nroggendorff commited on
Commit
b1e38e3
·
verified ·
1 Parent(s): e001176

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +3 -3
train.py CHANGED
@@ -8,11 +8,11 @@ from datasets import load_dataset, DatasetDict, Dataset
8
  from tokenizers import ByteLevelBPETokenizer
9
 
10
  MAX_SEQ_LENGTH = 128
11
- BATCH_SIZE = 64
12
  EPOCHS = 2
13
- LEARNING_RATE = 1e-4
14
  FACTOR = 1024
15
- VOCAB_SIZE = 3200
16
  INPUT_DATASET = "HuggingFaceTB/smollm-corpus"
17
  INSTRUCT_DATASET = "nroggendorff/openhermes"
18
  OUTPUT_REPO = "smallama"
 
8
  from tokenizers import ByteLevelBPETokenizer
9
 
10
  MAX_SEQ_LENGTH = 128
11
+ BATCH_SIZE = 16
12
  EPOCHS = 2
13
+ LEARNING_RATE = 1e-3
14
  FACTOR = 1024
15
+ VOCAB_SIZE = 32000
16
  INPUT_DATASET = "HuggingFaceTB/smollm-corpus"
17
  INSTRUCT_DATASET = "nroggendorff/openhermes"
18
  OUTPUT_REPO = "smallama"