Add Deepseek-R1-Distill pre-tokenizer

This commit is contained in:
Iwan Kawrakow
2025-01-23 13:10:03 +02:00
parent dbf5d31d01
commit c2624b2fd3

View File

@@ -5591,7 +5591,7 @@ static void llm_load_vocab(
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_COMMAND_R;
vocab.tokenizer_clean_spaces = false;
} else if (
tokenizer_pre == "qwen2") {
tokenizer_pre == "qwen2" || tokenizer_pre == "deepseek-r1-qwen") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_QWEN2;
vocab.tokenizer_clean_spaces = false;
} else if (