PR #344 Rework loading - SemanticDiff

Rework loading #344

OlivierDehaene merged 27 commits into main from rework_loading

Narsil requested a review from

OlivierDehaene 3 years ago

Narsil force pushed from c58c0875 to fc2571de 3 years ago

Narsil force pushed from 869f8620 to 5260436a 3 years ago

Narsil force pushed from 2405e060 to 5e88900b 3 years ago

Narsil force pushed from c41ba911 to 275fb0ec 3 years ago

Narsil force pushed from e81133c6 to 202de69b 3 years ago

Narsil changed the title ~~[WIP] Rework loading~~ Rework loading 3 years ago

Narsil marked this pull request as ready for review 3 years ago

Fused all commits for saner rebase..

15bf3d49

Black + ruff + T5 w0 quant.

2362a80a

Missing import.

5c2a0e45

Typo.

680f26d6

T5?

e36e42a3

Neox (non flash) port + kernel.

55045be4

M********

c471e46c

Green ?

165bb4b6

Fix PositionalRotary loads.

4e071bf2

Fix logic.

7fa79f02

Narsil force pushed from 202de69b to 7fa79f02 3 years ago

Fix rebase.

2a1ecf38

Fixing flash rw.

d083d57d

Large attention ?

daf59b05

Updating starcoder

644e0a65

Adding integration for neox NON flash.

877d4d4a

Fix regular flash

c5995652

Removing flash attention env

c6ac50e4

OlivierDehaene commented on 2023-06-07

Apply suggestions from code review

6ddcd158

Manual fixes.

b8bfb2a9

Update server/text_generation_server/models/custom_modeling/flash_rw_…

5c82dcd2

Fixing Falcon 40b

cc843878

Just ditch the non flash integration tests. They work, but seem to mess

f3388d29

Last fixes hopefully.

4170de1b

skip instead of comment

5e0a6ea1

black + cleanup

b027f5f1

add CARGO_REGISTRIES_CRATES_IO_PROTOCOL

c66648d9

warn on unused snapshot

f245aa0c

OlivierDehaene merged abd58ff8 into main 3 years ago

OlivierDehaene deleted the rework_loading branch 3 years ago

Reviewers

OlivierDehaene

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

text-generation-inference Rework loading #344 Merged

Rework loading #344

text-generation-inference
Rework loading
#344

Merged