nroggendorff commited on
Commit
3be17e6
·
verified ·
1 Parent(s): e2cf6d7

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +2 -2
train.py CHANGED
@@ -8,10 +8,10 @@ from datasets import load_dataset, DatasetDict, Dataset
8
  from tokenizers import ByteLevelBPETokenizer
9
 
10
  MAX_SEQ_LENGTH = 512
11
- BATCH_SIZE = 128
12
  EPOCHS = 3
13
  LEARNING_RATE = 2.5e-5
14
- FACTOR = 256
15
  VOCAB_SIZE = 32000
16
  INPUT_DATASET = "HuggingFaceTB/smollm-corpus"
17
  INSTRUCT_DATASET = "nroggendorff/openhermes"
 
8
  from tokenizers import ByteLevelBPETokenizer
9
 
10
  MAX_SEQ_LENGTH = 512
11
+ BATCH_SIZE = 16
12
  EPOCHS = 3
13
  LEARNING_RATE = 2.5e-5
14
+ FACTOR = 1024
15
  VOCAB_SIZE = 32000
16
  INPUT_DATASET = "HuggingFaceTB/smollm-corpus"
17
  INSTRUCT_DATASET = "nroggendorff/openhermes"