baalajimaestro/ollama
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Update llama.cpp submodule to 5921b8f0 (#4731)

Browse source 
* update llama.cpp submodule to `5921b8f089d3b7bda86aac5a66825df6a6c10603`

* add patch
This commit is contained in:
Jeffrey Morgan 2024-05-30 16:20:22 -07:00 committed by GitHub
parent 0fc0cfc6d2
commit 22f5c12ced
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 22 additions and 25 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
llm/llama.cpp

@ -1 +1 @@
Subproject commit 74f33adf5f8b20b08fc5a6aa17ce081abe86ef2f
Subproject commit 5921b8f089d3b7bda86aac5a66825df6a6c10603

45
llm/patches/05-default-pretokenizer.diff
View file

@ -1,35 +1,32 @@
From d02a06f3f45a09255ace8684a66590e06ce44605 Mon Sep 17 00:00:00 2001
From: Michael Yang <mxyng@pm.me>
Date: Thu, 23 May 2024 11:33:20 -0700
Subject: [PATCH] default pretokenizer on unrecognized type
---
llama.cpp | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)
diff --git a/llama.cpp b/llama.cpp
index 15c66077..af1aede3 100644
index 40d2ec2c..74f3ee9c 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -4504,9 +4504,6 @@ static void llm_load_vocab(
LLAMA_LOG_WARN("%s: ************************************ \n", __func__);
LLAMA_LOG_WARN("%s: \n", __func__);
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
- } else if (
- tokenizer_pre == "default") {
@@ -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__);
- vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
- } else if (
+ if (
tokenizer_pre == "default") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
} else if (
tokenizer_pre == "llama3" ||
tokenizer_pre == "llama-v3" ||
@@ -4553,7 +4550,7 @@ static void llm_load_vocab(
tokenizer_pre == "dbrx") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DBRX;
@@ -4703,7 +4694,8 @@ static void llm_load_vocab(
tokenizer_pre == "smaug-bpe") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_SMAUG;
} 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__);
+ vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
}
} else {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT;
--
2.45.1