train-diffuser

Paused

nroggendorff commited on Jul 17, 2024

Commit

177a90e

verified ·

1 Parent(s): c26b20b

Update app.py

Files changed (1) hide show

app.py CHANGED Viewed

@@ -9,7 +9,7 @@ import requests as rq
 import gc
 from tokenizers import ByteLevelBPETokenizer
-dataset = load_dataset("nroggendorff/openhermes", split="train").select(range(int(1e+4)))
 def get_training_corpus():
     for i in range(0, len(dataset), 1000):

 import gc
 from tokenizers import ByteLevelBPETokenizer
+dataset = load_dataset("nroggendorff/openhermes", split="train")#.select(range(int(1e+5)))
 def get_training_corpus():
     for i in range(0, len(dataset), 1000):