llama: update vendored code to commit 46e3556 (#8308)

This commit is contained in:
Jeffrey Morgan
2025-01-08 11:22:01 -08:00
committed by GitHub
parent 57f038ec7b
commit 1deafd8254
305 changed files with 16048 additions and 12926 deletions

View File

@@ -4,14 +4,14 @@ Date: Mon, 16 Sep 2024 15:53:13 -0700
Subject: [PATCH] pretokenizer
---
src/llama.cpp | 14 +++-----------
src/llama-model.cpp | 14 +++-----------
1 file changed, 3 insertions(+), 11 deletions(-)
diff --git a/src/llama.cpp b/src/llama.cpp
index abc1252e..626c3e3f 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -6400,16 +6400,7 @@ static void llm_load_vocab(
diff --git a/src/llama-model.cpp b/src/llama-model.cpp
index 405e0528..00b80c52 100644
--- a/src/llama-model.cpp
+++ b/src/llama-model.cpp
@@ -1249,16 +1249,7 @@ void llm_load_vocab(llama_model_loader & ml, llama_model & model) {
if (vocab.type == LLAMA_VOCAB_TYPE_BPE) {
vocab.tokenizer_add_space_prefix = false;
vocab.tokenizer_clean_spaces = true;
@@ -29,9 +29,9 @@ index abc1252e..626c3e3f 100644
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
} else if (
tokenizer_pre == "llama3" ||
@@ -6514,7 +6505,8 @@ static void llm_load_vocab(
tokenizer_pre == "minerva-7b") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_MINERVA;
@@ -1373,7 +1364,8 @@ void llm_load_vocab(llama_model_loader & ml, llama_model & model) {
tokenizer_pre == "megrez") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_QWEN2;
} 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__);