anlp-assignment1-c1-byte300-aligned / source_tokenizer.json
Anivarth's picture
Upload C1 best checkpoint and metrics
8668eb9 verified
Raw History Blame Contribute Delete
1.48 kB
{
"version": 1,
"type": "byte_bpe",
"special_tokens": [
"[PAD]",
"[UNK]",
"[BOS]",
"[EOS]"
],
"max_length": 2304,
"binary_chunk_size": 8,
"merges": [
[
45,
116
],
[
89,
22
],
[
91,
26
],
[
40,
114
],
[
91,
20
],
[
87,
101
],
[
57,
22
],
[
47,
112
],
[
22,
72
],
[
22,
70
],
[
69,
26
],
[
69,
20
],
[
54,
114
],
[
67,
116
],
[
73,
101
],
[
65,
112
],
[
40,
94
],
[
74,
92
],
[
74,
98
],
[
57,
42
],
[
60,
60
],
[
62,
40
],
[
46,
112
],
[
90,
26
],
[
44,
116
],
[
86,
101
],
[
39,
22
],
[
71,
22
],
[
90,
20
],
[
41,
114
],
[
36,
112
],
[
38,
116
],
[
96,
26
],
[
87,
70
],
[
51,
114
],
[
96,
20
],
[
91,
72
],
[
87,
72
],
[
41,
66
],
[
91,
70
]
]
}