add config
Browse files- riksdag_200h.yaml +58 -0
riksdag_200h.yaml
ADDED
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# @package _group_
|
2 |
+
|
3 |
+
common:
|
4 |
+
fp16: true
|
5 |
+
log_format: json
|
6 |
+
log_interval: 200
|
7 |
+
|
8 |
+
checkpoint:
|
9 |
+
no_epoch_checkpoints: true
|
10 |
+
best_checkpoint_metric: wer
|
11 |
+
|
12 |
+
task:
|
13 |
+
_name: audio_finetuning
|
14 |
+
data: /home/joregan/fairseq-data2/
|
15 |
+
normalize: true
|
16 |
+
labels: ltr
|
17 |
+
|
18 |
+
dataset:
|
19 |
+
num_workers: 6
|
20 |
+
max_tokens: 128000
|
21 |
+
skip_invalid_size_inputs_valid_test: true
|
22 |
+
valid_subset: valid
|
23 |
+
|
24 |
+
distributed_training:
|
25 |
+
ddp_backend: legacy_ddp
|
26 |
+
distributed_world_size: 8
|
27 |
+
|
28 |
+
criterion:
|
29 |
+
_name: ctc
|
30 |
+
zero_infinity: true
|
31 |
+
|
32 |
+
optimization:
|
33 |
+
max_update: 80000
|
34 |
+
lr: [0.00003]
|
35 |
+
sentence_avg: true
|
36 |
+
update_freq: [5]
|
37 |
+
|
38 |
+
optimizer:
|
39 |
+
_name: adam
|
40 |
+
adam_betas: (0.9,0.98)
|
41 |
+
adam_eps: 1e-08
|
42 |
+
|
43 |
+
lr_scheduler:
|
44 |
+
_name: tri_stage
|
45 |
+
phase_ratio: [0.1, 0.4, 0.5]
|
46 |
+
final_lr_scale: 0.05
|
47 |
+
|
48 |
+
model:
|
49 |
+
_name: wav2vec_ctc
|
50 |
+
w2v_path: /home/joregan/kb_wav2vec2_checkpoint/checkpoint_last.pt
|
51 |
+
apply_mask: true
|
52 |
+
mask_prob: 0.5
|
53 |
+
mask_channel_prob: 0.5
|
54 |
+
mask_channel_length: 64
|
55 |
+
layerdrop: 0.1
|
56 |
+
activation_dropout: 0.1
|
57 |
+
feature_grad_mult: 0.0
|
58 |
+
freeze_finetune_updates: 10000
|