Perceiver
The Perceiver model was released in the previous version:
Perceiver
Eight new models are released as part of the Perceiver implementation:
PerceiverModel
,PerceiverForMaskedLM
,PerceiverForSequenceClassification
,PerceiverForImageClassificationLearned
,PerceiverForImageClassificationFourier
,PerceiverForImageClassificationConvProcessing
,PerceiverForOpticalFlow
,PerceiverForMultimodalAutoencoding
, in PyTorch.The Perceiver IO model was proposed in Perceiver IO: A General Architecture for Structured Inputs & Outputs by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch,
Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M.
Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
- Add Perceiver IO by @NielsRogge in #14487
Compatible checkpoints can be found on the hub: https://huggingface.co/models?other=perceiver
Version v4.14.0 adds support for Perceiver in multiple pipelines, including the fill mask and sequence classification pipelines.
Keras model cards
The Keras push to hub callback now generates model cards when pushing to the model hub. Additionally to the callback, model cards will be generated by default by the model.push_to_hub() method.
- TF model cards by @Rocketknight1 in #14720
What's Changed
-
Fix : wrong link in the documentation (ConvBERT vs DistilBERT) by @Tikquuss in #14705
-
Fix doc examples: 'CausalLMOutput...' object has no attribute 'last_hidden_state' by @ydshieh in #14678
-
Fix doc examples: unexpected keyword argument by @ydshieh in #14689
-
[doc] document MoE model approach and current solutions by @stas00 in #14725
-
[Flax examples] remove dependancy on pytorch training args by @patil-suraj in #14636
-
Update bug-report.md by @patrickvonplaten in #14715
-
[Adafactor] Fix adafactor by @patrickvonplaten in #14713
-
Fix doc examples: modify config before super().init by @ydshieh in #14697
-
Improve documentation of some models by @NielsRogge in #14695
-
Skip Perceiver tests by @LysandreJik in #14745
-
Add ability to get a list of supported pipeline tasks by @codesue in #14732
-
Fix the perceiver docs by @LysandreJik in #14748
-
Swap TF and PT code inside two blocks by @LucienShui in #14742
-
Mention no images added to repository by @LysandreJik in #14738
-
Avoid using tf.tile in embeddings for TF models by @ydshieh in #14735
-
Change how to load config of XLNetLMHeadModel by @josutk in #14746
-
Improve perceiver by @NielsRogge in #14750
-
Make data shuffling in
run_clm_flax.py
respect global seed by @bminixhofer in #13410 -
Adding support for multiple mask tokens. by @Narsil in #14716
-
Fix broken links to distillation on index page of documentation by @amitness in #14722
-
[doc] performance: groups of operations by compute-intensity by @stas00 in #14757
-
Fix preprocess_function in run_summarization_flax.py by @ydshieh in #14769
-
Update Perceiver code examples by @NielsRogge in #14783
New Contributors
- @Tikquuss made their first contribution in #14705
- @codesue made their first contribution in #14732
- @LucienShui made their first contribution in #14742
- @josutk made their first contribution in #14746
- @amitness made their first contribution in #14722
Full Changelog: v4.13.0...v4.14.0