transformers
[cache refactor] Move all the caching logic to a per-layer approach
#39106
Merged

[cache refactor] Move all the caching logic to a per-layer approach #39106

manueldeprada
manueldeprada Squash for refactor: Replace monolithic cache classes with modular La…
1c3cbccc
manueldeprada manueldeprada requested a review from gante gante 1 year ago
HuggingFaceDocBuilderDev
manueldeprada fix quantized, add tests
04d7a0b1
manueldeprada manueldeprada force pushed from a2fe24c8 to 04d7a0b1 1 year ago
manueldeprada
manueldeprada remove CacheProcessorList
26c28af6
manueldeprada manueldeprada force pushed from d97a02da to 26c28af6 1 year ago
manueldeprada manueldeprada requested a review from ArthurZucker ArthurZucker 1 year ago
zucchini-nlp
zucchini-nlp commented on 2025-07-01
ArthurZucker
ArthurZucker commented on 2025-07-01
manueldeprada raushan review, arthur review
16a66240
manueldeprada manueldeprada requested a review from ArthurZucker ArthurZucker 1 year ago
manueldeprada
gante
gante commented on 2025-07-02
gante
manueldeprada joao review: minor things
aec9ccd6
manueldeprada manueldeprada force pushed from c1d93472 to 52ac7934 1 year ago
manueldeprada manueldeprada force pushed from 52ac7934 to d6843398 1 year ago
manueldeprada manueldeprada force pushed from d6843398 to 4c03e0f3 1 year ago
manueldeprada remove cache configs, make CacheLayer a mixin (joaos review)
e80c68a6
manueldeprada manueldeprada force pushed from 88089304 to e80c68a6 1 year ago
manueldeprada manueldeprada force pushed from 65fbce68 to a21b2f74 1 year ago
manueldeprada manueldeprada force pushed from a21b2f74 to c538ad75 1 year ago
manueldeprada manueldeprada force pushed from c538ad75 to 5dc5fb40 1 year ago
manueldeprada manueldeprada force pushed from 5dc5fb40 to a6b7562b 1 year ago
manueldeprada back to storage inside Cache()
27916bc2
manueldeprada manueldeprada force pushed from a6b7562b to 27916bc2 1 year ago
manueldeprada remove cachebase for decorator
fd83e14b
manueldeprada manueldeprada force pushed from 9e37266d to fd83e14b 1 year ago
manueldeprada manueldeprada force pushed from 6c0dd19f to 129119dd 1 year ago
manueldeprada no more __getattr__
c2004471
manueldeprada manueldeprada force pushed from 129119dd to c2004471 1 year ago
manueldeprada Merge branch 'main' of github.com:huggingface/transformers into cache…
f3275705
manueldeprada fix tests
5b1b1f17
manueldeprada manueldeprada force pushed from 295c8d8a to 5b1b1f17 1 year ago
gante
gante commented on 2025-07-10
manueldeprada manueldeprada force pushed from ba1d14ed to 5b1b1f17 1 year ago
manueldeprada joaos review except docs
58dbcfe2
manueldeprada manueldeprada force pushed from 2b4e7bf7 to 58dbcfe2 1 year ago
manueldeprada fix ast deprecations for python 3.14: replace node.n by node.value an…
0c6d2ff6
Cyrilvallez
Cyrilvallez commented on 2025-07-11
manueldeprada manueldeprada force pushed from b22b1179 to 87f399fd 1 year ago
manueldeprada Revert "back to storage inside Cache()"
6a77408a
manueldeprada manueldeprada force pushed from 87f399fd to 6a77408a 1 year ago
manueldeprada cyril review
13ec4a44
manueldeprada simplify cache export
7029a90c
manueldeprada Merge branch 'main' of github.com:huggingface/transformers into cache…
5fa99012
manueldeprada fix lfm2 cache
dd7458b5
manueldeprada HybridChunked to layer
dc08253c
manueldeprada manueldeprada force pushed from d434a998 to dc08253c 1 year ago
manueldeprada BC proxy object for cache.key_cache[i]=...
a9521249
manueldeprada reorder classes
dbbc4d51
manueldeprada bfff come on LFM2
4bb48fcf
manueldeprada manueldeprada requested a review from Cyrilvallez Cyrilvallez 1 year ago
manueldeprada better tests for hybrid and hybridChunked
00b1f96a
manueldeprada Merge branch 'main' of github.com:huggingface/transformers into cache…
def346eb
manueldeprada complete coverage for hybrid chunked caches (prefill chunking)
38e86034
manueldeprada reimplementing HybridChunked
34a3022f
Cyrilvallez
Cyrilvallez commented on 2025-07-15
ArthurZucker ArthurZucker removed review request from ArthurZucker ArthurZucker 1 year ago
manueldeprada cyril review
42226536
manueldeprada
manueldeprada Merge branch 'main' of github.com:huggingface/transformers into cache…
063459c5
manueldeprada manueldeprada force pushed from 51d1505b to 83a6016b 1 year ago
manueldeprada fix ci
1acc648f
manueldeprada manueldeprada force pushed from 83a6016b to 1acc648f 1 year ago
manueldeprada docs for cache refactor
ca39ffec
manueldeprada manueldeprada force pushed from bd4c9a41 to ca39ffec 1 year ago
manueldeprada docs
731d0b73
manueldeprada Merge branch 'main' of github.com:huggingface/transformers into cache…
574b8205
manueldeprada oopsie
a4794705
manueldeprada oopsie
9c0bdcc5
manueldeprada Merge branch 'main' into cache-refactor-1
04091ed8
manueldeprada Merge branch 'main' of github.com:huggingface/transformers into cache…
8642bb6b
manueldeprada Merge branch 'cache-refactor-1' of https://github.com/manueldeprada/t…
83968bd9
manueldeprada fix after merge
0c4700dc
Cyrilvallez
Cyrilvallez commented on 2025-07-21
Cyrilvallez Cyrilvallez changed the title [cache refactor] PR 1/3: layered caches with full backward compatibility [cache refactor] Move all the caching logic to a per-layer approach 1 year ago
manueldeprada cyril review
b3a35e92
ArthurZucker
ArthurZucker commented on 2025-07-22
manueldeprada arthur review
e4878ad0
manueldeprada Merge branch 'main' of github.com:huggingface/transformers into cache…
38fb99d4
manueldeprada opsie
8df15953
manueldeprada fix lfm2
ad65a02d
Cyrilvallez
Cyrilvallez commented on 2025-07-22
github-actions
manueldeprada opsie2
d9fbb044
Cyrilvallez
Cyrilvallez approved these changes on 2025-07-22
Cyrilvallez Cyrilvallez merged c338fd43 into main 1 year ago
tdoublep
Cyrilvallez
tdoublep
tdoublep
tdoublep
manueldeprada
manueldeprada
tdoublep
Kokunekosan
Cyrilvallez

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone