transformers
[new model] Add zaya1 vl
#46011
Open

[new model] Add zaya1 vl #46011

JJJYmmm wants to merge 30 commits into huggingface:main from JJJYmmm:add_zaya1_vl
JJJYmmm
JJJYmmm zaya1 support
b35c5e08
JJJYmmm add test
d26fffc9
JJJYmmm update example
8191d397
JJJYmmm new config
c125ef31
JJJYmmm remove empty line
c90df6f3
JJJYmmm pass ci
b90759f1
JJJYmmm modify config, laguna-sytle rope
7e299992
JJJYmmm use existing cache
cf083aa1
JJJYmmm cca refine + use llama attn
69d09f3f
JJJYmmm use dict for 2d/4d mask
d936d54a
JJJYmmm optimize, reuse existing code
733e687c
JJJYmmm inherit from AfmoeForCausalLM,
eb7c8cc7
JJJYmmm checkpoint conversion
4d5bda4e
JJJYmmm align with official implement, check 74b conversion
f3e8e02c
JJJYmmm easier test
f4f206c5
tarekziade Merge branch 'main' into add_zaya1
059912d6
JJJYmmm remove mapping since we convert the ckpt
7c48ee10
JJJYmmm use default_swa_theta
498c2522
JJJYmmm update date
3d630612
JJJYmmm temp init
4d742969
JJJYmmm modular
d77d5d47
JJJYmmm better residual scaling
1c16fecb
JJJYmmm better cache
3f53fbca
JJJYmmm ops forget init again
dc7ac50d
JJJYmmm better naming
8be4b1ee
JJJYmmm llama decoderlayer
7bb5122a
JJJYmmm improve
b315ae07
JJJYmmm update date
0df3204d
JJJYmmm Add ZAYA1-VL model support
b53060fb
JJJYmmm improve
96c275f0
JJJYmmm JJJYmmm marked this pull request as draft 36 days ago
github-actions
github-actions

Login to write a write a comment.

Login via GitHub

Reviewers
No reviews
Assignees
No one assigned
Labels
Milestone