YUSS new trained #sparse #autoencoder has FOUND THE TEXT OBSESSION in #CLIP #AI!🥳🤩 Only 1 smol problem..🤣 It's not just *ONE* typographic cluster.🤯 Left: 3293 encodes CLIP neurons for English, probably EN text signs. Right: 2052 encodes East Asian + German + Mirrored. 👇🧵

zer0int1's tweet image. YUSS new trained #sparse #autoencoder has FOUND THE TEXT OBSESSION in #CLIP #AI!🥳🤩

Only 1 smol problem..🤣
It's not just *ONE* typographic cluster.🤯

Left: 3293 encodes CLIP neurons for English, probably EN text signs.
Right: 2052 encodes East Asian + German + Mirrored.
👇🧵
zer0int1's tweet image. YUSS new trained #sparse #autoencoder has FOUND THE TEXT OBSESSION in #CLIP #AI!🥳🤩

Only 1 smol problem..🤣
It's not just *ONE* typographic cluster.🤯

Left: 3293 encodes CLIP neurons for English, probably EN text signs.
Right: 2052 encodes East Asian + German + Mirrored.
👇🧵

Thanks for the en-/discouragement, #GPT4o 😂 Now #sparse #autoencoder #2 learns to be a #babelfish, translating #logits to #token sequences.🤯 It could help decode a sparse #CLIP embedding, it could help decode a gradient ascent #CLIP #opinion! God luck & good speed, #SAE ✊😬

zer0int1's tweet image. Thanks for the en-/discouragement, #GPT4o 😂
Now #sparse #autoencoder #2 learns to be a #babelfish, translating #logits to #token sequences.🤯

It could help decode a sparse #CLIP embedding, it could help decode a gradient ascent #CLIP #opinion!

God luck & good speed, #SAE ✊😬

Fun with #CLIP's #sparse #autoencoder: First glimpse, I thought [Act idx 20] was encoding "sports / tennis". But that's not the shared feature. It's a "people wearing a thing around their head that makes them look stupid" feature. 🤣😂 #lmao #AI #AIweirdness

zer0int1's tweet image. Fun with #CLIP's #sparse #autoencoder: First glimpse, I thought [Act idx 20] was encoding "sports / tennis".

But that's not the shared feature. It's a "people wearing a thing around their head that makes them look stupid" feature. 🤣😂
#lmao #AI #AIweirdness

w00t best #sparse #autoencoder for the #AI so far \o/ I kinda maxed out the reconstruction quality @ 97% cossim. I can stop optimizing this ball of mathematmadness now. 😅 Tied encoder / decoder weights + extra training on #CLIP's "hallucinwacky ooooowords" ('opinions'). 😂

zer0int1's tweet image. w00t best #sparse #autoencoder for the #AI so far \o/

I kinda maxed out the reconstruction quality @ 97% cossim. I can stop optimizing this ball of mathematmadness now. 😅

Tied encoder / decoder weights + extra training on #CLIP's "hallucinwacky ooooowords" ('opinions'). 😂

高積算高s/nのMRS dataで学習させたStacked Autoencoderを用い、MRS積算回数をごく少なくしても充分評価に耐えるspectraを生成する用意したというpreprint ヒト脳の低積算画像では,SNRが43.8%増加し,MSEが68.8%減少し、定量性は保たれた #MRS #autoencoder #papers arxiv.org/abs/2303.16503…

thmonk_pf's tweet image. 高積算高s/nのMRS dataで学習させたStacked Autoencoderを用い、MRS積算回数をごく少なくしても充分評価に耐えるspectraを生成する用意したというpreprint

ヒト脳の低積算画像では,SNRが43.8%増加し,MSEが68.8%減少し、定量性は保たれた

#MRS 
#autoencoder 
#papers 

arxiv.org/abs/2303.16503…

Attention Schema-based Attention Control (ASAC): A Cognitive-Inspired Approach for Attention Management in Transformers openreview.net/forum?id=cxRlo… #attention #autoencoder #neural


A Novel Method for Time Series Counterfactual Inference Based on Penalized Autoencoders openreview.net/forum?id=X6lrz… #autoencoders #autoencoder #counterfactual


Here is the one of the rare papers arxiv.org/abs/2504.12418 we did where a supervised event classifier is compared with an unsupervised #autoencoder using exactly the same input and a similar neural network architecture for the hidden layers. The example uses double-#Higgs


Amazing paper:"Arousal as a universal embedding for spatiotemporal brain dynamics"🧠🐁 𒅒𒈔 nature.com/articles/s4158… you can even find the #autoencoder based code: github.com/ryraut/arousal… but dear #Brain people🧠, why do you find cool brain dynamics, state trajectories, embedding?

Dr_Alex_Crimi's tweet image. Amazing paper:"Arousal as a universal embedding for spatiotemporal brain dynamics"🧠🐁 𒅒𒈔
nature.com/articles/s4158… you can even find the #autoencoder based code: github.com/ryraut/arousal…
but dear #Brain people🧠,
why do you find cool brain dynamics, state trajectories, embedding?

'Autoencoders in Function Space', by Justin Bunker, Mark Girolami, Hefin Lambley, Andrew M. Stuart, T. J. Sullivan. jmlr.org/papers/v26/25-… #autoencoders #autoencoder #generative


We continue at the @SwanseaPPCTh @SwanseaUni #machinelearning & #lattice workshop with a talk by Simran Singh (@unibielefeld) on application of #autoencoder|s to exploration of the phase structure of the strong interactions governing quarks & gluons. @dfg_public #WomeninPhysics

rothkopfAK's tweet image. We continue at the @SwanseaPPCTh @SwanseaUni #machinelearning & #lattice workshop with a talk by Simran Singh (@unibielefeld) on application of #autoencoder|s to exploration of the phase structure of the strong interactions governing quarks & gluons. @dfg_public #WomeninPhysics
rothkopfAK's tweet image. We continue at the @SwanseaPPCTh @SwanseaUni #machinelearning & #lattice workshop with a talk by Simran Singh (@unibielefeld) on application of #autoencoder|s to exploration of the phase structure of the strong interactions governing quarks & gluons. @dfg_public #WomeninPhysics

A Deep Bayesian Nonparametric Framework for Robust Mutual Information Estimation openreview.net/forum?id=mqGzG… #regularization #nonparametric #autoencoder


Causal Dynamic Variational Autoencoder for Counterfactual Regression in Longitudinal Data Mouad El Bouchattaoui, Myriam Tami, BENOIT LEPETIT, Paul-Henry Cournède. Action editor: Amit Sharma. openreview.net/forum?id=atf9q… #unobserved #autoencoder #confound


Causal Dynamic Variational Autoencoder for Counterfactual Regression in Longitudinal Data Mouad El Bouchattaoui, Myriam Tami, BENOIT LEPETIT, Paul-Henry Cournède. Action editor: Amit Sharma. openreview.net/forum?id=atf9q… #unobserved #autoencoder #confound


Approximately Equivariant Recurrent Generative Models for Quasi-Periodic Time Series with a Progressive Training Scheme openreview.net/forum?id=KHk5E… #recurrent #autoencoder #generative


New article! Interpretable and efficient data-driven discovery and control of distributed systems 👉bit.ly/449T0HJ By Florian Wolf, Nicolò Botteghi, Urban Fasel & Andrea Manzoni Part of Physics-Enhancing #MachineLearning special collection #autoencoder #data

DCE_Journal's tweet image. New article!

Interpretable and efficient data-driven discovery and control of distributed systems

👉bit.ly/449T0HJ

By Florian Wolf, Nicolò Botteghi, Urban Fasel &
Andrea Manzoni

Part of Physics-Enhancing #MachineLearning special collection

#autoencoder #data

Cross-Layer Discrete Concept Discovery for Interpreting Language Models openreview.net/forum?id=xBVTq… #autoencoder #quantization #representations


New patent application #US20250344987A1 by #BiosenseWebster explores reducing noise in intracardiac ECGs using a denoised #autoencoder. The system refines ECGs with #DeepLearning, enhancing signal clarity by encoding and decoding raw data to remove noise. Key features include…

PatentPulse's tweet image. New patent application #US20250344987A1 by #BiosenseWebster explores reducing noise in intracardiac ECGs using a denoised #autoencoder. 

The system refines ECGs with #DeepLearning, enhancing signal clarity by encoding and decoding raw data to remove noise. 

Key features include…
PatentPulse's tweet image. New patent application #US20250344987A1 by #BiosenseWebster explores reducing noise in intracardiac ECGs using a denoised #autoencoder. 

The system refines ECGs with #DeepLearning, enhancing signal clarity by encoding and decoding raw data to remove noise. 

Key features include…
PatentPulse's tweet image. New patent application #US20250344987A1 by #BiosenseWebster explores reducing noise in intracardiac ECGs using a denoised #autoencoder. 

The system refines ECGs with #DeepLearning, enhancing signal clarity by encoding and decoding raw data to remove noise. 

Key features include…
PatentPulse's tweet image. New patent application #US20250344987A1 by #BiosenseWebster explores reducing noise in intracardiac ECGs using a denoised #autoencoder. 

The system refines ECGs with #DeepLearning, enhancing signal clarity by encoding and decoding raw data to remove noise. 

Key features include…

LLM-based Contrastive Self-Supervised AMR Learning with Masked Graph Autoencoders for Fake News Detection openreview.net/forum?id=puFYj… #misinformation #supervised #autoencoder


Conditional Latent Space Molecular Scaffold Optimization for Accelerated Molecular Design Onur Boyar, Hiroyuki Hanada, Ichiro Takeuchi. Action editor: Stanislaw Jastrzebski. openreview.net/forum?id=KhxVc… #molecular #molecule #autoencoder


#Call4Reading 📖 Deep #Autoencoder-Based Integrated Model for Anomaly Detection and Efficient Feature Extraction in #IoT Networks ✍ Authors: Khaled A. Alaghbari, Heng-Siong Lim, Mohamad Hanif Md Saad and Yik Seng Yong 🔗 Read the full paper here: mdpi.com/2624-831X/4/3/…

Iot_MDPI's tweet image. #Call4Reading
📖 Deep #Autoencoder-Based Integrated Model for Anomaly Detection and Efficient Feature Extraction in #IoT Networks
✍ Authors: Khaled A. Alaghbari, Heng-Siong Lim, Mohamad Hanif Md Saad and Yik Seng Yong

🔗 Read the full paper here: mdpi.com/2624-831X/4/3/…

A Novel Method for Time Series Counterfactual Inference Based on Penalized Autoencoders openreview.net/forum?id=X6lrz… #autoencoders #autoencoder #counterfactual


A Deep Bayesian Nonparametric Framework for Robust Mutual Information Estimation openreview.net/forum?id=mqGzG… #regularization #nonparametric #autoencoder


Learning Encoding-Decoding Direction Pairs to Unveil Concepts of Influence in Deep Vision Networks openreview.net/forum?id=lIeyZ… #embeddings #autoencoder #decoding


Attention Schema-based Attention Control (ASAC): A Cognitive-Inspired Approach for Attention Management in Transformers openreview.net/forum?id=cxRlo… #attention #autoencoder #neural


Learning the Language of Protein Structure Jérémie DONA, Benoit Gaujac, Timothy Atkinson, Liviu Copoiu, Thomas Pierrot, Thomas D Barrett. Action editor: Lingpeng Kong. openreview.net/forum?id=SRRPQ… #proteins #autoencoder #representations


PredLDM: Spatiotemporal Sequence Prediction with Latent Diffusion Models openreview.net/forum?id=TWmnO… #autoencoder #spatiotemporal #predicting


Automated Attention Pattern Discovery at Scale in Large Language Models openreview.net/forum?id=KpsUN… #attention #predicts #autoencoder


'Autoencoders in Function Space', by Justin Bunker, Mark Girolami, Hefin Lambley, Andrew M. Stuart, T. J. Sullivan. jmlr.org/papers/v26/25-… #autoencoders #autoencoder #generative


Amazing paper:"Arousal as a universal embedding for spatiotemporal brain dynamics"🧠🐁 𒅒𒈔 nature.com/articles/s4158… you can even find the #autoencoder based code: github.com/ryraut/arousal… but dear #Brain people🧠, why do you find cool brain dynamics, state trajectories, embedding?

Dr_Alex_Crimi's tweet image. Amazing paper:"Arousal as a universal embedding for spatiotemporal brain dynamics"🧠🐁 𒅒𒈔
nature.com/articles/s4158… you can even find the #autoencoder based code: github.com/ryraut/arousal…
but dear #Brain people🧠,
why do you find cool brain dynamics, state trajectories, embedding?

Emergence of Quantised Representations Isolated to Anisotropic Functions openreview.net/forum?id=aokVp… #representations #representational #autoencoder


STLDM: Spatio-Temporal Latent Diffusion Model for Precipitation Nowcasting openreview.net/forum?id=f4oJw… #autoencoder #precipitation #prediction


Semi-Symmetrical, Fully Convolutional Masked #Autoencoder for TBM Muck #ImageSegmentation ✏️ Ke Lei et al. 🔗 brnw.ch/21wVdRA Viewed: 2021; Cited: 10 #mdpisymmetry #selfsupervised #instancesegmentation

Symmetry_MDPI's tweet image. Semi-Symmetrical, Fully Convolutional Masked #Autoencoder for TBM Muck #ImageSegmentation
✏️ Ke Lei et al.
🔗 brnw.ch/21wVdRA
Viewed: 2021; Cited: 10
#mdpisymmetry #selfsupervised #instancesegmentation

لا توجد نتائج لـ "#autoencoder"
Loading...

Something went wrong.


Something went wrong.


United States Trends