BlinkDL 3 years ago
parent 1945cb58ed
commit 4ca274aad7

@ -183,9 +183,12 @@ class MyDataset(Dataset):
if (args.my_qa_mask == 0) or (data == self.data_pile): if (args.my_qa_mask == 0) or (data == self.data_pile):
i = i + args.my_pile_shift i = i + args.my_pile_shift
# print(f"epoch {epoch} idx {idx} rank {rank}/{world_size} ii {ii} pos {round(i / self.data_size, 3)}") # print(f"epoch {epoch} idx {idx} rank {rank}/{world_size} ii {ii} pos {round(i / self.data_size, 3)}")
else: elif args.my_pile_stage == 4:
# cheat: pick a random spot in dataset # cheat: pick a random spot in dataset
if args.my_pile_version == 1:
i = np.random.randint(0, self.data_size - req_len) i = np.random.randint(0, self.data_size - req_len)
else:
i = np.random.randint(0, self.data_size)
if args.data_type == "binidx": if args.data_type == "binidx":
if args.my_pile_version == 1: if args.my_pile_version == 1:

@ -177,7 +177,7 @@ if __name__ == "__main__":
args.magic_prime = 423736637 args.magic_prime = 423736637
args.epoch_count = 10508 args.epoch_count = 10508
elif args.ctx_len == 8192: elif args.ctx_len == 8192:
args.magic_prime = 211868309 args.magic_prime = 211868243
args.epoch_count = 5253 args.epoch_count = 5253
if args.my_pile_shift < 0: if args.my_pile_shift < 0:
args.my_pile_shift = 0 args.my_pile_shift = 0

Loading…
Cancel
Save