nroggendorff commited on
Commit
577a57e
·
verified ·
1 Parent(s): a956d3f

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +2 -2
train.py CHANGED
@@ -13,7 +13,7 @@ EPOCHS = 3
13
  LEARNING_RATE = 2e-2
14
  FACTOR = 256
15
  VOCAB_SIZE = 32000
16
- INPUT_DATASET = "nroggendorff/openhermes"
17
  OUTPUT_REPO = "smallama"
18
  FP16 = True
19
  WARMUP_STEPS = 20
@@ -23,7 +23,7 @@ CLIPPING = 1.0
23
  PUSH_TO_HUB = True
24
 
25
  def load_data():
26
- dataset = load_dataset(INPUT_DATASET, split="train").select(range(int(2e+5)))
27
  return dataset
28
 
29
  def create_tokenizer(training_corpus):
 
13
  LEARNING_RATE = 2e-2
14
  FACTOR = 256
15
  VOCAB_SIZE = 32000
16
+ INPUT_DATASET = "nroggendorff/elephant"
17
  OUTPUT_REPO = "smallama"
18
  FP16 = True
19
  WARMUP_STEPS = 20
 
23
  PUSH_TO_HUB = True
24
 
25
  def load_data():
26
+ dataset = load_dataset(INPUT_DATASET, split="train")#.select(range(int(2e+5)))
27
  return dataset
28
 
29
  def create_tokenizer(training_corpus):