Search results

Html

Status: Open.

#114 In microsoft/torchscale;

· Sifat1312 opened on Jun 15, 2025

torchscale 0.3.0 does not include LongNet

Status: Open.

#113 In microsoft/torchscale;

· tsy935 opened on Dec 9, 2024

train.py: error: unrecognized arguments: --flash-attention --segment-length [2048,4096] --dilated-ratio [1,2]

Status: Open.

#112 In microsoft/torchscale;

· github2657529567 opened on Nov 8, 2024

No module named 'sentencepiece'

Status: Open.

#111 In microsoft/torchscale;

· github2657529567 opened on Nov 8, 2024

Question about LongNet attention map overlap

Status: Open.

#108 In microsoft/torchscale;

· RmZeta2718 opened on May 27, 2024

Different batch sizes lead to different evalution results for LongVIT

Status: Open.

#107 In microsoft/torchscale;

· HHHedo opened on May 14, 2024

How to test the model

Status: Open.

#106 In microsoft/torchscale;

· ReloJeffrey opened on May 10, 2024

pip error

Status: Open.

#105 In microsoft/torchscale;

· wanghaoran-ucas opened on Apr 28, 2024

Where is the offset implemented in Multi-head dilated attention ?

Status: Open.

#104 In microsoft/torchscale;

· AshStuff opened on Apr 20, 2024

can't use longvit

Status: Open.

#103 In microsoft/torchscale;

· abebe9849 opened on Apr 18, 2024

Question about learnable segment lengths and dilation rates

Status: Open.

#102 In microsoft/torchscale;

· benrousePUC opened on Apr 10, 2024

How to use retention in RetNet for cross-attention?

Status: Open.

#101 In microsoft/torchscale;

· yxchng opened on Mar 23, 2024