Upload config_resolved.yaml with huggingface_hub
Browse files- config_resolved.yaml +97 -0
config_resolved.yaml
ADDED
|
@@ -0,0 +1,97 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
project:
|
| 2 |
+
slug: ssb_rtx6000x2
|
| 3 |
+
name: SSB RTX6000x2
|
| 4 |
+
owner: null
|
| 5 |
+
experiment:
|
| 6 |
+
name: ssb_npz_rtx6000x2_run
|
| 7 |
+
output_dir: artifacts/ssb_rtx6000x2/runs
|
| 8 |
+
tags:
|
| 9 |
+
provider: local
|
| 10 |
+
gpu: rtx6000x2
|
| 11 |
+
dataset: ssb_npz_v1
|
| 12 |
+
notes: null
|
| 13 |
+
data:
|
| 14 |
+
dataset_name: npz_classification
|
| 15 |
+
dataset_version: ssb_npz_v1
|
| 16 |
+
source_uri: data/processed/ssb_npz_v1
|
| 17 |
+
batch_size: 305
|
| 18 |
+
num_workers: 8
|
| 19 |
+
shuffle: true
|
| 20 |
+
pin_memory: true
|
| 21 |
+
split:
|
| 22 |
+
train: 0.8
|
| 23 |
+
validation: 0.1
|
| 24 |
+
test: 0.1
|
| 25 |
+
preprocessing:
|
| 26 |
+
class_weighted_loss: true
|
| 27 |
+
augmentations: []
|
| 28 |
+
cache_dir: data/cache
|
| 29 |
+
model:
|
| 30 |
+
name: mlp_classifier
|
| 31 |
+
parameters:
|
| 32 |
+
input_dim: 144
|
| 33 |
+
hidden_dims:
|
| 34 |
+
- 512
|
| 35 |
+
- 256
|
| 36 |
+
- 128
|
| 37 |
+
num_classes: 3
|
| 38 |
+
dropout: 0.226947161178787
|
| 39 |
+
args: null
|
| 40 |
+
pretrained: false
|
| 41 |
+
checkpoint_path: null
|
| 42 |
+
training:
|
| 43 |
+
epochs: 45
|
| 44 |
+
max_steps: null
|
| 45 |
+
log_every_n_steps: 25
|
| 46 |
+
val_every_n_steps: null
|
| 47 |
+
gradient_clip_norm: 1.0
|
| 48 |
+
optimizer:
|
| 49 |
+
name: adamw
|
| 50 |
+
lr: 0.0002547762821605831
|
| 51 |
+
weight_decay: 0.010098717791485455
|
| 52 |
+
eps: 1.0e-08
|
| 53 |
+
betas: null
|
| 54 |
+
args: {}
|
| 55 |
+
scheduler:
|
| 56 |
+
name: cosine
|
| 57 |
+
args:
|
| 58 |
+
t_max: 43
|
| 59 |
+
early_stop_patience: 0
|
| 60 |
+
distributed:
|
| 61 |
+
backend: nccl
|
| 62 |
+
world_size: 2
|
| 63 |
+
num_nodes: 1
|
| 64 |
+
node_rank: 0
|
| 65 |
+
master_addr: localhost
|
| 66 |
+
master_port: 29500
|
| 67 |
+
gradient_accumulation_steps: 4
|
| 68 |
+
mixed_precision: true
|
| 69 |
+
seed: 2026
|
| 70 |
+
checkpoint:
|
| 71 |
+
dir: artifacts/ssb_rtx6000x2/checkpoints
|
| 72 |
+
monitor: val_loss
|
| 73 |
+
mode: min
|
| 74 |
+
save_top_k: 2
|
| 75 |
+
save_every_n_epochs: 2
|
| 76 |
+
save_last: true
|
| 77 |
+
resume_from: null
|
| 78 |
+
logging:
|
| 79 |
+
level: INFO
|
| 80 |
+
log_dir: logs/ssb_rtx6000x2
|
| 81 |
+
tensorboard: true
|
| 82 |
+
mlflow_tracking_uri: null
|
| 83 |
+
mlflow_experiment: null
|
| 84 |
+
flush_steps: 50
|
| 85 |
+
timeout_seconds: null
|
| 86 |
+
evaluation:
|
| 87 |
+
metrics:
|
| 88 |
+
- accuracy
|
| 89 |
+
- f1
|
| 90 |
+
- precision
|
| 91 |
+
- recall
|
| 92 |
+
batch_size: 1024
|
| 93 |
+
save_predictions: false
|
| 94 |
+
output_dir: artifacts/ssb_rtx6000x2/eval
|
| 95 |
+
generate_confusion_matrix: true
|
| 96 |
+
generate_calibration_plot: false
|
| 97 |
+
calibration_bins: 10
|