File size: 3,738 Bytes
53e0dae | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 | import json
import torch
from litdata import TokensLoader, optimize
from torch.utils._pytree import tree_map
from litgpt.data.text_files import TextFiles
class Tokenizer:
bos_id = 0
def encode(self, text, bos, eos):
assert bos
assert not eos
return [self.bos_id] + [ord(c) for c in text]
def tokenize(data):
for story in data:
yield torch.tensor(story)
def fake_chunk(path, data):
optimize(
fn=tokenize,
inputs=[data] * len(data),
output_dir=str(path),
num_workers=1,
chunk_bytes="200MB",
item_loader=TokensLoader(),
)
def test_textfiles_datamodule(tmp_path):
from litgpt.data.text_files import TextFiles
data_dir = tmp_path / "textfiles"
datamodule = TextFiles(train_data_path=data_dir, num_workers=1)
datamodule.connect(max_seq_length=2, tokenizer=Tokenizer())
# simulate `datamodule.prepare_data`
train_data_dir = data_dir / "train"
train_data_dir.mkdir(parents=True)
fake_chunk(train_data_dir, [[12], [0, 23, 15, 63, 0], [73, 5, 0, 1, 1999, 0, 13]])
datamodule.setup()
tr_dataloader = datamodule.train_dataloader()
tr_dataloader.shuffle = False
actual = tree_map(torch.Tensor.tolist, list(tr_dataloader))
# there is 1 sample per index in the data (13)
assert actual == [
[[73, 5, 0]],
[[12, 0, 23]],
[[5, 0, 1]],
[[0, 73, 5]],
[[1999, 0, 13]],
[[0, 1, 1999]],
[[1, 1999, 0]],
[[0, 23, 15]],
[[13, 12, 0]],
[[63, 0, 73]],
[[23, 15, 63]],
[[15, 63, 0]],
[[0, 13, 12]],
]
class MockTokenizer:
bos_id = 0
eos_id = 1
use_bos = True
def encode(self, text, bos=True, eos=False, device=None, max_length=-1):
# Simple: map each character to its ordinal + 2
tokens = [ord(c) + 2 for c in text]
if bos:
tokens = [self.bos_id] + tokens
if eos:
tokens.append(self.eos_id)
if max_length > 0:
tokens = tokens[:max_length]
return torch.tensor(tokens, dtype=torch.long, device=device)
def decode(self, tensor):
ids = tensor.tolist() if tensor.ndim > 0 else [tensor.item()]
chars = []
for tid in ids:
if tid == self.bos_id:
chars.append("<BOS>")
elif tid == self.eos_id:
chars.append("<EOS>")
else:
chars.append(chr(tid - 2))
return "".join(chars)
def decode_stream(self, token_stream, device=None):
for token in token_stream:
yield self.decode(token)
@property
def vocab_size(self):
return 130
def test_textfiles_token_loader(tmp_path):
# Create the directory for text files
data_dir = tmp_path / "textfiles"
data_dir.mkdir(parents=True, exist_ok=True)
# Write sample training data to the directory
sample_texts = ["hello world", "foo bar", "lorem ipsum"]
for i, text in enumerate(sample_texts):
(data_dir / f"{i}.txt").write_text(text)
datamodule = TextFiles(train_data_path=data_dir, num_workers=1)
datamodule.connect(max_seq_length=2, tokenizer=MockTokenizer())
datamodule.prepare_data()
# ensure training set uses tokens loader
index_json = data_dir / "train" / "index.json"
assert index_json.exists()
meta = json.loads(index_json.read_text())
assert meta["config"]["item_loader"] == "TokensLoader"
# ensure validation set uses tokens loader
index_json = data_dir / "val" / "index.json"
assert index_json.exists()
meta = json.loads(index_json.read_text())
assert meta["config"]["item_loader"] == "TokensLoader"
|