nroggendorff commited on
Commit
f957dfd
·
verified ·
1 Parent(s): 2a359e8

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +2 -2
train.py CHANGED
@@ -10,8 +10,8 @@ from tokenizers import ByteLevelBPETokenizer
10
  BATCH_SIZE = 64
11
  EPOCHS = 20
12
  LEARNING_RATE = 2e-4
13
- FACTOR = 1440
14
- MAX_SEQ_LENGTH = FACTOR // 8
15
  VOCAB_SIZE = 32000
16
  INPUT_DATASET = "HuggingFaceTB/smollm-corpus"
17
  INSTRUCT_DATASET = "nroggendorff/elephant"
 
10
  BATCH_SIZE = 64
11
  EPOCHS = 20
12
  LEARNING_RATE = 2e-4
13
+ FACTOR = 1408
14
+ MAX_SEQ_LENGTH = 128
15
  VOCAB_SIZE = 32000
16
  INPUT_DATASET = "HuggingFaceTB/smollm-corpus"
17
  INSTRUCT_DATASET = "nroggendorff/elephant"