nroggendorff commited on
Commit
2aad646
·
verified ·
1 Parent(s): 4c29d32

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +2 -2
train.py CHANGED
@@ -8,9 +8,9 @@ from datasets import load_dataset, DatasetDict, Dataset
8
  from tokenizers import ByteLevelBPETokenizer
9
 
10
  MAX_SEQ_LENGTH = 128
11
- BATCH_SIZE = 16
12
  EPOCHS = 2
13
- LEARNING_RATE = 2e-3
14
  FACTOR = 1024
15
  VOCAB_SIZE = 32000
16
  INPUT_DATASET = "HuggingFaceTB/smollm-corpus"
 
8
  from tokenizers import ByteLevelBPETokenizer
9
 
10
  MAX_SEQ_LENGTH = 128
11
+ BATCH_SIZE = 64
12
  EPOCHS = 2
13
+ LEARNING_RATE = 2e-2
14
  FACTOR = 1024
15
  VOCAB_SIZE = 32000
16
  INPUT_DATASET = "HuggingFaceTB/smollm-corpus"