LlaMA-4 support (text only) (#321)

* llama4: WIP

* llama4: this seems to be working

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2025-04-10 09:05:21 +02:00
committed by GitHub
parent 5f44f4b3d0
commit 474435f58b
3 changed files with 295 additions and 21 deletions

View File

@@ -100,7 +100,12 @@ extern "C" {
LLAMA_VOCAB_PRE_TYPE_TEKKEN = 20,
LLAMA_VOCAB_PRE_TYPE_SMOLLM = 21,
LLAMA_VOCAB_PRE_TYPE_CODESHELL = 22,
LLAMA_VOCAB_PRE_TYPE_DEEPSEEK3_LLM = 23, //llama.cpp lists this as 28
LLAMA_VOCAB_PRE_TYPE_DEEPSEEK3_LLM = 28, //llama.cpp lists this as 28
LLAMA_VOCAB_PRE_TYPE_GPT4O = 29,
LLAMA_VOCAB_PRE_TYPE_SUPERBPE = 30,
LLAMA_VOCAB_PRE_TYPE_TRILLION = 31,
LLAMA_VOCAB_PRE_TYPE_BAILINGMOE = 32,
LLAMA_VOCAB_PRE_TYPE_LLAMA4 = 33,
};
// note: these values should be synchronized with ggml_rope