053,
|
| 218 |
+
"eval_macro_f1": 0.7725768398803927,
|
| 219 |
+
"eval_runtime": 16.5983,
|
| 220 |
+
"eval_samples_per_second": 292.018,
|
| 221 |
+
"eval_steps_per_second": 2.832,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.4305,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.4142,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3401,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3451,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.352,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3708,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3681,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3474,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.8046214153084382,
|
| 275 |
+
"eval_loss": 0.4309827387332916,
|
| 276 |
+
"eval_macro_f1": 0.8045654803987876,
|
| 277 |
+
"eval_runtime": 16.7398,
|
| 278 |
+
"eval_samples_per_second": 289.549,
|
| 279 |
+
"eval_steps_per_second": 2.808,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.32,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.3421,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.3059,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2963,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.3126,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.3733,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2956,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.80647823395915,
|
| 327 |
+
"eval_loss": 0.4286085367202759,
|
| 328 |
+
"eval_macro_f1": 0.8027428225326398,
|
| 329 |
+
"eval_runtime": 17.2051,
|
| 330 |
+
"eval_samples_per_second": 281.719,
|
| 331 |
+
"eval_steps_per_second": 2.732,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.3157,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.258,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.2673,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2591,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2875,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.271,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2895,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.3092,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8132865690117599,
|
| 385 |
+
"eval_loss": 0.417743980884552,
|
| 386 |
+
"eval_macro_f1": 0.8114060731274264,
|
| 387 |
+
"eval_runtime": 16.8223,
|
| 388 |
+
"eval_samples_per_second": 288.13,
|
| 389 |
+
"eval_steps_per_second": 2.794,
|
| 390 |
+
"step": 1309
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.09,
|
| 394 |
+
"learning_rate": 4.533898305084746e-05,
|
| 395 |
+
"loss": 0.301,
|
| 396 |
+
"step": 1325
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.22,
|
| 400 |
+
"learning_rate": 4.519774011299435e-05,
|
| 401 |
+
"loss": 0.2618,
|
| 402 |
+
"step": 1350
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.35,
|
| 406 |
+
"learning_rate": 4.505649717514124e-05,
|
| 407 |
+
"loss": 0.2755,
|
| 408 |
+
"step": 1375
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.49,
|
| 412 |
+
"learning_rate": 4.491525423728814e-05,
|
| 413 |
+
"loss": 0.2284,
|
| 414 |
+
"step": 1400
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.62,
|
| 418 |
+
"learning_rate": 4.477401129943503e-05,
|
| 419 |
+
"loss": 0.2394,
|
| 420 |
+
"step": 1425
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.75,
|
| 424 |
+
"learning_rate": 4.463276836158192e-05,
|
| 425 |
+
"loss": 0.2741,
|
| 426 |
+
"step": 1450
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.89,
|
| 430 |
+
"learning_rate": 4.4491525423728816e-05,
|
| 431 |
+
"loss": 0.2554,
|
| 432 |
+
"step": 1475
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 8.0,
|
| 436 |
+
"eval_accuracy": 0.7893542397359191,
|
| 437 |
+
"eval_loss": 0.5291871428489685,
|
| 438 |
+
"eval_macro_f1": 0.781122122703469,
|
| 439 |
+
"eval_runtime": 16.8608,
|
| 440 |
+
"eval_samples_per_second": 287.472,
|
| 441 |
+
"eval_steps_per_second": 2.788,
|
| 442 |
+
"step": 1496
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 8.02,
|
| 446 |
+
"learning_rate": 4.435028248587571e-05,
|
| 447 |
+
"loss": 0.247,
|
| 448 |
+
"step": 1500
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 8.16,
|
| 452 |
+
"learning_rate": 4.42090395480226e-05,
|
| 453 |
+
"loss": 0.2439,
|
| 454 |
+
"step": 1525
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 8.29,
|
| 458 |
+
"learning_rate": 4.4067796610169495e-05,
|
| 459 |
+
"loss": 0.2337,
|
| 460 |
+
"step": 1550
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 8.42,
|
| 464 |
+
"learning_rate": 4.392655367231638e-05,
|
| 465 |
+
"loss": 0.2158,
|
| 466 |
+
"step": 1575
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 8.56,
|
| 470 |
+
"learning_rate": 4.378531073446328e-05,
|
| 471 |
+
"loss": 0.244,
|
| 472 |
+
"step": 1600
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 8.69,
|
| 476 |
+
"learning_rate": 4.3644067796610175e-05,
|
| 477 |
+
"loss": 0.2885,
|
| 478 |
+
"step": 1625
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 8.82,
|
| 482 |
+
"learning_rate": 4.350282485875706e-05,
|
| 483 |
+
"loss": 0.2534,
|
| 484 |
+
"step": 1650
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 8.96,
|
| 488 |
+
"learning_rate": 4.3361581920903954e-05,
|
| 489 |
+
"loss": 0.2277,
|
| 490 |
+
"step": 1675
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 9.0,
|
| 494 |
+
"eval_accuracy": 0.7924489374871054,
|
| 495 |
+
"eval_loss": 0.5165997743606567,
|
| 496 |
+
"eval_macro_f1": 0.7853909691785066,
|
| 497 |
+
"eval_runtime": 17.4093,
|
| 498 |
+
"eval_samples_per_second": 278.415,
|
| 499 |
+
"eval_steps_per_second": 2.7,
|
| 500 |
+
"step": 1683
|
| 501 |
+
}
|
| 502 |
+
],
|
| 503 |
+
"max_steps": 9350,
|
| 504 |
+
"num_train_epochs": 50,
|
| 505 |
+
"total_flos": 6923004316286976.0,
|
| 506 |
+
"trial_name": null,
|
| 507 |
+
"trial_params": null
|
| 508 |
+
}
|
cross_cell_type_generization/L6/NKT/fold0/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 6,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L6/NKT/fold0/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.96,
|
| 3 |
+
"test_loss": 0.1425999402999878,
|
| 4 |
+
"test_macro_f1": 0.9564185753716528,
|
| 5 |
+
"test_runtime": 3.1636,
|
| 6 |
+
"test_samples_per_second": 284.49,
|
| 7 |
+
"test_steps_per_second": 2.845
|
| 8 |
+
}
|
cross_cell_type_generization/L6/NKT/fold1/.DS_Store
ADDED
|
Binary file (6.15 kB). View file
|
|
|
cross_cell_type_generization/L6/NKT/fold1/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9511111111111111,
|
| 3 |
+
"test_loss": 0.1433023363351822,
|
| 4 |
+
"test_macro_f1": 0.9459545035184165,
|
| 5 |
+
"test_runtime": 3.0683,
|
| 6 |
+
"test_samples_per_second": 293.317,
|
| 7 |
+
"test_steps_per_second": 2.933
|
| 8 |
+
}
|
cross_cell_type_generization/L6/NKT/fold1/checkpoint-1870/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 6,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L6/NKT/fold1/checkpoint-1870/trainer_state.json
ADDED
|
@@ -0,0 +1,560 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3829525113105774,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L6/NKT/fold1/checkpoint-1496",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 1870,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.5e-06,
|
| 13 |
+
"loss": 0.6951,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6938,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 7.5e-06,
|
| 25 |
+
"loss": 0.6933,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1e-05,
|
| 31 |
+
"loss": 0.6931,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.693,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 1.5e-05,
|
| 43 |
+
"loss": 0.6925,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.94,
|
| 48 |
+
"learning_rate": 1.75e-05,
|
| 49 |
+
"loss": 0.6893,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5303342963268676,
|
| 55 |
+
"eval_loss": 0.690353274345398,
|
| 56 |
+
"eval_macro_f1": 0.5104776766732619,
|
| 57 |
+
"eval_runtime": 17.1866,
|
| 58 |
+
"eval_samples_per_second": 281.963,
|
| 59 |
+
"eval_steps_per_second": 2.735,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6919,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6883,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6852,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6881,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6838,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6822,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6742,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.6027651671481634,
|
| 107 |
+
"eval_loss": 0.660984218120575,
|
| 108 |
+
"eval_macro_f1": 0.6025090901297125,
|
| 109 |
+
"eval_runtime": 16.6024,
|
| 110 |
+
"eval_samples_per_second": 291.885,
|
| 111 |
+
"eval_steps_per_second": 2.831,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6668,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6434,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6147,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.5699,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5541,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.493,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.4732,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4689,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7791993396615765,
|
| 165 |
+
"eval_loss": 0.47226572036743164,
|
| 166 |
+
"eval_macro_f1": 0.7736915474651473,
|
| 167 |
+
"eval_runtime": 16.87,
|
| 168 |
+
"eval_samples_per_second": 287.255,
|
| 169 |
+
"eval_steps_per_second": 2.786,
|
| 170 |
+
"step": 561
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.07,
|
| 174 |
+
"learning_rate": 4.957627118644068e-05,
|
| 175 |
+
"loss": 0.4375,
|
| 176 |
+
"step": 575
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.21,
|
| 180 |
+
"learning_rate": 4.9435028248587575e-05,
|
| 181 |
+
"loss": 0.4017,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.34,
|
| 186 |
+
"learning_rate": 4.929378531073446e-05,
|
| 187 |
+
"loss": 0.4196,
|
| 188 |
+
"step": 625
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.48,
|
| 192 |
+
"learning_rate": 4.915254237288136e-05,
|
| 193 |
+
"loss": 0.3939,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.3995,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.4203,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3996,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.8002476269087908,
|
| 217 |
+
"eval_loss": 0.4205825626850128,
|
| 218 |
+
"eval_macro_f1": 0.8001984841940988,
|
| 219 |
+
"eval_runtime": 17.7363,
|
| 220 |
+
"eval_samples_per_second": 273.225,
|
| 221 |
+
"eval_steps_per_second": 2.65,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3667,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3532,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3465,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3307,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3509,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3422,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.4064,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3749,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.8078827899298391,
|
| 275 |
+
"eval_loss": 0.41957414150238037,
|
| 276 |
+
"eval_macro_f1": 0.8035829663233104,
|
| 277 |
+
"eval_runtime": 17.682,
|
| 278 |
+
"eval_samples_per_second": 274.065,
|
| 279 |
+
"eval_steps_per_second": 2.658,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.3324,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.3093,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2799,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.3289,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.289,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.2873,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2904,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.7946760214609988,
|
| 327 |
+
"eval_loss": 0.46021702885627747,
|
| 328 |
+
"eval_macro_f1": 0.7929614261696445,
|
| 329 |
+
"eval_runtime": 17.0343,
|
| 330 |
+
"eval_samples_per_second": 284.485,
|
| 331 |
+
"eval_steps_per_second": 2.759,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2959,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2872,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.3003,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2779,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2427,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.2876,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2871,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.295,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8264548080891457,
|
| 385 |
+
"eval_loss": 0.391053169965744,
|
| 386 |
+
"eval_macro_f1": 0.8261616614660012,
|
| 387 |
+
"eval_runtime": 16.8533,
|
| 388 |
+
"eval_samples_per_second": 287.541,
|
| 389 |
+
"eval_steps_per_second": 2.789,
|
| 390 |
+
"step": 1309
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.09,
|
| 394 |
+
"learning_rate": 4.533898305084746e-05,
|
| 395 |
+
"loss": 0.2255,
|
| 396 |
+
"step": 1325
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.22,
|
| 400 |
+
"learning_rate": 4.519774011299435e-05,
|
| 401 |
+
"loss": 0.2355,
|
| 402 |
+
"step": 1350
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.35,
|
| 406 |
+
"learning_rate": 4.505649717514124e-05,
|
| 407 |
+
"loss": 0.2438,
|
| 408 |
+
"step": 1375
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.49,
|
| 412 |
+
"learning_rate": 4.491525423728814e-05,
|
| 413 |
+
"loss": 0.2269,
|
| 414 |
+
"step": 1400
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.62,
|
| 418 |
+
"learning_rate": 4.477401129943503e-05,
|
| 419 |
+
"loss": 0.266,
|
| 420 |
+
"step": 1425
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.75,
|
| 424 |
+
"learning_rate": 4.463276836158192e-05,
|
| 425 |
+
"loss": 0.2885,
|
| 426 |
+
"step": 1450
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.89,
|
| 430 |
+
"learning_rate": 4.4491525423728816e-05,
|
| 431 |
+
"loss": 0.235,
|
| 432 |
+
"step": 1475
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 8.0,
|
| 436 |
+
"eval_accuracy": 0.8326454808089145,
|
| 437 |
+
"eval_loss": 0.3829525113105774,
|
| 438 |
+
"eval_macro_f1": 0.8325607691090193,
|
| 439 |
+
"eval_runtime": 17.0079,
|
| 440 |
+
"eval_samples_per_second": 284.927,
|
| 441 |
+
"eval_steps_per_second": 2.763,
|
| 442 |
+
"step": 1496
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 8.02,
|
| 446 |
+
"learning_rate": 4.435028248587571e-05,
|
| 447 |
+
"loss": 0.2555,
|
| 448 |
+
"step": 1500
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 8.16,
|
| 452 |
+
"learning_rate": 4.42090395480226e-05,
|
| 453 |
+
"loss": 0.2347,
|
| 454 |
+
"step": 1525
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 8.29,
|
| 458 |
+
"learning_rate": 4.4067796610169495e-05,
|
| 459 |
+
"loss": 0.2535,
|
| 460 |
+
"step": 1550
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 8.42,
|
| 464 |
+
"learning_rate": 4.392655367231638e-05,
|
| 465 |
+
"loss": 0.2028,
|
| 466 |
+
"step": 1575
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 8.56,
|
| 470 |
+
"learning_rate": 4.378531073446328e-05,
|
| 471 |
+
"loss": 0.2398,
|
| 472 |
+
"step": 1600
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 8.69,
|
| 476 |
+
"learning_rate": 4.3644067796610175e-05,
|
| 477 |
+
"loss": 0.2299,
|
| 478 |
+
"step": 1625
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 8.82,
|
| 482 |
+
"learning_rate": 4.350282485875706e-05,
|
| 483 |
+
"loss": 0.2186,
|
| 484 |
+
"step": 1650
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 8.96,
|
| 488 |
+
"learning_rate": 4.3361581920903954e-05,
|
| 489 |
+
"loss": 0.2387,
|
| 490 |
+
"step": 1675
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 9.0,
|
| 494 |
+
"eval_accuracy": 0.8291374329343789,
|
| 495 |
+
"eval_loss": 0.4115827977657318,
|
| 496 |
+
"eval_macro_f1": 0.8287243066483084,
|
| 497 |
+
"eval_runtime": 16.9786,
|
| 498 |
+
"eval_samples_per_second": 285.418,
|
| 499 |
+
"eval_steps_per_second": 2.768,
|
| 500 |
+
"step": 1683
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 9.09,
|
| 504 |
+
"learning_rate": 4.3220338983050854e-05,
|
| 505 |
+
"loss": 0.2168,
|
| 506 |
+
"step": 1700
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 9.22,
|
| 510 |
+
"learning_rate": 4.307909604519774e-05,
|
| 511 |
+
"loss": 0.2242,
|
| 512 |
+
"step": 1725
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 9.36,
|
| 516 |
+
"learning_rate": 4.2937853107344634e-05,
|
| 517 |
+
"loss": 0.2278,
|
| 518 |
+
"step": 1750
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 9.49,
|
| 522 |
+
"learning_rate": 4.279661016949153e-05,
|
| 523 |
+
"loss": 0.2239,
|
| 524 |
+
"step": 1775
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 9.63,
|
| 528 |
+
"learning_rate": 4.265536723163842e-05,
|
| 529 |
+
"loss": 0.205,
|
| 530 |
+
"step": 1800
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 9.76,
|
| 534 |
+
"learning_rate": 4.251412429378531e-05,
|
| 535 |
+
"loss": 0.1794,
|
| 536 |
+
"step": 1825
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 9.89,
|
| 540 |
+
"learning_rate": 4.2372881355932206e-05,
|
| 541 |
+
"loss": 0.2228,
|
| 542 |
+
"step": 1850
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 10.0,
|
| 546 |
+
"eval_accuracy": 0.8295501444490301,
|
| 547 |
+
"eval_loss": 0.41377758979797363,
|
| 548 |
+
"eval_macro_f1": 0.8290193729257451,
|
| 549 |
+
"eval_runtime": 16.541,
|
| 550 |
+
"eval_samples_per_second": 292.969,
|
| 551 |
+
"eval_steps_per_second": 2.841,
|
| 552 |
+
"step": 1870
|
| 553 |
+
}
|
| 554 |
+
],
|
| 555 |
+
"max_steps": 9350,
|
| 556 |
+
"num_train_epochs": 50,
|
| 557 |
+
"total_flos": 7692623851929600.0,
|
| 558 |
+
"trial_name": null,
|
| 559 |
+
"trial_params": null
|
| 560 |
+
}
|
cross_cell_type_generization/L6/NKT/fold1/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 6,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L6/NKT/fold1/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9511111111111111,
|
| 3 |
+
"test_loss": 0.1433023363351822,
|
| 4 |
+
"test_macro_f1": 0.9459545035184165,
|
| 5 |
+
"test_runtime": 3.0683,
|
| 6 |
+
"test_samples_per_second": 293.317,
|
| 7 |
+
"test_steps_per_second": 2.933
|
| 8 |
+
}
|
cross_cell_type_generization/L6/NKT/fold2/.DS_Store
ADDED
|
Binary file (6.15 kB). View file
|
|
|
cross_cell_type_generization/L6/NKT/fold2/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9311111111111111,
|
| 3 |
+
"test_loss": 0.18912672996520996,
|
| 4 |
+
"test_macro_f1": 0.9234928922428922,
|
| 5 |
+
"test_runtime": 3.2414,
|
| 6 |
+
"test_samples_per_second": 277.658,
|
| 7 |
+
"test_steps_per_second": 2.777
|
| 8 |
+
}
|
cross_cell_type_generization/L6/NKT/fold2/checkpoint-1496/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 6,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L6/NKT/fold2/checkpoint-1496/trainer_state.json
ADDED
|
@@ -0,0 +1,450 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.41087305545806885,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L6/NKT/fold2/checkpoint-1122",
|
| 4 |
+
"epoch": 8.0,
|
| 5 |
+
"global_step": 1496,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.5e-06,
|
| 13 |
+
"loss": 0.6987,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6951,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 7.5e-06,
|
| 25 |
+
"loss": 0.6932,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1e-05,
|
| 31 |
+
"loss": 0.694,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6937,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 1.5e-05,
|
| 43 |
+
"loss": 0.6922,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.94,
|
| 48 |
+
"learning_rate": 1.75e-05,
|
| 49 |
+
"loss": 0.6889,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5307470078415187,
|
| 55 |
+
"eval_loss": 0.6906038522720337,
|
| 56 |
+
"eval_macro_f1": 0.49024302240380135,
|
| 57 |
+
"eval_runtime": 17.1655,
|
| 58 |
+
"eval_samples_per_second": 282.31,
|
| 59 |
+
"eval_steps_per_second": 2.738,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6925,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6918,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6885,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6851,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6863,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.682,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6781,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5938918695831614,
|
| 107 |
+
"eval_loss": 0.6697625517845154,
|
| 108 |
+
"eval_macro_f1": 0.5729142185663925,
|
| 109 |
+
"eval_runtime": 16.8355,
|
| 110 |
+
"eval_samples_per_second": 287.844,
|
| 111 |
+
"eval_steps_per_second": 2.792,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6756,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6543,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6365,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.5942,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5408,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5022,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.4708,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4587,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.767849773008667,
|
| 165 |
+
"eval_loss": 0.4629926085472107,
|
| 166 |
+
"eval_macro_f1": 0.7678314931261642,
|
| 167 |
+
"eval_runtime": 18.0211,
|
| 168 |
+
"eval_samples_per_second": 268.907,
|
| 169 |
+
"eval_steps_per_second": 2.608,
|
| 170 |
+
"step": 561
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.07,
|
| 174 |
+
"learning_rate": 4.957627118644068e-05,
|
| 175 |
+
"loss": 0.4487,
|
| 176 |
+
"step": 575
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.21,
|
| 180 |
+
"learning_rate": 4.9435028248587575e-05,
|
| 181 |
+
"loss": 0.4548,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.34,
|
| 186 |
+
"learning_rate": 4.929378531073446e-05,
|
| 187 |
+
"loss": 0.4656,
|
| 188 |
+
"step": 625
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.48,
|
| 192 |
+
"learning_rate": 4.915254237288136e-05,
|
| 193 |
+
"loss": 0.4017,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.3966,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.4144,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3979,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7891044160132068,
|
| 217 |
+
"eval_loss": 0.44995540380477905,
|
| 218 |
+
"eval_macro_f1": 0.7816395292955747,
|
| 219 |
+
"eval_runtime": 16.7493,
|
| 220 |
+
"eval_samples_per_second": 289.325,
|
| 221 |
+
"eval_steps_per_second": 2.806,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3979,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3668,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3506,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3736,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3474,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3536,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3387,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3688,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7981840693355344,
|
| 275 |
+
"eval_loss": 0.43727147579193115,
|
| 276 |
+
"eval_macro_f1": 0.7921452329326568,
|
| 277 |
+
"eval_runtime": 16.8861,
|
| 278 |
+
"eval_samples_per_second": 286.982,
|
| 279 |
+
"eval_steps_per_second": 2.783,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.3756,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.3078,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.3634,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.3636,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.303,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.3177,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.3206,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.8128353281056542,
|
| 327 |
+
"eval_loss": 0.41087305545806885,
|
| 328 |
+
"eval_macro_f1": 0.8109099341158665,
|
| 329 |
+
"eval_runtime": 16.9362,
|
| 330 |
+
"eval_samples_per_second": 286.132,
|
| 331 |
+
"eval_steps_per_second": 2.775,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.309,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.3354,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.2847,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.3138,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.283,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.293,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.3043,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.2837,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8126289723483285,
|
| 385 |
+
"eval_loss": 0.4173625707626343,
|
| 386 |
+
"eval_macro_f1": 0.8122696871662047,
|
| 387 |
+
"eval_runtime": 16.9306,
|
| 388 |
+
"eval_samples_per_second": 286.226,
|
| 389 |
+
"eval_steps_per_second": 2.776,
|
| 390 |
+
"step": 1309
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.09,
|
| 394 |
+
"learning_rate": 4.533898305084746e-05,
|
| 395 |
+
"loss": 0.2563,
|
| 396 |
+
"step": 1325
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.22,
|
| 400 |
+
"learning_rate": 4.519774011299435e-05,
|
| 401 |
+
"loss": 0.2583,
|
| 402 |
+
"step": 1350
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.35,
|
| 406 |
+
"learning_rate": 4.505649717514124e-05,
|
| 407 |
+
"loss": 0.2649,
|
| 408 |
+
"step": 1375
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.49,
|
| 412 |
+
"learning_rate": 4.491525423728814e-05,
|
| 413 |
+
"loss": 0.2525,
|
| 414 |
+
"step": 1400
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.62,
|
| 418 |
+
"learning_rate": 4.477401129943503e-05,
|
| 419 |
+
"loss": 0.2958,
|
| 420 |
+
"step": 1425
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.75,
|
| 424 |
+
"learning_rate": 4.463276836158192e-05,
|
| 425 |
+
"loss": 0.2901,
|
| 426 |
+
"step": 1450
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.89,
|
| 430 |
+
"learning_rate": 4.4491525423728816e-05,
|
| 431 |
+
"loss": 0.2582,
|
| 432 |
+
"step": 1475
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 8.0,
|
| 436 |
+
"eval_accuracy": 0.799009492364837,
|
| 437 |
+
"eval_loss": 0.47358283400535583,
|
| 438 |
+
"eval_macro_f1": 0.7982465589099031,
|
| 439 |
+
"eval_runtime": 17.7707,
|
| 440 |
+
"eval_samples_per_second": 272.695,
|
| 441 |
+
"eval_steps_per_second": 2.645,
|
| 442 |
+
"step": 1496
|
| 443 |
+
}
|
| 444 |
+
],
|
| 445 |
+
"max_steps": 9350,
|
| 446 |
+
"num_train_epochs": 50,
|
| 447 |
+
"total_flos": 6154099081543680.0,
|
| 448 |
+
"trial_name": null,
|
| 449 |
+
"trial_params": null
|
| 450 |
+
}
|
cross_cell_type_generization/L6/NKT/fold2/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForSequenceClassification"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.02,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.02,
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 512,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 2048,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 4,
|
| 16 |
+
"num_hidden_layers": 6,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"problem_type": "single_label_classification",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.28.0",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
cross_cell_type_generization/L6/NKT/fold2/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9311111111111111,
|
| 3 |
+
"test_loss": 0.18912672996520996,
|
| 4 |
+
"test_macro_f1": 0.9234928922428922,
|
| 5 |
+
"test_runtime": 3.2414,
|
| 6 |
+
"test_samples_per_second": 277.658,
|
| 7 |
+
"test_steps_per_second": 2.777
|
| 8 |
+
}
|
cross_cell_type_generization/L6/NKT/fold3/.DS_Store
ADDED
|
Binary file (6.15 kB). View file
|
|
|
cross_cell_type_generization/L6/NKT/fold3/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9366666666666666,
|
| 3 |
+
"test_loss": 0.1639990210533142,
|
| 4 |
+
"test_macro_f1": 0.9294972314263685,
|
| 5 |
+
"test_runtime": 3.3198,
|
| 6 |
+
"test_samples_per_second": 271.102,
|
| 7 |
+
"test_steps_per_second": 2.711
|
| 8 |
+
}
|