ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

mtmd : Support Pixtral 12B (#13065)

Browse source 
* add pixtral text model (vision is wip)

* cgraph ok, just missing 2D RoPE

* fix bad rebase

* first working version

* fix problem with img_break token

* support dynamic image size

* update docs

* update test script
This commit is contained in:
Xuan-Son Nguyen 2025-04-23 20:21:59 +02:00 • committed by GitHub
parent eb1776b15a
commit ecda2ec4b3
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
14 changed files with 643 additions and 31 deletions
Download patch file Download diff file Expand all files Collapse all files

3
src/llama-vocab.cpp
View file

@ -1506,7 +1506,8 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
tokenizer_pre == "llama3" ||
tokenizer_pre == "llama-v3" ||
tokenizer_pre == "llama-bpe"||
tokenizer_pre == "falcon3") {
tokenizer_pre == "falcon3" ||
tokenizer_pre == "pixtral") {
pre_type = LLAMA_VOCAB_PRE_TYPE_LLAMA3;
ignore_merges = true;
add_bos = true;