-
Since We adopt per-tensor quantization for activation and per-channel quantization for weight, I am confused that why the the entries in a_scale tensor at line 230 not share the same value, I suppos…
-
### Before asking:
1. search the issues.
- Maybe this #3163 is related
2. search the docs.
- Done
#### What is your question?
I am trying to understand the Quant-Noise approach with i…
-
**What is your question?**
Hi!
I'm looking to only run select parts of vector search on GPUs, specifically PQ processes.
1. Are these directly available through the API?
2. What functions should…
-
This issue is grouping a few things that might be optimized to improve performance:
- Faster dot product possible, see here: https://github.com/xenova/transformers.js/pull/667
- Add binary quanti…
do-me updated
3 months ago
-
## Tree-based approach
- [Optimised kd-trees for fast image descriptor matching](https://users.cecs.anu.edu.au/~hartley/Papers/PDF/SilpaAnan:CVPR08.pdf)
- [Data structures and algorithms for neare…
-
https://github.com/cgtuebingen/Product-Quantization-Tree
https://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Wieschollek_Efficient_Large-Scale_Approximate_CVPR_2016_paper.pdf
-
## v0.4.0
### Core:
- [ ] Product quantization.
- [ ] Scalar quantization (u8).
- [ ] Binary quantization.
- [ ] Enable IVF Index
- [ ] Add filter expression to match_dense/match_sparse/match_…
-
## 論文リンク
http://ml.cs.tsinghua.edu.cn/~lingxi/PDFs/Xie_ICMR15_ONE.pdf
## 公開日(yyyy/mm/dd)
2015/06/23
## 概要
画像の分類と検索は画像間の類似度を対象にしているという点で本質的に同じだという主張。
分類でも検索でも使える Online Nearest-neighbor Estim…
-
This is a big one
The only reason we use BLAS is that we don't have efficient implementation of `matrix x matrix` multiplication. Naively doing parallel dot products is not optimal. We need to impl…
-