29 Mar 2024
DeepCt: Predicting pharmacokinetic concentration-time curves and compartmental models from chemical structure using deep learning #machinelearning #compchem dx.doi.org/10.26434/chemrxiv…
1
1
8
1,180
This is gonna sound out of nowhere, but I believe that the misconception (is that the word?) of DeepCt stems from their cutesy final design, cuz if they looked like anything they did in their beta designs, it’d be more like “yea its so them” cuz they ACTUALLY looked rouge.
2
41
1,551
This just made me remember in game dialogue… DeepCt are EXTREMELY territorial of Splatlnds, so the girls would DEF feel betrayed after hearing what BgMan did…
25 Aug 2023
The thing is, Shiver and Frye obviously have a tendency to want to fight everything for fun. They are delinquent and bandits, they aren't perfectly cute and soft idols
Plus the way it is portrayed really makes me think of the girls seeing Big Man as cheating on them.
2
1
47
2,187
4 May 2023
DeepCT is a new neural network architecture which can learn complex interconnections of epigenetic features and infer unmeasured data from any available input.
Cell type–specific interpretation of noncoding variants using deep learning–based methods doi.org/10.1093/gigascience/…
1
2
2
503
23 Feb 2023
We don't have a sample app for it because SPLADE models are not open-source, but the building blocks are there. Example with deepct (which is a similar sparse weighting schema) docs.vespa.ai/en/using-wand-…
1
3
465
28 Feb 2022
It's National Invasive Species Awareness Week.
To learn more about aquatic invasive species and how to prevent their spread, please see the link below.
portal.ct.gov/DEEP/Fishing/G…
#boating #boatingct #deepct #boatlaunch #cleanwater #cleanwaterways
1
3
🚀 The «Bioinformatics» research group led by Olga Kardymon presented a new architecture of the «DeepCT» neural network that can help doctors determine the causes of hereditary diseases: biorxiv.org/content/biorxiv/…
1
2
11 Nov 2021
Ok. I'm researching for my first in-person talk since 2019 at @UnGaggedUK next Thursday, and down a rabbit hole. 60 tabs open on dense retrieval, first stage retrieval, ANCE, DeepCT. Currently very confused but hope clarity in part will follow. Super interesting topic
1
1
13
5 Nov 2021
Interesting example of DeepImpact - you can think of it as a hybrid of doc2query and DeepCT. #cikm2021 #pyterrier
1
2
5 Nov 2021
Sean @macavaney is now delivering Part 3 of the #PyTerrier tutorial at #cikm2021 on Contemporary Retrieval Architectures: Neural re-rankers such as BERT, EPIC, T5 and neural index augmentation such as DeepCT and Doc2Query
Slides & notebooks available at
github.com/terrier-org/cikm2…
3
3
16
9 Sep 2021
🚨New 🍻BEIR preprint on Arxiv🚨
What’s 🆕? Evaluated latest SOTA🚀 reranking (MiniLM), dense (TAS-B), and sparse (DeepCT, docT5query) models. Added a 🆕 dataset: Robust04. Using hole@10, found a few datasets with annotation biases. w/ @Nils_Reimers.
pdf: arxiv.org/abs/2104.08663
ALT BEIR research paper and it's abstract
20 Apr 2021
🚨New paper alert 🚨
🍻 BEIR: a heterogeneous benchmark for IR. 17 datasets, 9 tasks with diverse domains. 9 SOTA retrieval models evaluated in a zero-shot setup.
w/ @Nils_Reimers @arueckle @abhesrivas, IG at @UKPLab
pdf: arxiv.org/abs/2104.08663
More details, code 👇
#NLProc
ALT Cover Page for BEIR arxiv paper
ALT Flowchart describing tasks and datasets in BEIR
ALT Minimal code for running BEIR
3
3
28
6 Jul 2021
Really happy about the launch of 🍻 BEIR v0.2.0 🔥🔥🚀🚀
FAISS Indexes, DeepCT, BPR, Multilingual Datasets, MSMARCO-v3 models, Multi-GPU for question generation all now available.
Check out the new release below!
github.com/UKPLab/beir/relea…
3
13
69
If you're interested in DeepImpact (~ doc2query DeepCT) arxiv.org/abs/2104.12016 - Anserini's got you covered! Reproduce results in forthcoming #sigir2021 paper here: github.com/castorini/anserin… Thanks to @antonio_mallia and team for sharing data!
1
2
23
29 Apr 2021
Updated Vespa's high level WAND documentation docs.vespa.ai/en/using-wand-… with how to use DeepCT and also combine efficient retrieval over DeepCT weighting with traditional text ranking features like BM25.
1
9