-
Thanks for your great work!
In the paper, after the kv cache is quantized, a low-rank matrix is used to approximate the quantization error. I really want to know if this process needs training? S…
shhn1 updated
3 months ago
-
## 🚀 Feature
An incremental version of pca_lowrank - processing a mini-batch of samples in each iteration.
Similar to sklearn.decomposition.IncrementalPCA(), but in GPU.
## Motivation
The curr…
-
Hello, Prof. Zhang,
I can see you have modified part of the website.
However, please check [https://github.com/jwzhanggy/tinyBIG/blob/main/docs/examples/image/cifar10.md](url) and [https://githu…
-
## Fix the Op info test for `stft .. take_along_dim`
1. Find the lines 318 to 322 of [test_ops.py](test/test_ops.py) and remove
`stft` .. `take_along_dim` from `skip_list`
2. Run op_info test with…
qihqi updated
2 weeks ago
-
Hi,when i run the python2 ./vgg/lowrank_approx.py --model ./vgg/vgg_deploy.prototxt --config ./vgg/config.json --save_model ./vgg/vgg_lowrank_deploy.prototxt --weights ./vgg/models/vgg_finute.caffemo…
-
For example, the capabilities of 7b and 13b-lowrank.
-
### 🐛 Describe the bug
`torch.pca_lowrank` and `torch.svd_lowrank` does not work with automatic mixed precision, even if the inputs are 32 bit.
```python
import torch
x = torch.rand(1000, 3, d…
-
I succesfully compiled caffe version of the code but Iam getting error after executing below command
python2 lowrank_approx.py \
--model models_vgg/vgg_deploy.prototxt \
--config models_vgg/con…
-
Traceback (most recent call last):
File "/ossfs/workspace/sft/sft_all.py", line 161, in
train()
File "/ossfs/workspace/sft/sft_all.py", line 125, in train
Traceback (most recent call last…
-
I wonder if rather than this package defining `impute` and `impute_missing` it would be instead be worth implementing the interface provided by [`Impute.jl`](https://github.com/invenia/Impute.jl)?
…