WaveMind/EEG_Encoder/examples/train_classifier_only.yaml at master · FreedomIntelligence/WaveMind · GitHub

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
# Pure supervised classifier training (no CLIP loss)
# Usage: python EEG_Encoder/run_CLIPtraining.py --config-name=train_classifier_only

defaults:
  - base
  - _self_

# ============================================================================
# DATA AUGMENTATION
# ============================================================================
augmentation:
  train:
    enable: false
    normalization_method: "random"
    zscore_epsilon: 1.0e-8
    amplitude_fluctuation_enable: true
    amplitude_fluctuation_range: [0.9, 1.1]
  test:
    enable: false
    normalization_method: "zscore_global"
    zscore_epsilon: 1.0e-8

experiment:
  models: ['ATMSmodify']
  datasets: ['ImageNetEEG']  # Good starting point (6-class classification)
  # gpu_number: ['all'] inherited from base
  limit_sample: 7960

training:
  DEFAULT_EPOCHS: 30
  DEFAULT_BATCH_SIZE: 512
  DEFAULT_LEARNING_RATE: 1.0e-3

classifier:
  lambda_clip: 0.0  # Pure classification, no CLIP loss (classifier auto-enabled)
  use_simple_head: false  # Use MLP classifier for better capacity
  hidden_dim: 512
  dropout: 0.5
  label_smoothing: 0.2

modes:
  save_model: false

features:
  use_shm: true
  auto_clean_shm_when_exit: false