GitHub · Where software is built

Labels Milestones New issue

Truncation of sequences that are beyond the model's maximum length

feat/tokenization

#359

· MootezSaaD opened

on Jan 14, 2024

Add suggested PyTorch LLM optimizations

feat/generation

#356

· danieldk opened

on Dec 1, 2023

Move the old Falcon architecuture to the extras/addons pacakage

type/maintenance

#355

· shadeMe opened

on Oct 19, 2023

·

Add support for attention sinks

#350

· danieldk opened

on Oct 4, 2023

·

Support DeBERTa v2/3

#348

· danieldk opened

on Oct 3, 2023

·

Add a an extras/contrib package

type/maintenance

#347

· danieldk opened

on Oct 3, 2023

·

Expose more outputs through the <code>Generator</code> interface

feat/generation

#345

· danieldk opened

on Oct 3, 2023

·

Make <code>QkvMode</code> ADT-like

type/maintenance

#344

· danieldk opened

on Oct 3, 2023

·

Convert QKV projection splitting methods into Torch modules

type/maintenance

#343

· danieldk opened

on Oct 3, 2023

·

Option to only return the last hidden layer output from models

#342

· danieldk opened

on Oct 3, 2023

·

Add support for Mistral

#341

· danieldk opened

on Oct 3, 2023

·

Support for Encoder-Decoder-style architectures

#340

· bilelomrani1 opened

on Oct 2, 2023

·