Quang Dương Quỳnh

Pruning Swin transformer v2 MMPretrain: Index error

3

My scripts: ``` import torch from mmpretrain import get_model import torch_pruning as tp import torch.nn as nn from typing import Sequence from mmpretrain.models.utils import PatchMerging, WindowMSAV2 class SwinPatchMergingPruner(tp.BasePruningFunc): def prune_out_channels(self,...

Factorized Tensor slower than Neural Network Layer !!!

5

``` import tltorch import torch from torch.profiler import profile, record_function, ProfilerActivity data = torch.randn((4, 16), dtype=torch.float32) linear = torch.nn.Linear(16, 10) fact_linear = tltorch.FactorizedLinear.from_linear(linear, auto_tensorize=False, in_tensorized_features=(4, 4), out_tensorized_features=(2, 5), rank=0.1, factorization="tucker")...

HF version infer very slower than Original version?? How different between 2 version infer?

1

```bash python benchmarks/benchmark_generation_mamba_simple.py --model-name "AntonV/mamba2-130m-hf" --batch 1 --genlen 4096 --promptlen 600 ``` Output: Loading model AntonV/mamba2-130m-hf Number of parameters: 128989632 Prompt length: 600, generation length: 4096 AntonV/mamba2-130m-hf prompt processing +...

**Mesh Feature Embedding** is extract from character in the **Handle Predictor**, **But how to extract this feature embeding is not mention anywhere in your paper?** Thanks for your work.

parent_smpl = [-1, 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 9, 9, 12, 13, 14, 16, 17, 18, 19, 20, 21] # SMPL skeleton topology...