Exllama v2 #1211

Narsil merged 11 commits into main from tmp_exllamav2
Narsil
Narsil Narsil force pushed from 92804866 to 692fba1d 2 years ago
Narsil Narsil changed the title [WIP] Just showcasing some modification for exllama v2 Exllama v2 2 years ago
Narsil Narsil requested a review from OlivierDehaene OlivierDehaene 2 years ago
flozi00 draft exllamav2, first commit
0ae4523c
flozi00 build in docker
a02f6839
flozi00 use exllamav2QuantLinear instead of exllama1
f96d9974
Narsil Update exllamav2 (illegal address issued)
024bdb01
Narsil Adding scratch space.
fb64ce10
Fixing exllamav2.
a61f4325
Narsil Exllamav2 functional.
ff515893
Narsil Fix imports.
b041bf15
Narsil Deactivating v2 for sharded.
a7ed31cf
Narsil Rebased.
d9dffb55
Narsil Narsil force pushed from c7664263 to d9dffb55 2 years ago
Narsil Keep exllamav1 for sharded flows.
86009e28
Narsil Narsil merged ed2a3f61 into main 2 years ago
Narsil Narsil deleted the tmp_exllamav2 branch 2 years ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone