-
**TL;DR:** Implementing block-sparse operations for faster matrix-multiplication.
Is this something worth adding to PyTorch?
Goals:
1. Faster matrix-multiplication by taking advantage of block-…
-
I am trying to figure out what to use as optimal kernel parameter for different architectures.
For example, it looks like blis is using 8x4 for Sandy Bridge, but 8x6 for Haswell. Why? What lead the…
-
[The format of the issue]
Paper name/title:
Project link:
Paper link:
Code link:
amusi updated
2 weeks ago
-
在自动驾驶领域，基于激光雷达 (LiDAR) 的3D物体检测和运动行为预测是一种普遍的方案。目前绝大部分关于激光雷达的物体检测算法都是基于单帧的。激光雷达的多帧时序数据，提供了…
-
For some reason, someone added an `embeddings` argument to `BaseModule` but didn't update the other modules to pass it.
Why are we passing embeddings as an argument? Those should be an attribute of…
-
Ever since LeelaZero was released, people have been wanting to use it to analyze their games, and it's become incredibly easy and popular to review with LeelaZero. Now, LeelaZero has become insanely s…
-
각 주제에 대해 세밀한 검토를 하기 전 우선 아이디어 중심으로 제안
* Paper/Architecture figure to code
* AI 분야 논문이나 관련 description을 곧 바로 PyTorch 코드로 바꿔주는 프로그램
* nn.Module을 쌓아 구현하기 때문에 이러한 형태를 학습시킬 수 있을 것 같다는 추측..?
* Con…
-
Used this model: https://huggingface.co/meta-llama/Llama-2-70b
Used these commands:
```
$ convert-pth-to-ggml.py models/LLaMa2-70B-meta 1
```
```
$ ./quantize ./models/LLaMa2-70B-meta/ggml…
-
### 🐛 Describe the bug
Hi,
Running Phi3 Medium on LocalAI with OpenVINO backend I found that while the int8 quantization is working correctly, the int4 quant gives the following error after few to…
-
The given main TorchGFN example runs fine, including the line:
# 4 - We define the GFlowNet.
gfn = TBGFlowNet(init_logZ=0., pf=pf_estimator, pb=pb_estimator) # We initialize logZ to 0
…