explosion curated-transformers issues

explosion / curated-transformers

🤖 A PyTorch library of curated Transformer models and their composable components

MIT License

860 stars 34 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Backport: Fix activation lookup with Python 3.12.3 (#375)

#377 danieldk closed 4 months ago
0
Set version to 2.0.1

#376 danieldk closed 4 months ago
0
Fix activation lookup with Python 3.12.3

#375 danieldk closed 4 months ago
1
Finalize the API changes for 2.0

#374 danieldk closed 5 months ago
0
Set version to 2.0.0

#373 danieldk closed 5 months ago
0
Set version to 2.0.0.dev3, update curated-tokenizers dep to 2.0.0.dev0

#372 danieldk closed 5 months ago
0
Set version to 2.0.0.dev2

#371 danieldk closed 5 months ago
0
Add support for loading parameters in-place

#370 danieldk closed 5 months ago
0
Use small ELECTRA model for testing

#369 danieldk closed 5 months ago
0
Are there any Docker Images that are compatible with spacy==3.7 and CUDA <11.8?

#368 EY4L closed 5 months ago
1
Adjust two cross-tests for changes in HF transformers

#367 danieldk closed 5 months ago
0
Set version to 2.0.0.dev1

#366 danieldk closed 7 months ago
0
Set version to 1.3.1

#365 danieldk closed 7 months ago
0
Backport: Ensure that parameters are leaf nodes when loading a model

#364 danieldk closed 7 months ago
0
Set torch upper bound to <2.1.0

#363 danieldk closed 7 months ago
0
Ensure that parameters are leaf nodes when loading a model

#362 danieldk closed 7 months ago
0
Remove support for TorchScript tracing

#361 danieldk closed 7 months ago
0
Clear output of Torch SDPA for masked pieces

#360 danieldk closed 7 months ago
0
Truncation of sequences that are beyond the model's maximum length

#359 MootezSaaD opened 8 months ago
2
Added ELECTRA as a thin wrapper around BERT

#358 KennethEnevoldsen closed 5 months ago
9
Add `curated_transformers.__version__` and use it for doc generation

#357 danieldk closed 9 months ago
0
Add suggested PyTorch LLM optimizations

#356 danieldk opened 9 months ago
0
Move the old Falcon architecuture to the extras/addons pacakage

#355 shadeMe opened 11 months ago
0
Add support for file write/upload operations with `HfHubRepository`

#354 shadeMe closed 10 months ago
0
Bump the version to 2.0.0.dev0

#353 danieldk closed 11 months ago
0
AutoModel: let models check if the configuration is supported

#352 danieldk closed 11 months ago
0
Register models using `catalogue`

#351 danieldk closed 11 months ago
0
Add support for attention sinks

#350 danieldk opened 11 months ago
0
Add `AttentionScorer` abstraction

#349 shadeMe closed 11 months ago
0
Support DeBERTa v2/3

#348 danieldk opened 11 months ago
1
Add a an extras/contrib package

#347 danieldk opened 11 months ago
0
Register models/tokenizers through `catalogue`, so that the auto classes pick them up

#346 danieldk closed 5 months ago
2
Expose more outputs through the `Generator` interface

#345 danieldk opened 11 months ago
0
Make `QkvMode` ADT-like

#344 danieldk opened 11 months ago
0
Convert QKV projection splitting methods into Torch modules

#343 danieldk opened 11 months ago
0
Option to only return the last hidden layer output from models

#342 danieldk opened 11 months ago
0
Add support for Mistral

#341 danieldk opened 11 months ago
0
Support for Encoder-Decoder-style architectures

#340 bilelomrani1 opened 11 months ago
2
Update RTD links in Readme

#339 shadeMe closed 11 months ago
0
Set version to `1.3.0`

#338 shadeMe closed 11 months ago
0
Set version to `1.4.0.dev0`

#337 shadeMe closed 11 months ago
0
Add note to repo API on being experimental

#336 shadeMe closed 11 months ago
0
Remove unused model checkpoint context manager

#335 shadeMe closed 11 months ago
0
Restructure `util.serde`

#334 shadeMe closed 11 months ago
0
Add support for converting Curated Transformer configs to Hugging Face compatible configs

#333 shadeMe closed 10 months ago
0
Add support for converting Curated Transfomer state dicts Hugging Face compatible state dicts

#332 shadeMe closed 11 months ago
0
Add repository abstraction

#331 danieldk closed 11 months ago
0
Add a default `dtype` to model configs

#330 shadeMe closed 12 months ago
0
Pretrain from scratch and fine tune

#329 nimitpattanasri closed 12 months ago
1
Persist model config to aid serialization

#328 shadeMe closed 12 months ago
0