patrickvonplaten
Follow
Highlights
- Pro
Pinned
2,795 contributions in the last year
Contribution activity
February 2023
Created 65 commits in 5 repositories
Created 1 repository
- patrickvonplaten/UniPC Jupyter Notebook
Created a pull request in huggingface/transformers that received 6 comments
[Variant] Make sure variant files are not incorrectly deleted
What does this PR do? As discussed offline with @sgugger and @ArthurZucker this PR makes sure that only sharded weight files can be deleted to prev…
+27
−0
•
6
comments
Opened 25 other pull requests in 2 repositories
huggingface/diffusers
24
merged
- fix transformers naming
- Fix deprecation warning
- Add ddim inversion pix2pix
- [Pix2Pix] Add utility function
- Revert "[Pix2Pix0] Add utility function to get edit vector"
- [Pix2Pix0] Add utility function to get edit vector
- [Examples] Make sure EMA works with any device
- [Dummy imports] Add missing if else statements for SD]
- [Tests] Add MPS skip decorator
- Fix callback type hints - no optional function argument
- [Tests] Remove unnecessary tests
- [Versatile Diffusion] Fix tests
- [Docs] Fix ethical guidelines docs
- [Tests] Refactor push tests
- Correct fast tests
- Fast CPU tests should also run on main
- [Variant] Add "variant" as input kwarg so to have better UX when downloading no_ema or fp16 weights
- [Latent Upscaling] Remove unused noise
- fix pix2pix docs
- [Examples] Test all examples on CPU
- Replace flake8 with ruff and update black
- [Tests] Fix slow tests
- [Examples] Remove datasets important that is not needed
- [Loading] Better error message on missing keys
wl-zhao/UniPC
1
merged
Reviewed 75 pull requests in 5 repositories
huggingface/diffusers
25 pull requests
- add MultiDiffusionPanorama pipeline
- Add ddim inversion pix2pix
- controlling generation docs
- add: inversion to pix2pix zero docs.
- [Pipelines] Add a section on generating captions and embeddings for Pix2Pix Zero
- Torch2.0 scaled_dot_product_attention processor
- [Pix2Pix] Add utility function
- Attend and excite 2
- [Examples] Make sure EMA works with any device
- Add semantic guidance pipeline
- [SchedulingPNDM ] reset cur_model_output after each call
-
enable_model_cpu_offload - Fix 3-way merging with the checkpoint_merger community pipeline
- [Variant] Add "variant" as input kwarg so to have better UX when downloading no_ema or fp16 weights
-
[Utils] Adds
store()andrestore()methods to EMAModel - checkpointing_steps_total_limit->checkpoints_total_limit
- Music Spectrogram diffusion pipeline
- [Tests] Add MPS skip decorator
- Add option to set dtype in pipeline.to() method
- [Pipelines] Adds pix2pix zero
- [Docs] initial docs about KarrasDiffusionSchedulers
- Add attend and excite pipeline
- Textual inv make save log both steps
- schedulers add glide noising schedule
- unCLIP variant
- Some pull request reviews not shown.
huggingface/transformers
4 pull requests
huggingface/accelerate
2 pull requests
AttendAndExcite/Attend-and-Excite
1 pull request
huggingface/api-inference-community
1 pull request
Created an issue in huggingface/huggingface_hub that received 12 comments
Smarter cache for snapshot_download
Is your feature request related to a problem? Please describe.
For diffusers many model repositories share identical checkpoint files. Currently wh…
12
comments