Hugging Face's logo

SLAM5566
/
huggingface_models

Model card Files
xet
Community
5
huggingface_models
File size: 2,191 Bytes
297c44b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
# Generated 2021-10-05 from:
# /mnt/md0/user_wayne/speechbrain/recipes/MATBN/LM/hparams/TransformerLM_seg_char.yaml
# yamllint disable
output_folder: results/TransformerLM_seg_char
save_folder: results/TransformerLM_seg_char/save
train_log: results/TransformerLM_seg_char/train_log.txt
num_workers: 4

data_folder: results/prepare_seg

tokenizer_file: results/tokenizer_seg_bpe5k_char/5000_char.model

tokenizer: &id001 !new:sentencepiece.SentencePieceProcessor

pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
  collect_in: results/TransformerLM_seg_char/tokenizer
  loadables:
    tokenizer: *id001
  paths:
    tokenizer: results/tokenizer_seg_bpe5k_char/5000_char.model

train_logger: !new:speechbrain.utils.train_logger.FileTrainLogger
  save_file: results/TransformerLM_seg_char/train_log.txt

# Training parameters
number_of_epochs: 20
batch_size: 64
lr: 1
accumulation_steps: 2
ckpt_interval_minutes: 15

epoch_counter: &id004 !new:speechbrain.utils.epoch_loop.EpochCounter

  limit: 20

# Dataloader options
train_dataloader_opts:
  batch_size: 64
  num_workers: 4
  shuffle: true
  pin_memory: true

valid_dataloader_opts:
  batch_size: 64
  num_workers: 4

test_dataloader_opts:
  batch_size: 64
  num_workers: 4

# Model parameters
d_model: 576

# Outputs
output_neurons: 5000
blank_index: 0
bos_index: 1
eos_index: 2
unk_index: 0
pad_index: 0

model: &id002 !new:speechbrain.lobes.models.transformer.TransformerLM.TransformerLM

  vocab: 5000
  d_model: 576
  nhead: 6
  num_encoder_layers: 6
  num_decoder_layers: 0
  d_ffn: 1538
  dropout: 0.2
  activation: !name:torch.nn.GELU
  normalize_before: false

modules:
  model: *id002
lr_annealing: &id003 !new:speechbrain.nnet.schedulers.NoamScheduler
  lr_initial: 1
  n_warmup_steps: 1000
  model_size: 576

checkpointer: !new:speechbrain.utils.checkpoints.Checkpointer
  checkpoints_dir: results/TransformerLM_seg_char/save
  recoverables:
    model: *id002
    scheduler: *id003
    counter: *id004
log_softmax: !new:speechbrain.nnet.activations.Softmax
  apply_log: true

optimizer: !name:torch.optim.Adam
  lr: 0
  betas: (0.9, 0.98)
  eps: 0.000000001

compute_cost: !name:speechbrain.nnet.losses.nll_loss