-
Try out
- masking/augmentation
- established techniques (TS-DAE, SimCSE)
Use a more-or-less standardized dataset for text similarity (e.g. from benchmarks used in the papers), and try to compare …
-
Hi @datquocnguyen ,
I'm using PhoBERT for finding semantic textual similarity between sentences. I'm wonder is it necessary to perform lowercase and remove punctuation in the input text before feedin…
-
Hi there
The current Blenderbot 2.0 work pretty well on answering questions. However, the bot sometimes asks **repetitive questions** or **asking questions in which the answer is already given** by…
-
We need to come up with a definition and guidelines as to what we mean when we say WGSL is bijective to SPIR-V. This would the help guide the language evolution.
-
Common metrics for evaluating QA like Exact Match or f1 score are very strict. They can be better applied to settings where an exact entity (name, date or number) needs to be extracted.
For more comp…
-
1. How is BlueGraph different compared to CDLib (with iGraph and NetworkX)?
- There are not enough variety of centrality measures compared to NetworkX
- https://github.com/networkx/networkx/is…
-
## 論文URL
https://arxiv.org/abs/2104.08821
## 著者
Tianyu Gao, Xingcheng Yao, Danqi Chen
## 会議
EMNLP 2021
## 背景
効果的な分散表現を獲得することは、それを用いたNLPタスクの精度を上げることに寄与する。
## 目的
効果的なContrasti…
-
Here you will find a long list of the articles thats need to be coded. They are divided into sections, one for each coder (TR = Timo, MR = Melanie, JC = Joseph, AB = Agata, LK = Liam). Each item in th…
-
Related issues: #1891 and #2629.
Suggestion: The call `Engine::Flush()` should be made MPI-collective and its effect should be to drain data current held in ADIOS buffers to disk. In streaming engi…
-
I trained an extension of model `sentence-transformers/paraphrase-multilingual-mpnet-base-v2` (see #235).
After training I used the script [save_pretrained_hf.py](https://github.com/JohnGiorgi/DeCL…