Skip to content

Commit 1f0487d

Browse files
authored
Update llama special tokens
Differential Revision: D61730848 Pull Request resolved: #4876
1 parent 2b2911b commit 1f0487d

File tree

1 file changed

+6
-6
lines changed

1 file changed

+6
-6
lines changed

examples/models/llama2/tokenizer/llama_tiktoken.cpp

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -23,15 +23,15 @@ _get_default_special_tokens() {
2323
"<|end_of_text|>",
2424
"<|reserved_special_token_0|>",
2525
"<|reserved_special_token_1|>",
26-
"<|reserved_special_token_2|>",
27-
"<|reserved_special_token_3|>",
26+
"<|finetune_right_pad_id|>",
27+
"<|step_id|>",
2828
"<|start_header_id|>",
2929
"<|end_header_id|>",
30-
"<|reserved_special_token_4|>",
31-
"<|eot_id|>"});
32-
30+
"<|eom_id|>",
31+
"<|eot_id|>",
32+
"<|python_tag|>"});
3333
// pad the rest of the special tokens with reserved tokens
34-
ssize_t reserved_special_token_num = 5;
34+
ssize_t reserved_special_token_num = 2;
3535
while (special_tokens->size() < kSpecialTokensSize) {
3636
special_tokens->emplace_back(
3737
"<|reserved_special_token_" +

0 commit comments

Comments
 (0)