Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

merge #1186

Merged
merged 30 commits into from
Dec 3, 2024
Merged

merge #1186

merged 30 commits into from
Dec 3, 2024

Conversation

bghira
Copy link
Owner

@bghira bghira commented Dec 3, 2024

No description provided.

bghira and others added 30 commits November 30, 2024 09:46
flux: use sage attention if available
57ec923
update deepspeed call to ensure compliance with new transformers version
68c5a5b
configurator should offer option to enable SageAttention for user
c8fe4e5
add --attention_mechanism option for sageattention
ab02d12
flux: remove model-specific sageattention code
2a5e028
sageattention cannot be enabled concurrently to xformers
e6b1919
sageattention should overwrite sdpa at startup if enabled
72e7d9e
flux: remove more model-specific sageattn code
bbf20d0
deprecate --enable_xformers_memory_efficient_attention in favour of -…
a86c256 
…-attention_mechanism=xformers
hackish awful workaround for VAE decode in SageAttention
73c6f1b
add more sageattention API choices
d406e3a
SD 1.5/2.x fix for SageAttention decode
cb286c2
set attention mechanism to the default for tests
11b3cf1
add sageattention to OPTIONS doc, update recommendations in --help ou…
d1c227c 
…tput
kolors: enable vae decode hack for sageattention
0976b5e
add sageattention to the quickstart docs and a specific section under…
c98d0f7 
… "Precision" in the OPTIONS doc
sd3 skip_layer_guidance fix from upstream for num images per prompt > 1
8719e73
disable nf4 + sageattention
964e065
@bghira
Merge pull request #1182 from bghira/feature/sage-attention
de9f988 
Add SageAttention for substantial training speed-up
add --sageattention_usage and default to inference-only
9bda632
update documentation
96eb116
@bghira
Merge pull request #1183 from bghira/feature/sageattention-inference-…
9cd974e 
…only

SageAttention: make it inference-only by default
update configure.py to use inference-only sageattention
62f1034
add --gradient_checkpointing_interval and disable gradient checkpoint…
a140657 
…ing before validations run for extra inference performance boost
remove print message
a4e5393
configurator should ask about gradient checkpointing interval
2ba64ab
configurator logic fix
96d477e
@bghira
Merge pull request #1184 from bghira/feature/gradient-checkpointing-s…
cd0644d 
…peedup

gradient checkpointing speed-up
add gradient checkpointing option to docs
8f8e1cf
@bghira
Merge pull request #1185 from bghira/feature/gradient-checkpointing-s…
aa42e05 
…peedup

add gradient checkpointing option to docs
@bghira bghira merged commit 07d9ea7 into release Dec 3, 2024
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@bghira