File size: 5,422 Bytes
cd4aacd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 |
[2023-12-07 03:35:49,908][hydra][INFO] -
estimator:
accelerator: gpu
precision: 32
deterministic: true
tf32_mode: high
callbacks:
timer:
_target_: energizer.active_learning.callbacks.Timer
save_outputs:
_target_: src.callbacks.SaveOutputs
dirpath: ./logs/
instance_level: false
batch_level: false
epoch_level: false
early_stopping:
_target_: energizer.callbacks.early_stopping.EarlyStopping
monitor: train/avg_f1_minclass
stage: train
interval: epoch
mode: max
min_delta: 1.0e-05
patience: 10
stopping_threshold: null
divergence_threshold: null
verbose: true
model_checkpoint:
_target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
dirpath: .checkpoints
monitor: train/avg_f1_minclass
stage: train
mode: max
save_last: false
save_top_k: 1
verbose: true
loggers:
tensorboard:
_target_: energizer.loggers.TensorBoardLogger
root_dir: ./
name: tb_logs
version: null
data:
batch_size: 32
eval_batch_size: 256
num_workers: 32
pin_memory: true
drop_last: false
persistent_workers: true
shuffle: true
seed: 123456
replacement: false
max_length: 512
active_data:
budget: 100
positive_budget: 5
seed: 654321
fit:
min_steps: 100
max_epochs: 10
learning_rate: 4.0e-05
optimizer: adamw
log_interval: 1
enable_progress_bar: false
limit_train_batches: null
limit_validation_batches: null
active_fit:
max_budget: 5000
query_size: 25
reinit_model: true
limit_pool_batches: null
limit_test_batches: null
test:
log_interval: 1
enable_progress_bar: false
limit_batches: null
strategy:
name: anchoral_entropy
args:
seed: 42
subpool_size: 10000
num_neighbours: 50
max_search_size: null
num_anchors: 10
anchor_strategy_minority: kmeans_pp_sampling
anchor_strategy_majority: kmeans_pp_sampling
minority_classes_ids: null
model:
name: deberta_v3-base
seed: 654321
dataset:
name: agnews-business-.01
text_column: text
label_column: labels
uid_column: uid
prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/agnews-business-.01
processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/agnews
minority_classes:
- 1
index_metric: all-mpnet-base-v2_cosine
log_interval: 1
enable_progress_bar: false
limit_batches: null
seed: 42
experiment_group: deberta_v3-base/new
run_name: agnews-business-.01/deberta_v3-base_anchoral_entropy_2023-12-06T10-52-24
data_path: /rds/user/pl487/hpc-work/anchoral/data
======================================================================
[2023-12-07 03:35:49,931][hydra][INFO] - Running active learning with strategy {'name': 'anchoral_entropy', 'args': {'seed': 42, 'subpool_size': 10000, 'num_neighbours': 50, 'max_search_size': None, 'num_anchors': 10, 'anchor_strategy_minority': 'kmeans_pp_sampling', 'anchor_strategy_majority': 'kmeans_pp_sampling', 'minority_classes_ids': None}}
[2023-12-07 03:35:49,949][hydra][INFO] - Seed enabled: 42
[2023-12-07 03:35:50,941][hydra][INFO] - loading index from /rds/user/pl487/hpc-work/anchoral/data/processed/agnews/all-mpnet-base-v2_cosine
[2023-12-07 03:35:52,895][hydra][INFO] - Labelled size: 100 Pool size: 90810 Test size: 7600
Label distribution:
| | labels | count | perc |
|---:|:---------|--------:|-------:|
| 0 | Negative | 95 | 0.95 |
| 1 | Positive | 5 | 0.05 |
[2023-12-07 03:35:52,910][hydra][INFO] - Batch:
{<InputKeys.INPUT_IDS: 'input_ids'>: tensor([[ 1, 1864, 294, 41142, 2729, 2152, 19979, 53884, 5147, 1050,
279, 7439, 2298, 927, 7710, 70907, 2]]), <InputKeys.ATT_MASK: 'attention_mask'>: tensor([[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]]), <InputKeys.LABELS: 'labels'>: tensor([0]), <InputKeys.ON_CPU: 'on_cpu'>: {<SpecialKeys.ID: 'uid'>: [122709]}}
[2023-12-07 03:35:54,719][hydra][INFO] - Loggers: {'tensorboard': <energizer.loggers.tensorboard.TensorBoardLogger object at 0x14699566a700>}
[2023-12-07 03:35:54,719][hydra][INFO] - Callbacks: {'timer': <energizer.active_learning.callbacks.Timer object at 0x1469954dbf40>, 'save_outputs': <src.callbacks.SaveOutputs object at 0x1469954db430>, 'early_stopping': <energizer.callbacks.early_stopping.EarlyStopping object at 0x146995583130>, 'model_checkpoint': <energizer.callbacks.model_checkpoint.ModelCheckpoint object at 0x146995583af0>}
[2023-12-07 03:35:54,773][hydra][INFO] -
| Name | Type | Params
----------------------------------------------
0 | deberta | DebertaV2Model | 183 M
1 | pooler | ContextPooler | 590 K
2 | classifier | Linear | 1.5 K
3 | dropout | StableDropout | 0
----------------------------------------------
184 M Trainable params
0 Non-trainable params
184 M Total params
737.695 Total estimated model params size (MB)
0.00 GB CUDA Memory used
[2023-12-07 09:33:08,507][submitit][INFO] - Job has timed out. Ran 357 minutes out of requested 360 minutes.
[2023-12-07 09:33:08,526][submitit][WARNING] - Caught signal SIGUSR2 on gpu-q-60: this job is timed-out.
[2023-12-07 09:33:08,543][submitit][INFO] - Calling checkpoint method.
[2023-12-07 09:33:08,568][submitit][INFO] - Job not requeued because: timed-out too many times.
[2023-12-07 09:33:08,568][submitit][WARNING] - Bypassing signal SIGCONT
[2023-12-07 09:33:08,586][submitit][INFO] - Job completed successfully
|