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

llama : support models without vocabulary (#5798)

Browse source 
* additional methods to read model and ctx parameters

* vocab size as a part of a model metadata

* models without vocabulary, convert.py part

* models without vocabulary, llama.cpp part

* PR clean up

* converter scrypt fixes

* llama_vocab_type update (renamed the new key)

* pr review fixes

* revert function renaming

* one more NoVocab assert
This commit is contained in:
Michael Podvitskiy 2024-03-14 17:21:56 +01:00 • committed by GitHub
parent 044ec4b2a5
commit 69ff61397d
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 142 additions and 88 deletions
Download patch file Download diff file Expand all files Collapse all files

7
llama.h
View file

@ -59,9 +59,10 @@ extern "C" {
typedef int32_t llama_seq_id;
enum llama_vocab_type {
LLAMA_VOCAB_TYPE_SPM = 0, // SentencePiece
LLAMA_VOCAB_TYPE_BPE = 1, // Byte Pair Encoding
LLAMA_VOCAB_TYPE_WPM = 2, // WordPiece
LLAMA_VOCAB_TYPE_NONE = 0, // For models without vocab
LLAMA_VOCAB_TYPE_SPM = 1, // SentencePiece
LLAMA_VOCAB_TYPE_BPE = 2, // Byte Pair Encoding
LLAMA_VOCAB_TYPE_WPM = 3, // WordPiece
};
// note: these values should be synchronized with ggml_rope