transformers
[CB] [Major] Upgrade the cache to support different attention types
#47809
Open

[CB] [Major] Upgrade the cache to support different attention types #47809

remi-or wants to merge 32 commits into main from cb-cache-upgrade
remi-or
remi-or Consolidate TP
b0bc55dd
remi-or rename
1a515c8a
remi-or The most breaking commit ever
22b627fa
remi-or Rename attributes in cfg
751c59da
remi-or Disable compile around attn
4b4ad8d0
remi-or Nits
0240a9e0
remi-or Cahges to make mask more efficient
08a7a6e8
remi-or Rename nits
3e5784a0
remi-or Init fixes
4486df35
remi-or Further integration
f9951575
remi-or Fix cache allocators
c43a8edc
remi-or Nit
9445e44f
remi-or Sizing
8d2471a2
remi-or Add cache pool
feaaf07f
remi-or Better and more optimized cache pool
f1de8a4c
remi-or Block buggy path
c550e06e
remi-or Add forking
caeef1a9
remi-or De-duplication and prefix sharing
2e0d6ad2
remi-or Offloading is back
50aee742
remi-or Restore block table
9cc752ea
remi-or Update overall
a5ec4d17
remi-or Block table part 2
11713d36
remi-or Test
ba7698d7
remi-or Fixes
ce5fe45a
remi-or Rename
eb0f4f7f
remi-or Tests 1/n
032bda56
remi-or Test 2/n
f539e4e7
remi-or More renaming
acdc0b18
remi-or Rename back
8db0c7c7
remi-or More test fixes
30eb11aa
remi-or remi-or force pushed from 6142eefc to 30eb11aa 5 days ago
remi-or Mb
730ba388
HuggingFaceDocBuilderDev
remi-or Merge branch 'main' into cb-cache-upgrade
16e58590
github-actions

Login to write a write a comment.

Login via GitHub

Reviewers
No reviews
Assignees
No one assigned
Labels
Milestone