idiap/fast-transformers

Issues

• Cuda version

  #117 opened 2 years ago by jiaji-huang
  2

• ERROR: Could not build wheels for pytorch-fast-transformers, which is required to install pyproject.toml-based projects

  #128 opened a year ago by ouusan
  12

• TypeError: canonicalize_version() got an unexpected keyword argument 'strip_trailing_zero'

  #132 opened 7 months ago by luispintoc
  0

• Full Attention does not sum to 1

  #131 opened 8 months ago by yourj4m
  1

• Can't officially save Linear Attention model

  #114 opened 3 years ago by maulberto3
  2

• Speed of linear attention slower than the attention implemented in pytorch

  #130 opened 9 months ago by yzeng58
  0

• [WinError 2] The system cannot find the file specified: build_ext

  #129 opened a year ago by cliffordkleinsr
  1

• Error about `causal_product_cpu.cpython-38-darwin.so` on Mac

  #124 opened 2 years ago by XiaoqZhang
  2

• installation error

  #92 opened 4 years ago by davidliujiafeng
  6

• ImportError

  #127 opened a year ago by PaulaTeeuwen
  1

• causal-linear do not use attn_mask ?

  #105 opened 3 years ago by davidliujiafeng
  1

• `.causal_product_cuda` missing in pip installed version on linux

  #125 opened 2 years ago by Jackl-o-o-l
  4

• Provenance of algorithms

  #126 opened 2 years ago by taibai123abc
  0

• Got different result for the same batch

  #123 opened 2 years ago by gaoshan2006
  1

• Installing error on linux

  #112 opened 3 years ago by xxmlala
  4

• Windows installation - building wheel error

  #121 opened 2 years ago by BenoitDalFerro
  27

• Windows installation - Building wheel

  #106 opened 2 years ago by MaximeHoude
  3

• TransformerDecoderBuilder: decoder only self attention example

  #81 opened 4 years ago by patdflynn
  3

• Detailed implementation of `clustered_sparse_dot_product`

  #120 opened 2 years ago by HanielF
  0

• pip install and c++ compilation error, then name 'compute_hashes_cuda' is not defined

  #89 opened 4 years ago by nikjetchev
  3

• Understanding how to define key, query and value for the cross attention calculation

  #119 opened 2 years ago by neuronphysics
  0

• Example for NLP

  #118 opened 2 years ago by Bachstelze
  0

• Training Language Model

  #107 opened 3 years ago by lucasnfe
  2

• Speed of recurrent model

  #116 opened 2 years ago by mads-oestergaard
  2

• can offer built code for linus?

  #115 opened 2 years ago by li-car-fei
  0

• Runtime error on causal_product_cpu on GCC/G++ 11

  #110 opened 3 years ago by lsisoft
  3

• Any decoder example?

  #113 opened 3 years ago by ahmedraza1996
  1

• Installation failed on Windows

  #97 opened 3 years ago by WRKULOL
  2

• Parallel complexity of Linear Attention is O(N)?

  #108 opened 3 years ago by haozheji
  1

• Casual attention is cheating by looking in the future

  #111 opened 3 years ago by jogardi
  1

• how causal mask constructed in training batch model with linear causal attention?

  #109 opened 3 years ago by Howuhh
  0

• Huggingface Bert vs. Fast Transformer full attention

  #100 opened 3 years ago by lipmem
  9

• For recurrent models, are positional embeddings required?

  #102 opened 4 years ago by rongcuid
  5

• Quick start raise a ModuleNotFoundError

  #99 opened 4 years ago by CaoYiqingT
  2

• CUDA error: CUBLAS_STATUS_INVALID_VALUE

  #104 opened 4 years ago by huu4ontocord
  1

• layernorm eps is not copied properly when cloning HF_Bert

  #103 opened 4 years ago by huu4ontocord
  2

• Mask and QK not of the same shape ?

  #101 opened 4 years ago by Baldwin-disso
  1

• local_dot_product_cuda fails when queries and keys have different lengths

  #98 opened 4 years ago by tridao
  0

• Can't import causal_product_cuda

  #96 opened 4 years ago by 15805383399
  1

• support of cluster attention

  #95 opened 4 years ago by TianhaoFu
  1

• TypeError: forward() missing 3 required positional arguments: 'attn_mask', 'query_lengths', and 'key_lengths'

  #94 opened 4 years ago by TianhaoFu
  1

• ModuleNotFoundError: No module named 'aggregate.aggregate_cpu'

  #93 opened 4 years ago by TianhaoFu
  2

• Question over cuda implementation of causal product (forward)

  #91 opened 4 years ago by thomasw21
  1

• Make fast-transformers JIT Compilable

  #88 opened 4 years ago by AndriyMulyar
  1

• [FAVOR & friends] Orthogonal random matrix not uniformly drawn

  #87 opened 4 years ago by blefaudeux
  3

• Implementing clustering function of clustered_attention with Python.

  #85 opened 4 years ago by mHsuann
  2

• CUDA version and CausalDotProduct time

  #83 opened 4 years ago by caffeinetoomuch
  4

• Tips and tricks for training linear_att

  #84 opened 4 years ago by gaceladri
  7

• Where is the sum operation of KV?

  #82 opened 4 years ago by Yogurt928
  3

• Queries scaling is not consistent for recurrent wrappers

  #80 opened 4 years ago by hadaev8
  3