text-generation-inference
Revamp medusa implementation so that every model can benefit.
#1588
Merged

Revamp medusa implementation so that every model can benefit. #1588

Narsil merged 16 commits into main from fix_medusa
Narsil
Narsil [Tmp] Revamping medusa to make it orthogonal.
2446f3ec
Narsil Upgrade ALL the code.
ac419f5e
Narsil Small updates.
21b30722
Narsil Remove the old logic.
7a9998d4
Narsil Black.
64d38afa
Narsil Fix MPT, not sure about idefics.
f592df52
Narsil Fixing.
a0095b5b
Narsil Fix gemma + medusa.
ed95f198
Narsil Narsil changed the title [WIP] Revamp medusa implementation so that every model can benefit. Revamp medusa implementation so that every model can benefit. 2 years ago
OlivierDehaene
OlivierDehaene commented on 2024-02-26
Narsil Fix GPT2 detection.
680a52f2
Narsil Download safetensors directly.
c7793235
Narsil Remove dead file.
1445b951
Narsil Specify revision to force use safetensors files.
fa40801f
Narsil Fix .
e672f976
Narsil Narsil force pushed from 52c8e22f to e672f976 2 years ago
OlivierDehaene
OlivierDehaene dismissed these changes on 2024-02-26
Narsil Fixing revision for the medusa test.
bfec09ec
Narsil Narsil dismissed their stale review via bfec09ec 2 years ago
Narsil Forgot docker launcher.
915e5f08
Narsil Small fixes in the weights loading logic.
e69e68c8
Narsil Narsil merged bf700e7e into main 2 years ago
Narsil Narsil deleted the fix_medusa branch 2 years ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone