lucidrains routing-transformer issues

lucidrains / routing-transformer

Fully featured implementation of Routing Transformer

MIT License

281 stars 29 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

How to reconstruct the full attention matrix?

#33 FarzanT opened 1 year ago
2
ONNX export hangs

#32 genolve closed 1 year ago
1
Compound words

#31 wingedsheep closed 2 years ago
0
Enquiry about BPC Calculation

#30 ShiweiLiuFdu opened 2 years ago
0
大佬可否细讲讲KmeansAttention？原理和动机

#29 guotong1988 opened 3 years ago
0
Could you please illustrate more about KmeansAttention? Thank you very much!

#28 guotong1988 opened 3 years ago
0
input_mask behavior

#27 AliOskooeiTR opened 3 years ago
0
TPU support

#26 abcp4 opened 3 years ago
0
recieves_context cause tensor mismatch error

#25 WeForgot closed 3 years ago
1
results on wiki103 or enwiki8

#24 yelongshen opened 3 years ago
0
README Typo

#23 rainmaker712 closed 3 years ago
1
Music Routing Transformer Colab

#22 asigalov61 opened 3 years ago
1
Building and training a RoutingTransformerEncDec from pre-trained RoutingTransformerLMs

#21 AliOskooeiTR closed 3 years ago
7
LM slower than the encoder-decoder with the same depth and max_seq_len, window size

#20 AliOskooeiTR opened 3 years ago
3
Issue about input shape

#19 Henrykwokkk closed 3 years ago
1
Usage for image generation

#18 Baran-phys closed 3 years ago
11
Sequence length limited

#17 Henrykwokkk closed 3 years ago
14
Report an error in the training example of enwik8_simple

#16 JunZhan2000 closed 3 years ago
1
When running the RoutingTransformerLM's example, there is an error that the tensor dimension does not match

#15 JunZhan2000 closed 3 years ago
1
Add ReZero and ScaleNorm support

#14 tomweingarten closed 3 years ago
7
MoE doesn't work with reversible layers

#13 tomweingarten closed 3 years ago
2
Batch size 1

#12 matthew-jurewicz closed 3 years ago
2
Long dependencies

#11 matthew-jurewicz opened 4 years ago
2
normalize queries and keys before dot product

#10 lucidrains closed 4 years ago
0
Why doesn't AutoregressiveWrapper sum the encoder aux loss?

#9 tomweingarten closed 4 years ago
8
What does autoregressive mean?

#8 matthew-jurewicz closed 4 years ago
8
One-hot encoded input?

#7 matthew-jurewicz closed 4 years ago
4
Missing key(s) in state_dict

#6 epetros closed 4 years ago
4
AutoregressiveWrapper expects different input lengths based on type

#5 tomweingarten closed 4 years ago
3
Encoder-decoder fails at KMeans attention

#4 tomweingarten closed 4 years ago
16
use_evonorm no longer supported in PKM

#3 tomweingarten closed 4 years ago
1
Fix top_p to define threshold similarly to top_k and not garble output.

#2 tomweingarten closed 4 years ago
1
top_p returns wrong values and re-orders the data

#1 tomweingarten closed 4 years ago
7