Giters
lucidrains
/
vector-quantize-pytorch
Vector (and Scalar) Quantization, in Pytorch
Geek Repo:
Geek Repo
Github PK Tool:
Github PK Tool
Stargazers:
2009
Watchers:
30
Issues:
97
Forks:
170
lucidrains/vector-quantize-pytorch Issues
Import error of LatentQuantize
Updated
a month ago
Comments count
1
Possible bug in LFQ distance calculation
Closed
a month ago
Comments count
3
Modernizing the setup
Closed
a month ago
Comments count
2
Missing dependencies for running examples
Closed
a month ago
accelerate launch test.py error
Closed
a month ago
Comments count
1
Potential Bug for Lower codebook Dimension/ get_output_from_indices (improved VQGAN)
Closed
2 months ago
Comments count
2
`bfloat16` cannot utilize some codes
Closed
2 months ago
Comments count
6
RuntimeError: mat1 and mat2 must have the same dtype, but got Float and BFloat16
Closed
2 months ago
Comments count
5
How to recover quantized from the GroupedResidualVQ indices?
Closed
3 months ago
How to train VectorQuantize codebook?
Updated
3 months ago
Examples Codebook Utilization does not generalize
Updated
3 months ago
DDP hang when using vector-quantize-pytorch within torchrun
Closed
3 months ago
Comments count
6
Gumbel max trick does not seem to make sense in here
Updated
3 months ago
FSQ-indices of codes out of range
Closed
3 months ago
Comments count
5
LFQ loss is negative?
Updated
3 months ago
Comments count
2
nan encountered in `cdist()`
Closed
4 months ago
Comments count
2
Synchronization Bug in Distributed Environments
Closed
5 months ago
Comments count
5
How to freeze codebook ?
Closed
10 months ago
Comments count
9
How to get the original code by saved codebook and index
Updated
4 months ago
Comments count
1
About requirements
Closed
4 months ago
Comments count
1
How to Train Residual LFQ
Updated
4 months ago
learnable_codebook and in-place optimization
Updated
5 months ago
Wrong implementation on residual fsq
Closed
5 months ago
Comments count
2
FSQ implementations' bug
Closed
6 months ago
Comments count
4
ResidualVQ: the freeze_codebook in VectorQuantize is not implemented
Closed
6 months ago
Comments count
1
Conducting experiments on ImageNet using the implementation of FSQ from this repository.
Closed
6 months ago
Comments count
25
Is it possible to use Cosine Distance in Residual-VQ?
Updated
6 months ago
Comments count
1
Perhaps wrong formula in finite_scalar_quantization.py
Closed
6 months ago
Comments count
1
Multi headed VQ get_codes_from_indices doesnt return a correct output
Closed
7 months ago
Comments count
5
How to use vq loss to update codebook ?
Closed
7 months ago
Comments count
4
indices to codes doesn't work as expected for batched sequences
Closed
7 months ago
Comments count
5
[feature] advice for add 'Additive VQ'
Closed
8 months ago
Comments count
3
Implementation of LFQ
Closed
8 months ago
Comments count
10
PyTorch Python 3.8 CI failures caused by `list[int]`
Closed
8 months ago
Comments count
3
[feature] Finite Scalar Quantization: VQ-VAE Made Simple
Closed
8 months ago
Comments count
6
Cannot reconstruct Quantized from trained codeblock and indices
Closed
8 months ago
Comments count
2
Examples issue on how to flatten the image-like input.
Closed
8 months ago
Comments count
2
commitment loss too large and how to choose codebook dim
Updated
9 months ago
Decode function
Closed
9 months ago
Comments count
4
Allow for sample_codebook_temp to be annealed?
Updated
10 months ago
Comments count
1
K-means initialization does not works
Closed
10 months ago
Comments count
2
How about ChamferDistance instead of cdist to calculate KNN?
Updated
10 months ago
Comments count
2
Support perplexity to measure index collapse
Updated
10 months ago
Comments count
3
Training when input to be be quantized contains padding
Closed
a year ago
Comments count
4
Hi,I meet some debug when i inference the model
Updated
a year ago
Comments count
3
Gradient of gumbel straight through multiplier was not correct.
Closed
a year ago
Comments count
3
Crash on Mac M1/M2 chip when using MPS support
Closed
a year ago
Comments count
3
Memory leak on 1.6.14
Closed
a year ago
Comments count
9
Why L2 Normed l2 distance is cosine?
Closed
a year ago
Comments count
1
Bug with distributed training and Cosine codebook
Closed
a year ago
Comments count
1
Previous
Next