Add Photon model and pipeline support - #12456

Merged

sayakpaul merged 52 commits into

huggingface:mainfrom

Photoroom:photon

Oct 21, 2025

Commits on Oct 21, 2025

Add Photon model and pipeline support
davidb
authored and
DavidBert
committed
c46a649
just store the T5Gemma encoder
davidb
authored and
DavidBert
committed
8e78a99
enhance_vae_properties if vae is provided only
davidb
authored and
DavidBert
committed
3aeada7
remove autocast for text encoder forwad
davidb
authored and
DavidBert
committed
d5c3272
BF16 example
david-PHR
authored and
DavidBert
committed
234c5e3
conditioned CFG
davidb
authored and
DavidBert
committed
49528a4
remove enhance vae and use vae.config directly when possible
davidb
authored and
DavidBert
committed
0a4183c
move PhotonAttnProcessor2_0 in transformer_photon
davidb
authored and
DavidBert
committed
7f1199b
remove einops dependency and now inherits from AttentionMixin
davidb
authored and
DavidBert
committed
cabde58
unify the structure of the forward block
davidb
authored and
DavidBert
committed
bca1f7c
update doc
davidb
authored and
DavidBert
committed
cdfa636
update doc
davidb
authored and
DavidBert
committed
03a7df3
fix T5Gemma loading from hub
davidb
authored and
DavidBert
committed
c248835
fix timestep shift
davidb
authored and
DavidBert
committed
94f469a
remove lora support from doc
davidb
authored and
DavidBert
committed
ec2381e
Rename EmbedND for PhotoEmbedND
DavidBert
committed
b445016
remove modulation dataclass
DavidBert
committed
7ae1af9
put _attn_forward and _ffn_forward logic in PhotonBlock's forward
DavidBert
committed
a8216f7
renam LastLayer for FinalLayer
DavidBert
committed
73261c8
remove lora related code
DavidBert
committed
f8f45fd
rename vae_spatial_compression_ratio for vae_scale_factor
DavidBert
committed
e65de5f
support prompt_embeds in call
DavidBert
committed
b765031
move xattention conditionning out computation out of the denoising loop
DavidBert
committed
87cd6d2
add negative prompts
DavidBert
committed
1b61bb2
Use _import_structure for lazy loading
DavidBert
committed
9ca25fa
make quality + style
DavidBert
committed
98ed747
add pipeline test + corresponding fixes
DavidBert
committed
9819ff1
utility function that determines the default resolution given the VAE
DavidBert
committed
77b4f8f
Refactor PhotonAttention to match Flux pattern
DavidBert
committed
19f9c47
built-in RMSNorm
DavidBert
committed
cd78032
Revert accidental .gitignore change
DavidBert
committed
6987240
parameter names match the standard diffusers conventions
DavidBert
committed
c92ee55
renaming and remove unecessary attributes setting
DavidBert
committed
4f74d94
Update docs/source/en/api/pipelines/photon.md
DavidBert
and
stevhliu
committed
d219e8c
quantization example
DavidBert
committed
5d57f44
added doc to toctree
DavidBert
committed
5270316
Update docs/source/en/api/pipelines/photon.md
DavidBert
and
stevhliu
committed
f7f516f
Update docs/source/en/api/pipelines/photon.md
DavidBert
and
stevhliu
committed
fb98a3a
Update docs/source/en/api/pipelines/photon.md
DavidBert
and
stevhliu
committed
f891323
use dispatch_attention_fn for multiple attention backend support
DavidBert
committed
5fcb8e6
naming changes
DavidBert
committed
23392b0
make fix copy
DavidBert
committed
1f88313
Update docs/source/en/api/pipelines/photon.md
DavidBert
and
dg845
committed
167030c
Add PhotonTransformer2DModel to TYPE_CHECKING imports
DavidBert
committed
8b551d2
make fix-copies
DavidBert
committed
7f6bb8a
Use Tuple instead of tuple
DavidBert
and
dg845
committed
4264606
restrict the version of transformers
DavidBert
and
dg845
committed
756fe95
Update tests/pipelines/photon/test_pipeline_photon.py
DavidBert
and
dg845
committed
4402b41
Update tests/pipelines/photon/test_pipeline_photon.py
DavidBert
and
dg845
committed
3b953ff
change | for Optional
DavidBert
committed
fda1f78
fix nits.
sayakpaul
authored and
DavidBert
committed
3e2d292
use typing Dict
DavidBert
committed
803d0d1