Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- .DS_Store +0 -0
- cross_cell_type_generization/.DS_Store +0 -0
- cross_cell_type_generization/L4/.DS_Store +0 -0
- cross_cell_type_generization/L4/Epitheliums/.DS_Store +0 -0
- cross_cell_type_generization/L4/Epitheliums/config.json +25 -0
- cross_cell_type_generization/L4/Epitheliums/eval_results.json +8 -0
- cross_cell_type_generization/L4/Macrophages/.DS_Store +0 -0
- cross_cell_type_generization/L4/Macrophages/all_results.json +8 -0
- cross_cell_type_generization/L4/Macrophages/config.json +25 -0
- cross_cell_type_generization/L4/Macrophages/eval_results.json +8 -0
- cross_cell_type_generization/L4/Macrophages/trainer_state.json +450 -0
- cross_cell_type_generization/L4/NKT/.DS_Store +0 -0
- cross_cell_type_generization/L4/NKT/all_results.json +8 -0
- cross_cell_type_generization/L4/NKT/config.json +25 -0
- cross_cell_type_generization/L4/NKT/eval_results.json +8 -0
- cross_cell_type_generization/L4/NKT/trainer_state.json +450 -0
- cross_cell_type_generization/L4/NKs/.DS_Store +0 -0
- cross_cell_type_generization/L4/NKs/all_results.json +8 -0
- cross_cell_type_generization/L4/NKs/config.json +25 -0
- cross_cell_type_generization/L4/NKs/eval_results.json +8 -0
- cross_cell_type_generization/L4/NKs/trainer_state.json +450 -0
- cross_cell_type_generization/L4/T cells/.DS_Store +0 -0
- cross_cell_type_generization/L4/T cells/all_results.json +8 -0
- cross_cell_type_generization/L4/T cells/config.json +25 -0
- cross_cell_type_generization/L4/T cells/eval_results.json +8 -0
- cross_cell_type_generization/L4/T cells/trainer_state.json +450 -0
- data_curation&frozen_layers/.DS_Store +0 -0
- data_curation&frozen_layers/dataset1_all/.DS_Store +0 -0
- data_curation&frozen_layers/dataset1_all/F0/.DS_Store +0 -0
- data_curation&frozen_layers/dataset1_all/F0/all_results.json +8 -0
- data_curation&frozen_layers/dataset1_all/F0/config.json +27 -0
- data_curation&frozen_layers/dataset1_all/F0/eval_results.json +8 -0
- data_curation&frozen_layers/dataset1_all/F0/trainer_state.json +398 -0
- data_curation&frozen_layers/dataset1_all/F2/.DS_Store +0 -0
- data_curation&frozen_layers/dataset1_all/F2/all_results.json +8 -0
- data_curation&frozen_layers/dataset1_all/F2/config.json +27 -0
- data_curation&frozen_layers/dataset1_all/F2/eval_results.json +8 -0
- data_curation&frozen_layers/dataset1_all/F2/predictions.pickle +3 -0
- data_curation&frozen_layers/dataset1_all/F2/rng_state.pth +3 -0
- data_curation&frozen_layers/dataset1_all/F2/scheduler.pt +3 -0
- data_curation&frozen_layers/dataset1_all/F2/trainer_state.json +566 -0
- data_curation&frozen_layers/dataset1_all/F4/.DS_Store +0 -0
- data_curation&frozen_layers/dataset1_all/F4/all_results.json +8 -0
- data_curation&frozen_layers/dataset1_all/F4/config.json +27 -0
- data_curation&frozen_layers/dataset1_all/F4/eval_results.json +8 -0
- data_curation&frozen_layers/dataset1_all/F4/optimizer.pt +3 -0
- data_curation&frozen_layers/dataset1_all/F4/predictions.pickle +3 -0
- data_curation&frozen_layers/dataset1_all/F4/pytorch_model.bin +3 -0
- data_curation&frozen_layers/dataset1_all/F4/rng_state.pth +3 -0
- data_curation&frozen_layers/dataset1_all/F4/scheduler.pt +3 -0
.DS_Store
CHANGED
|
Binary files a/.DS_Store and b/.DS_Store differ
|
|
|
cross_cell_type_generization/.DS_Store
CHANGED
|
Binary files a/cross_cell_type_generization/.DS_Store and b/cross_cell_type_generization/.DS_Store differ
|
|
|
cross_cell_type_generization/L4/.DS_Store
CHANGED
|
Binary files a/cross_cell_type_generization/L4/.DS_Store and b/cross_cell_type_generization/L4/.DS_Store differ
|
|
|
cross_cell_type_generization/L4/Epitheliums/.DS_Store
CHANGED
|
Binary files a/cross_cell_type_generization/L4/Epitheliums/.DS_Store and b/cross_cell_type_generization/L4/Epitheliums/.DS_Store differ
|
|
|
cross_cell_type_generization/L4/Epitheliums/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": 4,
|
| 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/L4/Epitheliums/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7655555555555555,
|
| 3 |
+
"test_loss": 0.5546273589134216,
|
| 4 |
+
"test_macro_f1": 0.7257171579155659,
|
| 5 |
+
"test_runtime": 2.5959,
|
| 6 |
+
"test_samples_per_second": 346.703,
|
| 7 |
+
"test_steps_per_second": 3.467
|
| 8 |
+
}
|
cross_cell_type_generization/L4/Macrophages/.DS_Store
CHANGED
|
Binary files a/cross_cell_type_generization/L4/Macrophages/.DS_Store and b/cross_cell_type_generization/L4/Macrophages/.DS_Store differ
|
|
|
cross_cell_type_generization/L4/Macrophages/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.8377777777777777,
|
| 3 |
+
"test_loss": 0.3709336817264557,
|
| 4 |
+
"test_macro_f1": 0.8374237227740413,
|
| 5 |
+
"test_runtime": 2.6942,
|
| 6 |
+
"test_samples_per_second": 334.056,
|
| 7 |
+
"test_steps_per_second": 3.341
|
| 8 |
+
}
|
cross_cell_type_generization/L4/Macrophages/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": 4,
|
| 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/L4/Macrophages/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.8377777777777777,
|
| 3 |
+
"test_loss": 0.3709336817264557,
|
| 4 |
+
"test_macro_f1": 0.8374237227740413,
|
| 5 |
+
"test_runtime": 2.6942,
|
| 6 |
+
"test_samples_per_second": 334.056,
|
| 7 |
+
"test_steps_per_second": 3.341
|
| 8 |
+
}
|
cross_cell_type_generization/L4/Macrophages/trainer_state.json
ADDED
|
@@ -0,0 +1,450 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.39814653992652893,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L4/Macrophages/fold4/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": 0.6917,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6919,
|
| 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": 0.6934,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6944,
|
| 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": 0.6906,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5171275278580273,
|
| 55 |
+
"eval_loss": 0.6910191178321838,
|
| 56 |
+
"eval_macro_f1": 0.4080778330325605,
|
| 57 |
+
"eval_runtime": 14.041,
|
| 58 |
+
"eval_samples_per_second": 345.132,
|
| 59 |
+
"eval_steps_per_second": 3.347,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6912,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6911,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6875,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6873,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6844,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6814,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6795,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5870821295914156,
|
| 107 |
+
"eval_loss": 0.6697810292243958,
|
| 108 |
+
"eval_macro_f1": 0.5825487540387015,
|
| 109 |
+
"eval_runtime": 14.066,
|
| 110 |
+
"eval_samples_per_second": 344.519,
|
| 111 |
+
"eval_steps_per_second": 3.341,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6781,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6564,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6297,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6131,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5841,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5394,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.492,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4617,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7686751960379694,
|
| 165 |
+
"eval_loss": 0.46376147866249084,
|
| 166 |
+
"eval_macro_f1": 0.7685935958162224,
|
| 167 |
+
"eval_runtime": 14.296,
|
| 168 |
+
"eval_samples_per_second": 338.976,
|
| 169 |
+
"eval_steps_per_second": 3.288,
|
| 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": 0.407,
|
| 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": 0.4284,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.4237,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.3709,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3842,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7944696657036732,
|
| 217 |
+
"eval_loss": 0.4250829815864563,
|
| 218 |
+
"eval_macro_f1": 0.7944693506298842,
|
| 219 |
+
"eval_runtime": 14.2291,
|
| 220 |
+
"eval_samples_per_second": 340.57,
|
| 221 |
+
"eval_steps_per_second": 3.303,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3891,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3184,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3392,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3404,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3744,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3229,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3298,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3446,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7911679735864631,
|
| 275 |
+
"eval_loss": 0.4325283169746399,
|
| 276 |
+
"eval_macro_f1": 0.7906017246983076,
|
| 277 |
+
"eval_runtime": 13.721,
|
| 278 |
+
"eval_samples_per_second": 353.181,
|
| 279 |
+
"eval_steps_per_second": 3.425,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.319,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.266,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2992,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2987,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2906,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.3031,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2794,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.819026000825423,
|
| 327 |
+
"eval_loss": 0.39814653992652893,
|
| 328 |
+
"eval_macro_f1": 0.8178884568287126,
|
| 329 |
+
"eval_runtime": 13.8654,
|
| 330 |
+
"eval_samples_per_second": 349.504,
|
| 331 |
+
"eval_steps_per_second": 3.39,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2802,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2637,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.274,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2754,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2332,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.259,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2446,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.246,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8208832026413537,
|
| 385 |
+
"eval_loss": 0.4227023422718048,
|
| 386 |
+
"eval_macro_f1": 0.8184059652831023,
|
| 387 |
+
"eval_runtime": 13.7547,
|
| 388 |
+
"eval_samples_per_second": 352.315,
|
| 389 |
+
"eval_steps_per_second": 3.417,
|
| 390 |
+
"step": 1309
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.09,
|
| 394 |
+
"learning_rate": 4.533898305084746e-05,
|
| 395 |
+
"loss": 0.232,
|
| 396 |
+
"step": 1325
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.22,
|
| 400 |
+
"learning_rate": 4.519774011299435e-05,
|
| 401 |
+
"loss": 0.2373,
|
| 402 |
+
"step": 1350
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.35,
|
| 406 |
+
"learning_rate": 4.505649717514124e-05,
|
| 407 |
+
"loss": 0.2381,
|
| 408 |
+
"step": 1375
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.49,
|
| 412 |
+
"learning_rate": 4.491525423728814e-05,
|
| 413 |
+
"loss": 0.2607,
|
| 414 |
+
"step": 1400
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.62,
|
| 418 |
+
"learning_rate": 4.477401129943503e-05,
|
| 419 |
+
"loss": 0.221,
|
| 420 |
+
"step": 1425
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.75,
|
| 424 |
+
"learning_rate": 4.463276836158192e-05,
|
| 425 |
+
"loss": 0.2267,
|
| 426 |
+
"step": 1450
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.89,
|
| 430 |
+
"learning_rate": 4.4491525423728816e-05,
|
| 431 |
+
"loss": 0.21,
|
| 432 |
+
"step": 1475
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 8.0,
|
| 436 |
+
"eval_accuracy": 0.8012794056954189,
|
| 437 |
+
"eval_loss": 0.47252345085144043,
|
| 438 |
+
"eval_macro_f1": 0.8003815623782113,
|
| 439 |
+
"eval_runtime": 14.1303,
|
| 440 |
+
"eval_samples_per_second": 342.951,
|
| 441 |
+
"eval_steps_per_second": 3.326,
|
| 442 |
+
"step": 1496
|
| 443 |
+
}
|
| 444 |
+
],
|
| 445 |
+
"max_steps": 9350,
|
| 446 |
+
"num_train_epochs": 50,
|
| 447 |
+
"total_flos": 4145176027791360.0,
|
| 448 |
+
"trial_name": null,
|
| 449 |
+
"trial_params": null
|
| 450 |
+
}
|
cross_cell_type_generization/L4/NKT/.DS_Store
CHANGED
|
Binary files a/cross_cell_type_generization/L4/NKT/.DS_Store and b/cross_cell_type_generization/L4/NKT/.DS_Store differ
|
|
|
cross_cell_type_generization/L4/NKT/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9466666666666667,
|
| 3 |
+
"test_loss": 0.16052913665771484,
|
| 4 |
+
"test_macro_f1": 0.9411302982731554,
|
| 5 |
+
"test_runtime": 2.5904,
|
| 6 |
+
"test_samples_per_second": 347.442,
|
| 7 |
+
"test_steps_per_second": 3.474
|
| 8 |
+
}
|
cross_cell_type_generization/L4/NKT/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": 4,
|
| 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/L4/NKT/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9466666666666667,
|
| 3 |
+
"test_loss": 0.16052913665771484,
|
| 4 |
+
"test_macro_f1": 0.9411302982731554,
|
| 5 |
+
"test_runtime": 2.5904,
|
| 6 |
+
"test_samples_per_second": 347.442,
|
| 7 |
+
"test_steps_per_second": 3.474
|
| 8 |
+
}
|
cross_cell_type_generization/L4/NKT/trainer_state.json
ADDED
|
@@ -0,0 +1,450 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.39814653992652893,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L4/NKT/fold4/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": 0.6917,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6919,
|
| 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": 0.6934,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6944,
|
| 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": 0.6906,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5171275278580273,
|
| 55 |
+
"eval_loss": 0.6910191178321838,
|
| 56 |
+
"eval_macro_f1": 0.4080778330325605,
|
| 57 |
+
"eval_runtime": 13.555,
|
| 58 |
+
"eval_samples_per_second": 357.506,
|
| 59 |
+
"eval_steps_per_second": 3.467,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6912,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6911,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6875,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6873,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6844,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6814,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6795,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5870821295914156,
|
| 107 |
+
"eval_loss": 0.6697810292243958,
|
| 108 |
+
"eval_macro_f1": 0.5825487540387015,
|
| 109 |
+
"eval_runtime": 14.1178,
|
| 110 |
+
"eval_samples_per_second": 343.255,
|
| 111 |
+
"eval_steps_per_second": 3.329,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6781,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6564,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6297,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6131,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5841,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5394,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.492,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4617,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7686751960379694,
|
| 165 |
+
"eval_loss": 0.46376147866249084,
|
| 166 |
+
"eval_macro_f1": 0.7685935958162224,
|
| 167 |
+
"eval_runtime": 14.3072,
|
| 168 |
+
"eval_samples_per_second": 338.711,
|
| 169 |
+
"eval_steps_per_second": 3.285,
|
| 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": 0.407,
|
| 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": 0.4284,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.4237,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.3709,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3842,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7944696657036732,
|
| 217 |
+
"eval_loss": 0.4250829815864563,
|
| 218 |
+
"eval_macro_f1": 0.7944693506298842,
|
| 219 |
+
"eval_runtime": 14.4864,
|
| 220 |
+
"eval_samples_per_second": 334.521,
|
| 221 |
+
"eval_steps_per_second": 3.244,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3891,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3184,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3392,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3404,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3744,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3229,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3298,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3446,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7911679735864631,
|
| 275 |
+
"eval_loss": 0.4325283169746399,
|
| 276 |
+
"eval_macro_f1": 0.7906017246983076,
|
| 277 |
+
"eval_runtime": 14.5425,
|
| 278 |
+
"eval_samples_per_second": 333.229,
|
| 279 |
+
"eval_steps_per_second": 3.232,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.319,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.266,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2992,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2987,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2906,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.3031,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2794,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.819026000825423,
|
| 327 |
+
"eval_loss": 0.39814653992652893,
|
| 328 |
+
"eval_macro_f1": 0.8178884568287126,
|
| 329 |
+
"eval_runtime": 14.7665,
|
| 330 |
+
"eval_samples_per_second": 328.175,
|
| 331 |
+
"eval_steps_per_second": 3.183,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2802,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2637,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.274,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2754,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2332,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.259,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2446,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.246,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8208832026413537,
|
| 385 |
+
"eval_loss": 0.4227023422718048,
|
| 386 |
+
"eval_macro_f1": 0.8184059652831023,
|
| 387 |
+
"eval_runtime": 13.9954,
|
| 388 |
+
"eval_samples_per_second": 346.256,
|
| 389 |
+
"eval_steps_per_second": 3.358,
|
| 390 |
+
"step": 1309
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.09,
|
| 394 |
+
"learning_rate": 4.533898305084746e-05,
|
| 395 |
+
"loss": 0.232,
|
| 396 |
+
"step": 1325
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.22,
|
| 400 |
+
"learning_rate": 4.519774011299435e-05,
|
| 401 |
+
"loss": 0.2373,
|
| 402 |
+
"step": 1350
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.35,
|
| 406 |
+
"learning_rate": 4.505649717514124e-05,
|
| 407 |
+
"loss": 0.2381,
|
| 408 |
+
"step": 1375
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.49,
|
| 412 |
+
"learning_rate": 4.491525423728814e-05,
|
| 413 |
+
"loss": 0.2607,
|
| 414 |
+
"step": 1400
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.62,
|
| 418 |
+
"learning_rate": 4.477401129943503e-05,
|
| 419 |
+
"loss": 0.221,
|
| 420 |
+
"step": 1425
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.75,
|
| 424 |
+
"learning_rate": 4.463276836158192e-05,
|
| 425 |
+
"loss": 0.2267,
|
| 426 |
+
"step": 1450
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.89,
|
| 430 |
+
"learning_rate": 4.4491525423728816e-05,
|
| 431 |
+
"loss": 0.21,
|
| 432 |
+
"step": 1475
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 8.0,
|
| 436 |
+
"eval_accuracy": 0.8012794056954189,
|
| 437 |
+
"eval_loss": 0.47252345085144043,
|
| 438 |
+
"eval_macro_f1": 0.8003815623782113,
|
| 439 |
+
"eval_runtime": 14.3033,
|
| 440 |
+
"eval_samples_per_second": 338.804,
|
| 441 |
+
"eval_steps_per_second": 3.286,
|
| 442 |
+
"step": 1496
|
| 443 |
+
}
|
| 444 |
+
],
|
| 445 |
+
"max_steps": 9350,
|
| 446 |
+
"num_train_epochs": 50,
|
| 447 |
+
"total_flos": 4145176027791360.0,
|
| 448 |
+
"trial_name": null,
|
| 449 |
+
"trial_params": null
|
| 450 |
+
}
|
cross_cell_type_generization/L4/NKs/.DS_Store
CHANGED
|
Binary files a/cross_cell_type_generization/L4/NKs/.DS_Store and b/cross_cell_type_generization/L4/NKs/.DS_Store differ
|
|
|
cross_cell_type_generization/L4/NKs/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9044444444444445,
|
| 3 |
+
"test_loss": 0.24761898815631866,
|
| 4 |
+
"test_macro_f1": 0.861289328238912,
|
| 5 |
+
"test_runtime": 2.5343,
|
| 6 |
+
"test_samples_per_second": 355.126,
|
| 7 |
+
"test_steps_per_second": 3.551
|
| 8 |
+
}
|
cross_cell_type_generization/L4/NKs/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": 4,
|
| 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/L4/NKs/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9044444444444445,
|
| 3 |
+
"test_loss": 0.24761898815631866,
|
| 4 |
+
"test_macro_f1": 0.861289328238912,
|
| 5 |
+
"test_runtime": 2.5343,
|
| 6 |
+
"test_samples_per_second": 355.126,
|
| 7 |
+
"test_steps_per_second": 3.551
|
| 8 |
+
}
|
cross_cell_type_generization/L4/NKs/trainer_state.json
ADDED
|
@@ -0,0 +1,450 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.39814653992652893,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L4/NKs/fold4/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": 0.6917,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6919,
|
| 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": 0.6934,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6944,
|
| 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": 0.6906,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5171275278580273,
|
| 55 |
+
"eval_loss": 0.6910191178321838,
|
| 56 |
+
"eval_macro_f1": 0.4080778330325605,
|
| 57 |
+
"eval_runtime": 14.5066,
|
| 58 |
+
"eval_samples_per_second": 334.055,
|
| 59 |
+
"eval_steps_per_second": 3.24,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6912,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6911,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6875,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6873,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6844,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6814,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6795,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5870821295914156,
|
| 107 |
+
"eval_loss": 0.6697810292243958,
|
| 108 |
+
"eval_macro_f1": 0.5825487540387015,
|
| 109 |
+
"eval_runtime": 13.6383,
|
| 110 |
+
"eval_samples_per_second": 355.322,
|
| 111 |
+
"eval_steps_per_second": 3.446,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6781,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6564,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6297,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6131,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5841,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5394,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.492,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4617,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7686751960379694,
|
| 165 |
+
"eval_loss": 0.46376147866249084,
|
| 166 |
+
"eval_macro_f1": 0.7685935958162224,
|
| 167 |
+
"eval_runtime": 13.8492,
|
| 168 |
+
"eval_samples_per_second": 349.911,
|
| 169 |
+
"eval_steps_per_second": 3.394,
|
| 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": 0.407,
|
| 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": 0.4284,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.4237,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.3709,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3842,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7944696657036732,
|
| 217 |
+
"eval_loss": 0.4250829815864563,
|
| 218 |
+
"eval_macro_f1": 0.7944693506298842,
|
| 219 |
+
"eval_runtime": 14.4424,
|
| 220 |
+
"eval_samples_per_second": 335.539,
|
| 221 |
+
"eval_steps_per_second": 3.254,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3891,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3184,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3392,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3404,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3744,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3229,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3298,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3446,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7911679735864631,
|
| 275 |
+
"eval_loss": 0.4325283169746399,
|
| 276 |
+
"eval_macro_f1": 0.7906017246983076,
|
| 277 |
+
"eval_runtime": 13.9518,
|
| 278 |
+
"eval_samples_per_second": 347.339,
|
| 279 |
+
"eval_steps_per_second": 3.369,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.319,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.266,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2992,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2987,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2906,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.3031,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2794,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.819026000825423,
|
| 327 |
+
"eval_loss": 0.39814653992652893,
|
| 328 |
+
"eval_macro_f1": 0.8178884568287126,
|
| 329 |
+
"eval_runtime": 14.2206,
|
| 330 |
+
"eval_samples_per_second": 340.773,
|
| 331 |
+
"eval_steps_per_second": 3.305,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2802,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2637,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.274,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2754,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2332,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.259,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2446,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.246,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8208832026413537,
|
| 385 |
+
"eval_loss": 0.4227023422718048,
|
| 386 |
+
"eval_macro_f1": 0.8184059652831023,
|
| 387 |
+
"eval_runtime": 14.2381,
|
| 388 |
+
"eval_samples_per_second": 340.355,
|
| 389 |
+
"eval_steps_per_second": 3.301,
|
| 390 |
+
"step": 1309
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.09,
|
| 394 |
+
"learning_rate": 4.533898305084746e-05,
|
| 395 |
+
"loss": 0.232,
|
| 396 |
+
"step": 1325
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.22,
|
| 400 |
+
"learning_rate": 4.519774011299435e-05,
|
| 401 |
+
"loss": 0.2373,
|
| 402 |
+
"step": 1350
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.35,
|
| 406 |
+
"learning_rate": 4.505649717514124e-05,
|
| 407 |
+
"loss": 0.2381,
|
| 408 |
+
"step": 1375
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.49,
|
| 412 |
+
"learning_rate": 4.491525423728814e-05,
|
| 413 |
+
"loss": 0.2607,
|
| 414 |
+
"step": 1400
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.62,
|
| 418 |
+
"learning_rate": 4.477401129943503e-05,
|
| 419 |
+
"loss": 0.221,
|
| 420 |
+
"step": 1425
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.75,
|
| 424 |
+
"learning_rate": 4.463276836158192e-05,
|
| 425 |
+
"loss": 0.2267,
|
| 426 |
+
"step": 1450
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.89,
|
| 430 |
+
"learning_rate": 4.4491525423728816e-05,
|
| 431 |
+
"loss": 0.21,
|
| 432 |
+
"step": 1475
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 8.0,
|
| 436 |
+
"eval_accuracy": 0.8012794056954189,
|
| 437 |
+
"eval_loss": 0.47252345085144043,
|
| 438 |
+
"eval_macro_f1": 0.8003815623782113,
|
| 439 |
+
"eval_runtime": 13.6574,
|
| 440 |
+
"eval_samples_per_second": 354.827,
|
| 441 |
+
"eval_steps_per_second": 3.441,
|
| 442 |
+
"step": 1496
|
| 443 |
+
}
|
| 444 |
+
],
|
| 445 |
+
"max_steps": 9350,
|
| 446 |
+
"num_train_epochs": 50,
|
| 447 |
+
"total_flos": 4145176027791360.0,
|
| 448 |
+
"trial_name": null,
|
| 449 |
+
"trial_params": null
|
| 450 |
+
}
|
cross_cell_type_generization/L4/T cells/.DS_Store
CHANGED
|
Binary files a/cross_cell_type_generization/L4/T cells/.DS_Store and b/cross_cell_type_generization/L4/T cells/.DS_Store differ
|
|
|
cross_cell_type_generization/L4/T cells/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9255555555555556,
|
| 3 |
+
"test_loss": 0.19469526410102844,
|
| 4 |
+
"test_macro_f1": 0.9254980694980695,
|
| 5 |
+
"test_runtime": 2.8235,
|
| 6 |
+
"test_samples_per_second": 318.748,
|
| 7 |
+
"test_steps_per_second": 3.187
|
| 8 |
+
}
|
cross_cell_type_generization/L4/T cells/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": 4,
|
| 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/L4/T cells/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.9255555555555556,
|
| 3 |
+
"test_loss": 0.19469526410102844,
|
| 4 |
+
"test_macro_f1": 0.9254980694980695,
|
| 5 |
+
"test_runtime": 2.8235,
|
| 6 |
+
"test_samples_per_second": 318.748,
|
| 7 |
+
"test_steps_per_second": 3.187
|
| 8 |
+
}
|
cross_cell_type_generization/L4/T cells/trainer_state.json
ADDED
|
@@ -0,0 +1,450 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.39814653992652893,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_diversity/L4/T cells/fold4/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": 0.6917,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 5e-06,
|
| 19 |
+
"loss": 0.6919,
|
| 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": 0.6934,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.67,
|
| 36 |
+
"learning_rate": 1.25e-05,
|
| 37 |
+
"loss": 0.6944,
|
| 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": 0.6906,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.5171275278580273,
|
| 55 |
+
"eval_loss": 0.6910191178321838,
|
| 56 |
+
"eval_macro_f1": 0.4080778330325605,
|
| 57 |
+
"eval_runtime": 14.0812,
|
| 58 |
+
"eval_samples_per_second": 344.146,
|
| 59 |
+
"eval_steps_per_second": 3.338,
|
| 60 |
+
"step": 187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.07,
|
| 64 |
+
"learning_rate": 2e-05,
|
| 65 |
+
"loss": 0.6912,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.2,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 0.6911,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.34,
|
| 76 |
+
"learning_rate": 2.5e-05,
|
| 77 |
+
"loss": 0.6875,
|
| 78 |
+
"step": 250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.47,
|
| 82 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 83 |
+
"loss": 0.6873,
|
| 84 |
+
"step": 275
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.6,
|
| 88 |
+
"learning_rate": 3e-05,
|
| 89 |
+
"loss": 0.6844,
|
| 90 |
+
"step": 300
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.74,
|
| 94 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 95 |
+
"loss": 0.6814,
|
| 96 |
+
"step": 325
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.87,
|
| 100 |
+
"learning_rate": 3.5e-05,
|
| 101 |
+
"loss": 0.6795,
|
| 102 |
+
"step": 350
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.0,
|
| 106 |
+
"eval_accuracy": 0.5870821295914156,
|
| 107 |
+
"eval_loss": 0.6697810292243958,
|
| 108 |
+
"eval_macro_f1": 0.5825487540387015,
|
| 109 |
+
"eval_runtime": 13.5633,
|
| 110 |
+
"eval_samples_per_second": 357.289,
|
| 111 |
+
"eval_steps_per_second": 3.465,
|
| 112 |
+
"step": 374
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 2.01,
|
| 116 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 117 |
+
"loss": 0.6781,
|
| 118 |
+
"step": 375
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.14,
|
| 122 |
+
"learning_rate": 4e-05,
|
| 123 |
+
"loss": 0.6564,
|
| 124 |
+
"step": 400
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.27,
|
| 128 |
+
"learning_rate": 4.25e-05,
|
| 129 |
+
"loss": 0.6297,
|
| 130 |
+
"step": 425
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.41,
|
| 134 |
+
"learning_rate": 4.5e-05,
|
| 135 |
+
"loss": 0.6131,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.54,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 0.5841,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.67,
|
| 146 |
+
"learning_rate": 5e-05,
|
| 147 |
+
"loss": 0.5394,
|
| 148 |
+
"step": 500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.81,
|
| 152 |
+
"learning_rate": 4.9858757062146896e-05,
|
| 153 |
+
"loss": 0.492,
|
| 154 |
+
"step": 525
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.94,
|
| 158 |
+
"learning_rate": 4.971751412429379e-05,
|
| 159 |
+
"loss": 0.4617,
|
| 160 |
+
"step": 550
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7686751960379694,
|
| 165 |
+
"eval_loss": 0.46376147866249084,
|
| 166 |
+
"eval_macro_f1": 0.7685935958162224,
|
| 167 |
+
"eval_runtime": 14.5761,
|
| 168 |
+
"eval_samples_per_second": 332.462,
|
| 169 |
+
"eval_steps_per_second": 3.224,
|
| 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": 0.407,
|
| 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": 0.4284,
|
| 194 |
+
"step": 650
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.61,
|
| 198 |
+
"learning_rate": 4.9011299435028255e-05,
|
| 199 |
+
"loss": 0.4237,
|
| 200 |
+
"step": 675
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.74,
|
| 204 |
+
"learning_rate": 4.887005649717514e-05,
|
| 205 |
+
"loss": 0.3709,
|
| 206 |
+
"step": 700
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.88,
|
| 210 |
+
"learning_rate": 4.8728813559322034e-05,
|
| 211 |
+
"loss": 0.3842,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.0,
|
| 216 |
+
"eval_accuracy": 0.7944696657036732,
|
| 217 |
+
"eval_loss": 0.4250829815864563,
|
| 218 |
+
"eval_macro_f1": 0.7944693506298842,
|
| 219 |
+
"eval_runtime": 14.4363,
|
| 220 |
+
"eval_samples_per_second": 335.683,
|
| 221 |
+
"eval_steps_per_second": 3.256,
|
| 222 |
+
"step": 748
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 4.01,
|
| 226 |
+
"learning_rate": 4.8587570621468934e-05,
|
| 227 |
+
"loss": 0.3891,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.14,
|
| 232 |
+
"learning_rate": 4.844632768361582e-05,
|
| 233 |
+
"loss": 0.3184,
|
| 234 |
+
"step": 775
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.28,
|
| 238 |
+
"learning_rate": 4.8305084745762714e-05,
|
| 239 |
+
"loss": 0.3392,
|
| 240 |
+
"step": 800
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.41,
|
| 244 |
+
"learning_rate": 4.816384180790961e-05,
|
| 245 |
+
"loss": 0.3404,
|
| 246 |
+
"step": 825
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.55,
|
| 250 |
+
"learning_rate": 4.80225988700565e-05,
|
| 251 |
+
"loss": 0.3744,
|
| 252 |
+
"step": 850
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.68,
|
| 256 |
+
"learning_rate": 4.788135593220339e-05,
|
| 257 |
+
"loss": 0.3229,
|
| 258 |
+
"step": 875
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.81,
|
| 262 |
+
"learning_rate": 4.7740112994350286e-05,
|
| 263 |
+
"loss": 0.3298,
|
| 264 |
+
"step": 900
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.95,
|
| 268 |
+
"learning_rate": 4.759887005649718e-05,
|
| 269 |
+
"loss": 0.3446,
|
| 270 |
+
"step": 925
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7911679735864631,
|
| 275 |
+
"eval_loss": 0.4325283169746399,
|
| 276 |
+
"eval_macro_f1": 0.7906017246983076,
|
| 277 |
+
"eval_runtime": 14.0025,
|
| 278 |
+
"eval_samples_per_second": 346.08,
|
| 279 |
+
"eval_steps_per_second": 3.357,
|
| 280 |
+
"step": 935
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.08,
|
| 284 |
+
"learning_rate": 4.745762711864407e-05,
|
| 285 |
+
"loss": 0.319,
|
| 286 |
+
"step": 950
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.21,
|
| 290 |
+
"learning_rate": 4.7316384180790966e-05,
|
| 291 |
+
"loss": 0.266,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.35,
|
| 296 |
+
"learning_rate": 4.717514124293785e-05,
|
| 297 |
+
"loss": 0.2992,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.48,
|
| 302 |
+
"learning_rate": 4.703389830508475e-05,
|
| 303 |
+
"loss": 0.2987,
|
| 304 |
+
"step": 1025
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.61,
|
| 308 |
+
"learning_rate": 4.689265536723164e-05,
|
| 309 |
+
"loss": 0.2906,
|
| 310 |
+
"step": 1050
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.75,
|
| 314 |
+
"learning_rate": 4.675141242937853e-05,
|
| 315 |
+
"loss": 0.3031,
|
| 316 |
+
"step": 1075
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.88,
|
| 320 |
+
"learning_rate": 4.6610169491525425e-05,
|
| 321 |
+
"loss": 0.2794,
|
| 322 |
+
"step": 1100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 6.0,
|
| 326 |
+
"eval_accuracy": 0.819026000825423,
|
| 327 |
+
"eval_loss": 0.39814653992652893,
|
| 328 |
+
"eval_macro_f1": 0.8178884568287126,
|
| 329 |
+
"eval_runtime": 14.1681,
|
| 330 |
+
"eval_samples_per_second": 342.036,
|
| 331 |
+
"eval_steps_per_second": 3.317,
|
| 332 |
+
"step": 1122
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.02,
|
| 336 |
+
"learning_rate": 4.646892655367232e-05,
|
| 337 |
+
"loss": 0.2802,
|
| 338 |
+
"step": 1125
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.15,
|
| 342 |
+
"learning_rate": 4.632768361581921e-05,
|
| 343 |
+
"loss": 0.2637,
|
| 344 |
+
"step": 1150
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.28,
|
| 348 |
+
"learning_rate": 4.6186440677966104e-05,
|
| 349 |
+
"loss": 0.274,
|
| 350 |
+
"step": 1175
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.42,
|
| 354 |
+
"learning_rate": 4.6045197740113e-05,
|
| 355 |
+
"loss": 0.2754,
|
| 356 |
+
"step": 1200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.55,
|
| 360 |
+
"learning_rate": 4.590395480225989e-05,
|
| 361 |
+
"loss": 0.2332,
|
| 362 |
+
"step": 1225
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.68,
|
| 366 |
+
"learning_rate": 4.5762711864406784e-05,
|
| 367 |
+
"loss": 0.259,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.82,
|
| 372 |
+
"learning_rate": 4.562146892655367e-05,
|
| 373 |
+
"loss": 0.2446,
|
| 374 |
+
"step": 1275
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.95,
|
| 378 |
+
"learning_rate": 4.548022598870056e-05,
|
| 379 |
+
"loss": 0.246,
|
| 380 |
+
"step": 1300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.8208832026413537,
|
| 385 |
+
"eval_loss": 0.4227023422718048,
|
| 386 |
+
"eval_macro_f1": 0.8184059652831023,
|
| 387 |
+
"eval_runtime": 14.6282,
|
| 388 |
+
"eval_samples_per_second": 331.279,
|
| 389 |
+
"eval_steps_per_second": 3.213,
|
| 390 |
+
"step": 1309
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.09,
|
| 394 |
+
"learning_rate": 4.533898305084746e-05,
|
| 395 |
+
"loss": 0.232,
|
| 396 |
+
"step": 1325
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.22,
|
| 400 |
+
"learning_rate": 4.519774011299435e-05,
|
| 401 |
+
"loss": 0.2373,
|
| 402 |
+
"step": 1350
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.35,
|
| 406 |
+
"learning_rate": 4.505649717514124e-05,
|
| 407 |
+
"loss": 0.2381,
|
| 408 |
+
"step": 1375
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.49,
|
| 412 |
+
"learning_rate": 4.491525423728814e-05,
|
| 413 |
+
"loss": 0.2607,
|
| 414 |
+
"step": 1400
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.62,
|
| 418 |
+
"learning_rate": 4.477401129943503e-05,
|
| 419 |
+
"loss": 0.221,
|
| 420 |
+
"step": 1425
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.75,
|
| 424 |
+
"learning_rate": 4.463276836158192e-05,
|
| 425 |
+
"loss": 0.2267,
|
| 426 |
+
"step": 1450
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.89,
|
| 430 |
+
"learning_rate": 4.4491525423728816e-05,
|
| 431 |
+
"loss": 0.21,
|
| 432 |
+
"step": 1475
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 8.0,
|
| 436 |
+
"eval_accuracy": 0.8012794056954189,
|
| 437 |
+
"eval_loss": 0.47252345085144043,
|
| 438 |
+
"eval_macro_f1": 0.8003815623782113,
|
| 439 |
+
"eval_runtime": 14.1684,
|
| 440 |
+
"eval_samples_per_second": 342.028,
|
| 441 |
+
"eval_steps_per_second": 3.317,
|
| 442 |
+
"step": 1496
|
| 443 |
+
}
|
| 444 |
+
],
|
| 445 |
+
"max_steps": 9350,
|
| 446 |
+
"num_train_epochs": 50,
|
| 447 |
+
"total_flos": 4145176027791360.0,
|
| 448 |
+
"trial_name": null,
|
| 449 |
+
"trial_params": null
|
| 450 |
+
}
|
data_curation&frozen_layers/.DS_Store
CHANGED
|
Binary files a/data_curation&frozen_layers/.DS_Store and b/data_curation&frozen_layers/.DS_Store differ
|
|
|
data_curation&frozen_layers/dataset1_all/.DS_Store
CHANGED
|
Binary files a/data_curation&frozen_layers/dataset1_all/.DS_Store and b/data_curation&frozen_layers/dataset1_all/.DS_Store differ
|
|
|
data_curation&frozen_layers/dataset1_all/F0/.DS_Store
CHANGED
|
Binary files a/data_curation&frozen_layers/dataset1_all/F0/.DS_Store and b/data_curation&frozen_layers/dataset1_all/F0/.DS_Store differ
|
|
|
data_curation&frozen_layers/dataset1_all/F0/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7918886593454387,
|
| 3 |
+
"test_loss": 0.4178936779499054,
|
| 4 |
+
"test_macro_f1": 0.7900743751766766,
|
| 5 |
+
"test_runtime": 28.8947,
|
| 6 |
+
"test_samples_per_second": 318.294,
|
| 7 |
+
"test_steps_per_second": 3.08
|
| 8 |
+
}
|
data_curation&frozen_layers/dataset1_all/F0/config.json
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/vsphhome/fengguoqing/Geneformer",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"BertForSequenceClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.02,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"gradient_checkpointing": false,
|
| 9 |
+
"hidden_act": "relu",
|
| 10 |
+
"hidden_dropout_prob": 0.02,
|
| 11 |
+
"hidden_size": 256,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 512,
|
| 14 |
+
"layer_norm_eps": 1e-12,
|
| 15 |
+
"max_position_embeddings": 2048,
|
| 16 |
+
"model_type": "bert",
|
| 17 |
+
"num_attention_heads": 4,
|
| 18 |
+
"num_hidden_layers": 6,
|
| 19 |
+
"pad_token_id": 0,
|
| 20 |
+
"position_embedding_type": "absolute",
|
| 21 |
+
"problem_type": "single_label_classification",
|
| 22 |
+
"torch_dtype": "float32",
|
| 23 |
+
"transformers_version": "4.28.0",
|
| 24 |
+
"type_vocab_size": 2,
|
| 25 |
+
"use_cache": true,
|
| 26 |
+
"vocab_size": 25426
|
| 27 |
+
}
|
data_curation&frozen_layers/dataset1_all/F0/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7918886593454387,
|
| 3 |
+
"test_loss": 0.4178936779499054,
|
| 4 |
+
"test_macro_f1": 0.7900743751766766,
|
| 5 |
+
"test_runtime": 28.8947,
|
| 6 |
+
"test_samples_per_second": 318.294,
|
| 7 |
+
"test_steps_per_second": 3.08
|
| 8 |
+
}
|
data_curation&frozen_layers/dataset1_all/F0/trainer_state.json
ADDED
|
@@ -0,0 +1,398 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.4178936779499054,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_curation/240716_geneformer_CellClassifier_PM25_dataset1_L2048_B26_LR5e-05_LSlinear_WU500_E50_Oadamw_F0_fold4/checkpoint-1770",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"global_step": 2478,
|
| 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": 4.7e-06,
|
| 13 |
+
"loss": 0.699,
|
| 14 |
+
"step": 47
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 9.4e-06,
|
| 19 |
+
"loss": 0.6968,
|
| 20 |
+
"step": 94
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 1.4099999999999999e-05,
|
| 25 |
+
"loss": 0.6872,
|
| 26 |
+
"step": 141
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1.88e-05,
|
| 31 |
+
"loss": 0.6734,
|
| 32 |
+
"step": 188
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.66,
|
| 36 |
+
"learning_rate": 2.35e-05,
|
| 37 |
+
"loss": 0.6567,
|
| 38 |
+
"step": 235
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 2.8199999999999998e-05,
|
| 43 |
+
"loss": 0.6318,
|
| 44 |
+
"step": 282
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.93,
|
| 48 |
+
"learning_rate": 3.29e-05,
|
| 49 |
+
"loss": 0.6259,
|
| 50 |
+
"step": 329
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.6278134174187235,
|
| 55 |
+
"eval_loss": 0.6198515295982361,
|
| 56 |
+
"eval_macro_f1": 0.6250609592590812,
|
| 57 |
+
"eval_runtime": 29.085,
|
| 58 |
+
"eval_samples_per_second": 316.211,
|
| 59 |
+
"eval_steps_per_second": 3.06,
|
| 60 |
+
"step": 354
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.06,
|
| 64 |
+
"learning_rate": 3.76e-05,
|
| 65 |
+
"loss": 0.6026,
|
| 66 |
+
"step": 376
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.19,
|
| 70 |
+
"learning_rate": 4.23e-05,
|
| 71 |
+
"loss": 0.5907,
|
| 72 |
+
"step": 423
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.33,
|
| 76 |
+
"learning_rate": 4.7e-05,
|
| 77 |
+
"loss": 0.6026,
|
| 78 |
+
"step": 470
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.46,
|
| 82 |
+
"learning_rate": 4.995058139534884e-05,
|
| 83 |
+
"loss": 0.5725,
|
| 84 |
+
"step": 517
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.59,
|
| 88 |
+
"learning_rate": 4.981395348837209e-05,
|
| 89 |
+
"loss": 0.5673,
|
| 90 |
+
"step": 564
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.73,
|
| 94 |
+
"learning_rate": 4.967732558139535e-05,
|
| 95 |
+
"loss": 0.5563,
|
| 96 |
+
"step": 611
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.86,
|
| 100 |
+
"learning_rate": 4.954069767441861e-05,
|
| 101 |
+
"loss": 0.554,
|
| 102 |
+
"step": 658
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 1.99,
|
| 106 |
+
"learning_rate": 4.940406976744186e-05,
|
| 107 |
+
"loss": 0.5169,
|
| 108 |
+
"step": 705
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 2.0,
|
| 112 |
+
"eval_accuracy": 0.7059910840491465,
|
| 113 |
+
"eval_loss": 0.5674108266830444,
|
| 114 |
+
"eval_macro_f1": 0.704985311360717,
|
| 115 |
+
"eval_runtime": 28.6864,
|
| 116 |
+
"eval_samples_per_second": 320.605,
|
| 117 |
+
"eval_steps_per_second": 3.103,
|
| 118 |
+
"step": 708
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.12,
|
| 122 |
+
"learning_rate": 4.926744186046512e-05,
|
| 123 |
+
"loss": 0.544,
|
| 124 |
+
"step": 752
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.26,
|
| 128 |
+
"learning_rate": 4.9130813953488376e-05,
|
| 129 |
+
"loss": 0.5296,
|
| 130 |
+
"step": 799
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.39,
|
| 134 |
+
"learning_rate": 4.899418604651163e-05,
|
| 135 |
+
"loss": 0.5126,
|
| 136 |
+
"step": 846
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.52,
|
| 140 |
+
"learning_rate": 4.8857558139534885e-05,
|
| 141 |
+
"loss": 0.5277,
|
| 142 |
+
"step": 893
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.66,
|
| 146 |
+
"learning_rate": 4.8720930232558146e-05,
|
| 147 |
+
"loss": 0.4896,
|
| 148 |
+
"step": 940
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.79,
|
| 152 |
+
"learning_rate": 4.85843023255814e-05,
|
| 153 |
+
"loss": 0.4887,
|
| 154 |
+
"step": 987
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.92,
|
| 158 |
+
"learning_rate": 4.8447674418604654e-05,
|
| 159 |
+
"loss": 0.4916,
|
| 160 |
+
"step": 1034
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7490486028052625,
|
| 165 |
+
"eval_loss": 0.47779449820518494,
|
| 166 |
+
"eval_macro_f1": 0.7485365926009326,
|
| 167 |
+
"eval_runtime": 29.287,
|
| 168 |
+
"eval_samples_per_second": 314.03,
|
| 169 |
+
"eval_steps_per_second": 3.039,
|
| 170 |
+
"step": 1062
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.05,
|
| 174 |
+
"learning_rate": 4.831104651162791e-05,
|
| 175 |
+
"loss": 0.4725,
|
| 176 |
+
"step": 1081
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.19,
|
| 180 |
+
"learning_rate": 4.817441860465117e-05,
|
| 181 |
+
"loss": 0.457,
|
| 182 |
+
"step": 1128
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.32,
|
| 186 |
+
"learning_rate": 4.8037790697674424e-05,
|
| 187 |
+
"loss": 0.4775,
|
| 188 |
+
"step": 1175
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.45,
|
| 192 |
+
"learning_rate": 4.790116279069768e-05,
|
| 193 |
+
"loss": 0.4532,
|
| 194 |
+
"step": 1222
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.58,
|
| 198 |
+
"learning_rate": 4.776453488372093e-05,
|
| 199 |
+
"loss": 0.4651,
|
| 200 |
+
"step": 1269
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.72,
|
| 204 |
+
"learning_rate": 4.762790697674419e-05,
|
| 205 |
+
"loss": 0.4592,
|
| 206 |
+
"step": 1316
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.85,
|
| 210 |
+
"learning_rate": 4.749127906976744e-05,
|
| 211 |
+
"loss": 0.4466,
|
| 212 |
+
"step": 1363
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 3.98,
|
| 216 |
+
"learning_rate": 4.7354651162790695e-05,
|
| 217 |
+
"loss": 0.4476,
|
| 218 |
+
"step": 1410
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 4.0,
|
| 222 |
+
"eval_accuracy": 0.7726432532347505,
|
| 223 |
+
"eval_loss": 0.4460304081439972,
|
| 224 |
+
"eval_macro_f1": 0.7668150187606645,
|
| 225 |
+
"eval_runtime": 28.5449,
|
| 226 |
+
"eval_samples_per_second": 322.194,
|
| 227 |
+
"eval_steps_per_second": 3.118,
|
| 228 |
+
"step": 1416
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.12,
|
| 232 |
+
"learning_rate": 4.7218023255813956e-05,
|
| 233 |
+
"loss": 0.4375,
|
| 234 |
+
"step": 1457
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.25,
|
| 238 |
+
"learning_rate": 4.708139534883721e-05,
|
| 239 |
+
"loss": 0.4181,
|
| 240 |
+
"step": 1504
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.38,
|
| 244 |
+
"learning_rate": 4.6944767441860464e-05,
|
| 245 |
+
"loss": 0.408,
|
| 246 |
+
"step": 1551
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.51,
|
| 250 |
+
"learning_rate": 4.680813953488372e-05,
|
| 251 |
+
"loss": 0.4046,
|
| 252 |
+
"step": 1598
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.65,
|
| 256 |
+
"learning_rate": 4.667151162790698e-05,
|
| 257 |
+
"loss": 0.4175,
|
| 258 |
+
"step": 1645
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.78,
|
| 262 |
+
"learning_rate": 4.6534883720930234e-05,
|
| 263 |
+
"loss": 0.4072,
|
| 264 |
+
"step": 1692
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.91,
|
| 268 |
+
"learning_rate": 4.639825581395349e-05,
|
| 269 |
+
"loss": 0.3991,
|
| 270 |
+
"step": 1739
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7918886593454387,
|
| 275 |
+
"eval_loss": 0.4178936779499054,
|
| 276 |
+
"eval_macro_f1": 0.7900743751766766,
|
| 277 |
+
"eval_runtime": 29.1563,
|
| 278 |
+
"eval_samples_per_second": 315.438,
|
| 279 |
+
"eval_steps_per_second": 3.053,
|
| 280 |
+
"step": 1770
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.05,
|
| 284 |
+
"learning_rate": 4.626162790697675e-05,
|
| 285 |
+
"loss": 0.3955,
|
| 286 |
+
"step": 1786
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.18,
|
| 290 |
+
"learning_rate": 4.6125e-05,
|
| 291 |
+
"loss": 0.3857,
|
| 292 |
+
"step": 1833
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.31,
|
| 296 |
+
"learning_rate": 4.598837209302326e-05,
|
| 297 |
+
"loss": 0.3799,
|
| 298 |
+
"step": 1880
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.44,
|
| 302 |
+
"learning_rate": 4.585174418604651e-05,
|
| 303 |
+
"loss": 0.3798,
|
| 304 |
+
"step": 1927
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.58,
|
| 308 |
+
"learning_rate": 4.571511627906977e-05,
|
| 309 |
+
"loss": 0.3919,
|
| 310 |
+
"step": 1974
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.71,
|
| 314 |
+
"learning_rate": 4.557848837209303e-05,
|
| 315 |
+
"loss": 0.3751,
|
| 316 |
+
"step": 2021
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.84,
|
| 320 |
+
"learning_rate": 4.544186046511628e-05,
|
| 321 |
+
"loss": 0.4059,
|
| 322 |
+
"step": 2068
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 5.97,
|
| 326 |
+
"learning_rate": 4.5305232558139535e-05,
|
| 327 |
+
"loss": 0.3856,
|
| 328 |
+
"step": 2115
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 6.0,
|
| 332 |
+
"eval_accuracy": 0.7924323148852886,
|
| 333 |
+
"eval_loss": 0.4189501702785492,
|
| 334 |
+
"eval_macro_f1": 0.791898051259053,
|
| 335 |
+
"eval_runtime": 28.6324,
|
| 336 |
+
"eval_samples_per_second": 321.209,
|
| 337 |
+
"eval_steps_per_second": 3.108,
|
| 338 |
+
"step": 2124
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.11,
|
| 342 |
+
"learning_rate": 4.5168604651162796e-05,
|
| 343 |
+
"loss": 0.3676,
|
| 344 |
+
"step": 2162
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.24,
|
| 348 |
+
"learning_rate": 4.503197674418605e-05,
|
| 349 |
+
"loss": 0.3699,
|
| 350 |
+
"step": 2209
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.37,
|
| 354 |
+
"learning_rate": 4.4895348837209305e-05,
|
| 355 |
+
"loss": 0.3526,
|
| 356 |
+
"step": 2256
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.51,
|
| 360 |
+
"learning_rate": 4.475872093023256e-05,
|
| 361 |
+
"loss": 0.3744,
|
| 362 |
+
"step": 2303
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.64,
|
| 366 |
+
"learning_rate": 4.462209302325582e-05,
|
| 367 |
+
"loss": 0.3517,
|
| 368 |
+
"step": 2350
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.77,
|
| 372 |
+
"learning_rate": 4.4485465116279074e-05,
|
| 373 |
+
"loss": 0.37,
|
| 374 |
+
"step": 2397
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.9,
|
| 378 |
+
"learning_rate": 4.434883720930233e-05,
|
| 379 |
+
"loss": 0.3793,
|
| 380 |
+
"step": 2444
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.7862346417309992,
|
| 385 |
+
"eval_loss": 0.4403926134109497,
|
| 386 |
+
"eval_macro_f1": 0.7775918523788579,
|
| 387 |
+
"eval_runtime": 28.523,
|
| 388 |
+
"eval_samples_per_second": 322.441,
|
| 389 |
+
"eval_steps_per_second": 3.12,
|
| 390 |
+
"step": 2478
|
| 391 |
+
}
|
| 392 |
+
],
|
| 393 |
+
"max_steps": 17700,
|
| 394 |
+
"num_train_epochs": 50,
|
| 395 |
+
"total_flos": 1.021966170021888e+16,
|
| 396 |
+
"trial_name": null,
|
| 397 |
+
"trial_params": null
|
| 398 |
+
}
|
data_curation&frozen_layers/dataset1_all/F2/.DS_Store
CHANGED
|
Binary files a/data_curation&frozen_layers/dataset1_all/F2/.DS_Store and b/data_curation&frozen_layers/dataset1_all/F2/.DS_Store differ
|
|
|
data_curation&frozen_layers/dataset1_all/F2/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7876481461346091,
|
| 3 |
+
"test_loss": 0.42375245690345764,
|
| 4 |
+
"test_macro_f1": 0.7860128082271005,
|
| 5 |
+
"test_runtime": 28.2868,
|
| 6 |
+
"test_samples_per_second": 325.133,
|
| 7 |
+
"test_steps_per_second": 3.146
|
| 8 |
+
}
|
data_curation&frozen_layers/dataset1_all/F2/config.json
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/vsphhome/fengguoqing/Geneformer",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"BertForSequenceClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.02,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"gradient_checkpointing": false,
|
| 9 |
+
"hidden_act": "relu",
|
| 10 |
+
"hidden_dropout_prob": 0.02,
|
| 11 |
+
"hidden_size": 256,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 512,
|
| 14 |
+
"layer_norm_eps": 1e-12,
|
| 15 |
+
"max_position_embeddings": 2048,
|
| 16 |
+
"model_type": "bert",
|
| 17 |
+
"num_attention_heads": 4,
|
| 18 |
+
"num_hidden_layers": 6,
|
| 19 |
+
"pad_token_id": 0,
|
| 20 |
+
"position_embedding_type": "absolute",
|
| 21 |
+
"problem_type": "single_label_classification",
|
| 22 |
+
"torch_dtype": "float32",
|
| 23 |
+
"transformers_version": "4.28.0",
|
| 24 |
+
"type_vocab_size": 2,
|
| 25 |
+
"use_cache": true,
|
| 26 |
+
"vocab_size": 25426
|
| 27 |
+
}
|
data_curation&frozen_layers/dataset1_all/F2/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7876481461346091,
|
| 3 |
+
"test_loss": 0.42375245690345764,
|
| 4 |
+
"test_macro_f1": 0.7860128082271005,
|
| 5 |
+
"test_runtime": 28.2868,
|
| 6 |
+
"test_samples_per_second": 325.133,
|
| 7 |
+
"test_steps_per_second": 3.146
|
| 8 |
+
}
|
data_curation&frozen_layers/dataset1_all/F2/predictions.pickle
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:80418474e74868149cc0013cb607167bdbe4bf88c016ca8d6ca430a8be4599be
|
| 3 |
+
size 147612
|
data_curation&frozen_layers/dataset1_all/F2/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b59f0bd49e567d300464159bff0c123e0631c148c751ff5b6fae76ae15bc72bd
|
| 3 |
+
size 15006
|
data_curation&frozen_layers/dataset1_all/F2/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fba7ccc8bcbde81186750973c8893ef9d6aa9bf52befe50f650c061df92088a1
|
| 3 |
+
size 1064
|
data_curation&frozen_layers/dataset1_all/F2/trainer_state.json
ADDED
|
@@ -0,0 +1,566 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.42375245690345764,
|
| 3 |
+
"best_model_checkpoint": "/vsphhome/fengguoqing/Geneformer/models/data_curation/240716_geneformer_CellClassifier_PM25_dataset1_L2048_B26_LR5e-05_LSlinear_WU500_E50_Oadamw_F2_fold4/checkpoint-2832",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 3540,
|
| 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": 4.7e-06,
|
| 13 |
+
"loss": 0.6937,
|
| 14 |
+
"step": 47
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"learning_rate": 9.4e-06,
|
| 19 |
+
"loss": 0.69,
|
| 20 |
+
"step": 94
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.4,
|
| 24 |
+
"learning_rate": 1.4099999999999999e-05,
|
| 25 |
+
"loss": 0.6825,
|
| 26 |
+
"step": 141
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.53,
|
| 30 |
+
"learning_rate": 1.88e-05,
|
| 31 |
+
"loss": 0.6712,
|
| 32 |
+
"step": 188
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.66,
|
| 36 |
+
"learning_rate": 2.35e-05,
|
| 37 |
+
"loss": 0.6597,
|
| 38 |
+
"step": 235
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.8,
|
| 42 |
+
"learning_rate": 2.8199999999999998e-05,
|
| 43 |
+
"loss": 0.635,
|
| 44 |
+
"step": 282
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.93,
|
| 48 |
+
"learning_rate": 3.29e-05,
|
| 49 |
+
"loss": 0.6164,
|
| 50 |
+
"step": 329
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.0,
|
| 54 |
+
"eval_accuracy": 0.6319451995215831,
|
| 55 |
+
"eval_loss": 0.6155699491500854,
|
| 56 |
+
"eval_macro_f1": 0.6302866279772024,
|
| 57 |
+
"eval_runtime": 28.4084,
|
| 58 |
+
"eval_samples_per_second": 323.742,
|
| 59 |
+
"eval_steps_per_second": 3.133,
|
| 60 |
+
"step": 354
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 1.06,
|
| 64 |
+
"learning_rate": 3.76e-05,
|
| 65 |
+
"loss": 0.5959,
|
| 66 |
+
"step": 376
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.19,
|
| 70 |
+
"learning_rate": 4.23e-05,
|
| 71 |
+
"loss": 0.5948,
|
| 72 |
+
"step": 423
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.33,
|
| 76 |
+
"learning_rate": 4.7e-05,
|
| 77 |
+
"loss": 0.6044,
|
| 78 |
+
"step": 470
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.46,
|
| 82 |
+
"learning_rate": 4.995058139534884e-05,
|
| 83 |
+
"loss": 0.5994,
|
| 84 |
+
"step": 517
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.59,
|
| 88 |
+
"learning_rate": 4.981395348837209e-05,
|
| 89 |
+
"loss": 0.5673,
|
| 90 |
+
"step": 564
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.73,
|
| 94 |
+
"learning_rate": 4.967732558139535e-05,
|
| 95 |
+
"loss": 0.5619,
|
| 96 |
+
"step": 611
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.86,
|
| 100 |
+
"learning_rate": 4.954069767441861e-05,
|
| 101 |
+
"loss": 0.5816,
|
| 102 |
+
"step": 658
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 1.99,
|
| 106 |
+
"learning_rate": 4.940406976744186e-05,
|
| 107 |
+
"loss": 0.5384,
|
| 108 |
+
"step": 705
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 2.0,
|
| 112 |
+
"eval_accuracy": 0.7233880613243449,
|
| 113 |
+
"eval_loss": 0.5276534557342529,
|
| 114 |
+
"eval_macro_f1": 0.7229937346384986,
|
| 115 |
+
"eval_runtime": 28.9502,
|
| 116 |
+
"eval_samples_per_second": 317.684,
|
| 117 |
+
"eval_steps_per_second": 3.074,
|
| 118 |
+
"step": 708
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 2.12,
|
| 122 |
+
"learning_rate": 4.926744186046512e-05,
|
| 123 |
+
"loss": 0.5601,
|
| 124 |
+
"step": 752
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 2.26,
|
| 128 |
+
"learning_rate": 4.9130813953488376e-05,
|
| 129 |
+
"loss": 0.5334,
|
| 130 |
+
"step": 799
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 2.39,
|
| 134 |
+
"learning_rate": 4.899418604651163e-05,
|
| 135 |
+
"loss": 0.5024,
|
| 136 |
+
"step": 846
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.52,
|
| 140 |
+
"learning_rate": 4.8857558139534885e-05,
|
| 141 |
+
"loss": 0.5114,
|
| 142 |
+
"step": 893
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.66,
|
| 146 |
+
"learning_rate": 4.8720930232558146e-05,
|
| 147 |
+
"loss": 0.4943,
|
| 148 |
+
"step": 940
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 2.79,
|
| 152 |
+
"learning_rate": 4.85843023255814e-05,
|
| 153 |
+
"loss": 0.4806,
|
| 154 |
+
"step": 987
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.92,
|
| 158 |
+
"learning_rate": 4.8447674418604654e-05,
|
| 159 |
+
"loss": 0.4762,
|
| 160 |
+
"step": 1034
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.0,
|
| 164 |
+
"eval_accuracy": 0.7345873654452539,
|
| 165 |
+
"eval_loss": 0.49307793378829956,
|
| 166 |
+
"eval_macro_f1": 0.7239669430836269,
|
| 167 |
+
"eval_runtime": 29.4659,
|
| 168 |
+
"eval_samples_per_second": 312.123,
|
| 169 |
+
"eval_steps_per_second": 3.02,
|
| 170 |
+
"step": 1062
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.05,
|
| 174 |
+
"learning_rate": 4.831104651162791e-05,
|
| 175 |
+
"loss": 0.482,
|
| 176 |
+
"step": 1081
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.19,
|
| 180 |
+
"learning_rate": 4.817441860465117e-05,
|
| 181 |
+
"loss": 0.4553,
|
| 182 |
+
"step": 1128
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.32,
|
| 186 |
+
"learning_rate": 4.8037790697674424e-05,
|
| 187 |
+
"loss": 0.4866,
|
| 188 |
+
"step": 1175
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.45,
|
| 192 |
+
"learning_rate": 4.790116279069768e-05,
|
| 193 |
+
"loss": 0.4729,
|
| 194 |
+
"step": 1222
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.58,
|
| 198 |
+
"learning_rate": 4.776453488372093e-05,
|
| 199 |
+
"loss": 0.4637,
|
| 200 |
+
"step": 1269
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.72,
|
| 204 |
+
"learning_rate": 4.762790697674419e-05,
|
| 205 |
+
"loss": 0.464,
|
| 206 |
+
"step": 1316
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.85,
|
| 210 |
+
"learning_rate": 4.749127906976744e-05,
|
| 211 |
+
"loss": 0.47,
|
| 212 |
+
"step": 1363
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 3.98,
|
| 216 |
+
"learning_rate": 4.7354651162790695e-05,
|
| 217 |
+
"loss": 0.4643,
|
| 218 |
+
"step": 1410
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 4.0,
|
| 222 |
+
"eval_accuracy": 0.7752527998260302,
|
| 223 |
+
"eval_loss": 0.4479929208755493,
|
| 224 |
+
"eval_macro_f1": 0.7729510982990464,
|
| 225 |
+
"eval_runtime": 29.7623,
|
| 226 |
+
"eval_samples_per_second": 309.015,
|
| 227 |
+
"eval_steps_per_second": 2.99,
|
| 228 |
+
"step": 1416
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.12,
|
| 232 |
+
"learning_rate": 4.7218023255813956e-05,
|
| 233 |
+
"loss": 0.4516,
|
| 234 |
+
"step": 1457
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 4.25,
|
| 238 |
+
"learning_rate": 4.708139534883721e-05,
|
| 239 |
+
"loss": 0.4284,
|
| 240 |
+
"step": 1504
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.38,
|
| 244 |
+
"learning_rate": 4.6944767441860464e-05,
|
| 245 |
+
"loss": 0.4292,
|
| 246 |
+
"step": 1551
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 4.51,
|
| 250 |
+
"learning_rate": 4.680813953488372e-05,
|
| 251 |
+
"loss": 0.4251,
|
| 252 |
+
"step": 1598
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 4.65,
|
| 256 |
+
"learning_rate": 4.667151162790698e-05,
|
| 257 |
+
"loss": 0.4173,
|
| 258 |
+
"step": 1645
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 4.78,
|
| 262 |
+
"learning_rate": 4.6534883720930234e-05,
|
| 263 |
+
"loss": 0.4122,
|
| 264 |
+
"step": 1692
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 4.91,
|
| 268 |
+
"learning_rate": 4.639825581395349e-05,
|
| 269 |
+
"loss": 0.4161,
|
| 270 |
+
"step": 1739
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 5.0,
|
| 274 |
+
"eval_accuracy": 0.7766663042296401,
|
| 275 |
+
"eval_loss": 0.43961769342422485,
|
| 276 |
+
"eval_macro_f1": 0.7711253496732764,
|
| 277 |
+
"eval_runtime": 29.1171,
|
| 278 |
+
"eval_samples_per_second": 315.863,
|
| 279 |
+
"eval_steps_per_second": 3.057,
|
| 280 |
+
"step": 1770
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 5.05,
|
| 284 |
+
"learning_rate": 4.626162790697675e-05,
|
| 285 |
+
"loss": 0.4198,
|
| 286 |
+
"step": 1786
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 5.18,
|
| 290 |
+
"learning_rate": 4.6125e-05,
|
| 291 |
+
"loss": 0.4047,
|
| 292 |
+
"step": 1833
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 5.31,
|
| 296 |
+
"learning_rate": 4.598837209302326e-05,
|
| 297 |
+
"loss": 0.4109,
|
| 298 |
+
"step": 1880
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 5.44,
|
| 302 |
+
"learning_rate": 4.585174418604651e-05,
|
| 303 |
+
"loss": 0.402,
|
| 304 |
+
"step": 1927
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 5.58,
|
| 308 |
+
"learning_rate": 4.571511627906977e-05,
|
| 309 |
+
"loss": 0.4146,
|
| 310 |
+
"step": 1974
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 5.71,
|
| 314 |
+
"learning_rate": 4.557848837209303e-05,
|
| 315 |
+
"loss": 0.3954,
|
| 316 |
+
"step": 2021
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.84,
|
| 320 |
+
"learning_rate": 4.544186046511628e-05,
|
| 321 |
+
"loss": 0.4254,
|
| 322 |
+
"step": 2068
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 5.97,
|
| 326 |
+
"learning_rate": 4.5305232558139535e-05,
|
| 327 |
+
"loss": 0.4244,
|
| 328 |
+
"step": 2115
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 6.0,
|
| 332 |
+
"eval_accuracy": 0.7843862128955094,
|
| 333 |
+
"eval_loss": 0.4381129741668701,
|
| 334 |
+
"eval_macro_f1": 0.7827506323705966,
|
| 335 |
+
"eval_runtime": 29.0184,
|
| 336 |
+
"eval_samples_per_second": 316.937,
|
| 337 |
+
"eval_steps_per_second": 3.067,
|
| 338 |
+
"step": 2124
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.11,
|
| 342 |
+
"learning_rate": 4.5168604651162796e-05,
|
| 343 |
+
"loss": 0.4082,
|
| 344 |
+
"step": 2162
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 6.24,
|
| 348 |
+
"learning_rate": 4.503197674418605e-05,
|
| 349 |
+
"loss": 0.401,
|
| 350 |
+
"step": 2209
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 6.37,
|
| 354 |
+
"learning_rate": 4.4895348837209305e-05,
|
| 355 |
+
"loss": 0.3686,
|
| 356 |
+
"step": 2256
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 6.51,
|
| 360 |
+
"learning_rate": 4.475872093023256e-05,
|
| 361 |
+
"loss": 0.4074,
|
| 362 |
+
"step": 2303
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.64,
|
| 366 |
+
"learning_rate": 4.462209302325582e-05,
|
| 367 |
+
"loss": 0.3775,
|
| 368 |
+
"step": 2350
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 6.77,
|
| 372 |
+
"learning_rate": 4.4485465116279074e-05,
|
| 373 |
+
"loss": 0.4056,
|
| 374 |
+
"step": 2397
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 6.9,
|
| 378 |
+
"learning_rate": 4.434883720930233e-05,
|
| 379 |
+
"loss": 0.3998,
|
| 380 |
+
"step": 2444
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.0,
|
| 384 |
+
"eval_accuracy": 0.7610090246819615,
|
| 385 |
+
"eval_loss": 0.4613344073295593,
|
| 386 |
+
"eval_macro_f1": 0.7609479537198214,
|
| 387 |
+
"eval_runtime": 29.0367,
|
| 388 |
+
"eval_samples_per_second": 316.737,
|
| 389 |
+
"eval_steps_per_second": 3.065,
|
| 390 |
+
"step": 2478
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 7.04,
|
| 394 |
+
"learning_rate": 4.421220930232558e-05,
|
| 395 |
+
"loss": 0.3861,
|
| 396 |
+
"step": 2491
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 7.17,
|
| 400 |
+
"learning_rate": 4.4075581395348844e-05,
|
| 401 |
+
"loss": 0.3773,
|
| 402 |
+
"step": 2538
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 7.3,
|
| 406 |
+
"learning_rate": 4.39389534883721e-05,
|
| 407 |
+
"loss": 0.3902,
|
| 408 |
+
"step": 2585
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.44,
|
| 412 |
+
"learning_rate": 4.380232558139535e-05,
|
| 413 |
+
"loss": 0.3815,
|
| 414 |
+
"step": 2632
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 7.57,
|
| 418 |
+
"learning_rate": 4.3665697674418606e-05,
|
| 419 |
+
"loss": 0.377,
|
| 420 |
+
"step": 2679
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 7.7,
|
| 424 |
+
"learning_rate": 4.352906976744186e-05,
|
| 425 |
+
"loss": 0.3835,
|
| 426 |
+
"step": 2726
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 7.83,
|
| 430 |
+
"learning_rate": 4.3392441860465115e-05,
|
| 431 |
+
"loss": 0.4,
|
| 432 |
+
"step": 2773
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 7.97,
|
| 436 |
+
"learning_rate": 4.325581395348837e-05,
|
| 437 |
+
"loss": 0.3806,
|
| 438 |
+
"step": 2820
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 8.0,
|
| 442 |
+
"eval_accuracy": 0.7876481461346091,
|
| 443 |
+
"eval_loss": 0.42375245690345764,
|
| 444 |
+
"eval_macro_f1": 0.7860128082271005,
|
| 445 |
+
"eval_runtime": 29.7134,
|
| 446 |
+
"eval_samples_per_second": 309.523,
|
| 447 |
+
"eval_steps_per_second": 2.995,
|
| 448 |
+
"step": 2832
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 8.1,
|
| 452 |
+
"learning_rate": 4.311918604651163e-05,
|
| 453 |
+
"loss": 0.3679,
|
| 454 |
+
"step": 2867
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 8.23,
|
| 458 |
+
"learning_rate": 4.2982558139534884e-05,
|
| 459 |
+
"loss": 0.3563,
|
| 460 |
+
"step": 2914
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 8.36,
|
| 464 |
+
"learning_rate": 4.284593023255814e-05,
|
| 465 |
+
"loss": 0.3514,
|
| 466 |
+
"step": 2961
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 8.5,
|
| 470 |
+
"learning_rate": 4.27093023255814e-05,
|
| 471 |
+
"loss": 0.3527,
|
| 472 |
+
"step": 3008
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 8.63,
|
| 476 |
+
"learning_rate": 4.2572674418604654e-05,
|
| 477 |
+
"loss": 0.3829,
|
| 478 |
+
"step": 3055
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 8.76,
|
| 482 |
+
"learning_rate": 4.243604651162791e-05,
|
| 483 |
+
"loss": 0.3756,
|
| 484 |
+
"step": 3102
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 8.9,
|
| 488 |
+
"learning_rate": 4.229941860465116e-05,
|
| 489 |
+
"loss": 0.3553,
|
| 490 |
+
"step": 3149
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 9.0,
|
| 494 |
+
"eval_accuracy": 0.7750353376100902,
|
| 495 |
+
"eval_loss": 0.45743581652641296,
|
| 496 |
+
"eval_macro_f1": 0.7614865114284046,
|
| 497 |
+
"eval_runtime": 29.6041,
|
| 498 |
+
"eval_samples_per_second": 310.667,
|
| 499 |
+
"eval_steps_per_second": 3.006,
|
| 500 |
+
"step": 3186
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 9.03,
|
| 504 |
+
"learning_rate": 4.216279069767442e-05,
|
| 505 |
+
"loss": 0.3757,
|
| 506 |
+
"step": 3196
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 9.16,
|
| 510 |
+
"learning_rate": 4.202616279069768e-05,
|
| 511 |
+
"loss": 0.3445,
|
| 512 |
+
"step": 3243
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 9.29,
|
| 516 |
+
"learning_rate": 4.188953488372093e-05,
|
| 517 |
+
"loss": 0.3753,
|
| 518 |
+
"step": 3290
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 9.43,
|
| 522 |
+
"learning_rate": 4.1752906976744186e-05,
|
| 523 |
+
"loss": 0.3587,
|
| 524 |
+
"step": 3337
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 9.56,
|
| 528 |
+
"learning_rate": 4.161627906976745e-05,
|
| 529 |
+
"loss": 0.3513,
|
| 530 |
+
"step": 3384
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 9.69,
|
| 534 |
+
"learning_rate": 4.14796511627907e-05,
|
| 535 |
+
"loss": 0.3578,
|
| 536 |
+
"step": 3431
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 9.82,
|
| 540 |
+
"learning_rate": 4.1343023255813955e-05,
|
| 541 |
+
"loss": 0.3661,
|
| 542 |
+
"step": 3478
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 9.96,
|
| 546 |
+
"learning_rate": 4.120639534883721e-05,
|
| 547 |
+
"loss": 0.3506,
|
| 548 |
+
"step": 3525
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 10.0,
|
| 552 |
+
"eval_accuracy": 0.788409263890399,
|
| 553 |
+
"eval_loss": 0.4321058988571167,
|
| 554 |
+
"eval_macro_f1": 0.7788198668907367,
|
| 555 |
+
"eval_runtime": 28.9187,
|
| 556 |
+
"eval_samples_per_second": 318.03,
|
| 557 |
+
"eval_steps_per_second": 3.078,
|
| 558 |
+
"step": 3540
|
| 559 |
+
}
|
| 560 |
+
],
|
| 561 |
+
"max_steps": 17700,
|
| 562 |
+
"num_train_epochs": 50,
|
| 563 |
+
"total_flos": 1.45995167145984e+16,
|
| 564 |
+
"trial_name": null,
|
| 565 |
+
"trial_params": null
|
| 566 |
+
}
|
data_curation&frozen_layers/dataset1_all/F4/.DS_Store
CHANGED
|
Binary files a/data_curation&frozen_layers/dataset1_all/F4/.DS_Store and b/data_curation&frozen_layers/dataset1_all/F4/.DS_Store differ
|
|
|
data_curation&frozen_layers/dataset1_all/F4/all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7649233445688811,
|
| 3 |
+
"test_loss": 0.4680415391921997,
|
| 4 |
+
"test_macro_f1": 0.7626351149703829,
|
| 5 |
+
"test_runtime": 29.0877,
|
| 6 |
+
"test_samples_per_second": 316.182,
|
| 7 |
+
"test_steps_per_second": 3.06
|
| 8 |
+
}
|
data_curation&frozen_layers/dataset1_all/F4/config.json
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/vsphhome/fengguoqing/Geneformer",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"BertForSequenceClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.02,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"gradient_checkpointing": false,
|
| 9 |
+
"hidden_act": "relu",
|
| 10 |
+
"hidden_dropout_prob": 0.02,
|
| 11 |
+
"hidden_size": 256,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 512,
|
| 14 |
+
"layer_norm_eps": 1e-12,
|
| 15 |
+
"max_position_embeddings": 2048,
|
| 16 |
+
"model_type": "bert",
|
| 17 |
+
"num_attention_heads": 4,
|
| 18 |
+
"num_hidden_layers": 6,
|
| 19 |
+
"pad_token_id": 0,
|
| 20 |
+
"position_embedding_type": "absolute",
|
| 21 |
+
"problem_type": "single_label_classification",
|
| 22 |
+
"torch_dtype": "float32",
|
| 23 |
+
"transformers_version": "4.28.0",
|
| 24 |
+
"type_vocab_size": 2,
|
| 25 |
+
"use_cache": true,
|
| 26 |
+
"vocab_size": 25426
|
| 27 |
+
}
|
data_curation&frozen_layers/dataset1_all/F4/eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"test_accuracy": 0.7649233445688811,
|
| 3 |
+
"test_loss": 0.4680415391921997,
|
| 4 |
+
"test_macro_f1": 0.7626351149703829,
|
| 5 |
+
"test_runtime": 29.0877,
|
| 6 |
+
"test_samples_per_second": 316.182,
|
| 7 |
+
"test_steps_per_second": 3.06
|
| 8 |
+
}
|
data_curation&frozen_layers/dataset1_all/F4/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b4e30a5c69f99c38707598aceaf66d323941584b710b19d93b3988986bae5712
|
| 3 |
+
size 65263500
|
data_curation&frozen_layers/dataset1_all/F4/predictions.pickle
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fbdf534b6eb865c439392081dfc84a258031d58a19209b88cebb3d6dd5135b98
|
| 3 |
+
size 147612
|
data_curation&frozen_layers/dataset1_all/F4/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c53ac15664fb8080b3ea0c1aa4972bf7b42c11445f34425cc83da0049faacd5
|
| 3 |
+
size 41105778
|
data_curation&frozen_layers/dataset1_all/F4/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:65b67c12eae2f1f8f7e56cf5c4ad756416f58327aa916250e907b979787424ba
|
| 3 |
+
size 15006
|
data_curation&frozen_layers/dataset1_all/F4/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7c2abe87f776f0304299cfd007f13ebbea52990058bcfa21eeda52477f26c98d
|
| 3 |
+
size 1064
|