Clip floating point constants to bf16 range to avoid inf conversion#20562

Closed

sangeethabal wants to merge 26 commits intohuggingface:mainfrom

sangeethabal:add_b16_support_for_attnmask_-1e4

Commits on Dec 2, 2022

Clip floating point constants to either bf16 or fp16 range to avoid inf conversion
EC2 Default User
committed

Commits on Dec 5, 2022

Fix torch device issues (#20584)

ydshieh
authored and
EC2 Default User
committed
Fix flax GPT-J-6B linking model in tests (#20556)

JuanFKurucz
authored and
EC2 Default User
committed
[Vision] fix small nit on `BeitDropPath` layers (#20587)

younesbelkada
authored and
EC2 Default User
committed
Fix repo consistency

sgugger
authored and
EC2 Default User
committed
Install `natten` with CUDA version (#20546)

ydshieh
authored and
EC2 Default User
committed
Add entries to `FEATURE_EXTRACTOR_MAPPING_NAMES` (#20551)

ydshieh
authored and
EC2 Default User
committed
Cleanup some config attributes (#20554)

ydshieh
authored and
EC2 Default User
committed
[Whisper] Move decoder id method to tokenizer (#20589)

sanchit-gandhi
authored and
EC2 Default User
committed
Add `require_torch` to 2 pipeline tests (#20585)

ydshieh
authored and
EC2 Default User
committed
Install `tensorflow_probability` for TF pipeline CI (#20586)

ydshieh
authored and
EC2 Default User
committed
Add BioGPT (#20420)

kamalkraj
authored and
EC2 Default User
committed
Ci-whisper-asr (#20588)

ArthurZucker
authored and
EC2 Default User
committed
cross platform from_pretrained (#20538)

authored and
EC2 Default User
committed
Make convert_to_onnx runable as script again (#20009)

mcernusca
authored and
EC2 Default User
committed
ESM openfold_utils type hints (#20544)

authored and
EC2 Default User
committed
Add RemBERT ONNX config (#20520)

authored and
EC2 Default User
committed
Fix link to Swin Model contributor novice03 (#20557)

JuanFKurucz
authored and
EC2 Default User
committed
Fix link to swin transformers v2 microsoft model (#20558)

JuanFKurucz
authored and
EC2 Default User
committed
Fix link to table transformer detection microsoft model (#20560)

JuanFKurucz
authored and
EC2 Default User
committed
clean up unused `classifier_dropout` in config (#20596)

ydshieh
authored and
EC2 Default User
committed
Fix whisper and speech to text doc (#20595)

ArthurZucker
authored and
EC2 Default User
committed
Replace `set-output` by `$GITHUB_OUTPUT` (#20547)

ydshieh
authored and
EC2 Default User
committed
[Vision] `.to` function for ImageProcessors (#20536)

authored and
EC2 Default User
committed
[Whisper] Fix decoder ids methods (#20599)

sanchit-gandhi
authored and
EC2 Default User
committed
Add-whisper-conversion (#20600)

ArthurZucker
authored and
EC2 Default User
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Clip floating point constants to bf16 range to avoid inf conversion#20562

Clip floating point constants to bf16 range to avoid inf conversion#20562
sangeethabal wants to merge 26 commits intohuggingface:mainfrom
sangeethabal:add_b16_support_for_attnmask_-1e4

Commits on Dec 2, 2022

Clip floating point constants to either bf16 or fp16 range to avoid inf conversion

Commits on Dec 5, 2022

Fix torch device issues (#20584)

Fix flax GPT-J-6B linking model in tests (#20556)

[Vision] fix small nit on `BeitDropPath` layers (#20587)

Fix repo consistency

Install `natten` with CUDA version (#20546)

Add entries to `FEATURE_EXTRACTOR_MAPPING_NAMES` (#20551)

Cleanup some config attributes (#20554)

[Whisper] Move decoder id method to tokenizer (#20589)

Add `require_torch` to 2 pipeline tests (#20585)

Install `tensorflow_probability` for TF pipeline CI (#20586)

Add BioGPT (#20420)

Ci-whisper-asr (#20588)

cross platform from_pretrained (#20538)

Make convert_to_onnx runable as script again (#20009)

ESM openfold_utils type hints (#20544)

Add RemBERT ONNX config (#20520)

Fix link to Swin Model contributor novice03 (#20557)

Fix link to swin transformers v2 microsoft model (#20558)

Fix link to table transformer detection microsoft model (#20560)

clean up unused `classifier_dropout` in config (#20596)

Fix whisper and speech to text doc (#20595)

Replace `set-output` by `$GITHUB_OUTPUT` (#20547)

[Vision] `.to` function for ImageProcessors (#20536)

[Whisper] Fix decoder ids methods (#20599)

Add-whisper-conversion (#20600)