llama.cpp
Extend the support of T5 models with different encoder-decoder layers
#15909
Merged

Extend the support of T5 models with different encoder-decoder layers #15909

CISC merged 27 commits into ggml-org:master from DamonFool:t5-enhancement
DamonFool
DamonFool Extend the support of T5 models with different encoder-decoder layers
33163bfb
github-actions github-actions added python
DamonFool
CISC
DamonFool
CISC
DamonFool
CISC
CISC commented on 2025-09-10
DamonFool Update convert_hf_to_gguf.py
219eadab
DamonFool Update gguf-py/gguf/constants.py
2161c309
DamonFool Update gguf-py/gguf/gguf_writer.py
284ceb3d
DamonFool Update src/llama-arch.cpp
77f0f162
DamonFool Update src/llama-arch.h
7efe517a
DamonFool Update src/llama-model.cpp
12a909f1
DamonFool Update src/llama-model.cpp
634e5a96
DamonFool Update src/llama-model.cpp
ebef503a
DamonFool Update src/llama-model.cpp
0acda17a
DamonFool Update src/llama-hparams.h
19281fe1
DamonFool Update src/llama-model.cpp
5153072b
DamonFool Update src/llama-model.cpp
60821df6
DamonFool Update src/llama-model.cpp
de46320f
DamonFool Update src/llama-model.cpp
804a982d
DamonFool Update src/llama-model.cpp
92150875
DamonFool Update src/llama-model.cpp
11672692
DamonFool Update src/llama-model.cpp
678aa48a
DamonFool Update src/llama-model.cpp
d145ee13
DamonFool Update src/llama-model.cpp
ce90f806
DamonFool Update src/llama-model.cpp
01002dfb
DamonFool Update src/llama-model.cpp
3ee2193d
DamonFool Update src/llama-model.cpp
6cb51f2d
DamonFool Update src/llama-model.cpp
42f1fdba
CISC
DamonFool Update src/llama-model.cpp
69406502
DamonFool
CISC
ggerganov
ggerganov commented on 2025-09-10
DamonFool Rename n_dec_layer --> dec_n_layer
f16d8de5
DamonFool Adapt to cases when dec_n_layer > n_layer
84e5db4d
DamonFool
DamonFool commented on 2025-09-10
DamonFool
CISC
CISC approved these changes on 2025-09-10
CISC CISC merged 4f658855 into master 94 days ago
DamonFool
DamonFool DamonFool deleted the t5-enhancement branch 94 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone