ollama/llm/patches/05-default-pretokenizer.diff

33 lines
1.7 KiB
Diff
Raw Normal View History

2024-05-23 21:16:26 +00:00
diff --git a/llama.cpp b/llama.cpp
index 40d2ec2c..74f3ee9c 100644
2024-05-23 21:16:26 +00:00
--- a/llama.cpp
+++ b/llama.cpp
@@ -4642,16 +4642,7 @@ static void llm_load_vocab(
// for now, only BPE models have pre-tokenizers
if (vocab.type == LLAMA_VOCAB_TYPE_BPE) {
- if (tokenizer_pre.empty()) {
- LLAMA_LOG_WARN("%s: missing pre-tokenizer type, using: 'default'\n", __func__);
- LLAMA_LOG_WARN("%s: \n", __func__);
- LLAMA_LOG_WARN("%s: ************************************ \n", __func__);
- LLAMA_LOG_WARN("%s: GENERATION QUALITY WILL BE DEGRADED! \n", __func__);
- LLAMA_LOG_WARN("%s: CONSIDER REGENERATING THE MODEL \n", __func__);
- LLAMA_LOG_WARN("%s: ************************************ \n", __func__);
- LLAMA_LOG_WARN("%s: \n", __func__);
2024-05-23 21:16:26 +00:00
- vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
- } else if (
+ if (
tokenizer_pre == "default") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
2024-05-23 21:16:26 +00:00
} else if (
@@ -4703,7 +4694,8 @@ static void llm_load_vocab(
tokenizer_pre == "smaug-bpe") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_SMAUG;
2024-05-23 21:16:26 +00:00
} else {
- throw std::runtime_error(format("unknown pre-tokenizer type: '%s'", tokenizer_pre.c_str()));
+ LLAMA_LOG_WARN("%s: missing or unrecognized pre-tokenizer type, using: 'default'\n", __func__);
2024-05-23 21:16:26 +00:00
+ vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
}
} else {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;