Lior Wolf
YOU?
Author Swipe
View article: Server-side performance and environmental impact evaluation of a hybrid e-learning website
Server-side performance and environmental impact evaluation of a hybrid e-learning website Open
The importance of e-learning in modern education systems is increasing, and Server-Side Rendering (SSR) is making a comeback as a popular technique for building websites with dynamic content. This paper examines a real-world e-learning pla…
View article: ConsiStyle: Style Diversity in Training-Free Consistent T2I Generation
ConsiStyle: Style Diversity in Training-Free Consistent T2I Generation Open
In text-to-image models, consistent character generation is the task of achieving text alignment while maintaining the subject's appearance across different prompts. However, since style and appearance are often entangled, the existing met…
View article: Detecting and Pruning Prominent but Detrimental Neurons in Large Language Models
Detecting and Pruning Prominent but Detrimental Neurons in Large Language Models Open
Large language models (LLMs) often develop learned mechanisms specialized to specific datasets, such as reliance on domain-specific correlations, which yield high-confidence predictions without generalizable reasoning. While beneficial in …
View article: Execution Guided Line-by-Line Code Generation
Execution Guided Line-by-Line Code Generation Open
We present a novel approach to neural code generation that incorporates real-time execution signals into the language model generation process. While large language models (LLMs) have demonstrated impressive code generation capabilities, t…
View article: Overclocking LLM Reasoning: Monitoring and Controlling Thinking Path Lengths in LLMs
Overclocking LLM Reasoning: Monitoring and Controlling Thinking Path Lengths in LLMs Open
Recently, techniques such as explicit structured reasoning have demonstrated strong test-time scaling behavior by enforcing a separation between the model's internal "thinking" process and the final response. A key factor influencing answe…
View article: Revisiting LRP: Positional Attribution as the Missing Ingredient for Transformer Explainability
Revisiting LRP: Positional Attribution as the Missing Ingredient for Transformer Explainability Open
The development of effective explainability tools for Transformers is a crucial pursuit in deep learning research. One of the most promising approaches in this domain is Layer-wise Relevance Propagation (LRP), which propagates relevance sc…
View article: ConsiStyle: Style Diversity in Training-Free Consistent T2I Generation
ConsiStyle: Style Diversity in Training-Free Consistent T2I Generation Open
In text-to-image models, consistent character generation is the task of achieving text alignment while maintaining the subject's appearance across different prompts. However, since style and appearance are often entangled, the existing met…
View article: Overflow Prevention Enhances Long-Context Recurrent LLMs
Overflow Prevention Enhances Long-Context Recurrent LLMs Open
A recent trend in LLMs is developing recurrent sub-quadratic models that improve long-context processing efficiency. We investigate leading large long-context models, focusing on how their fixed-size recurrent memory affects their performa…
View article: IlluSign: Illustrating Sign Language Videos by Leveraging the Attention Mechanism
IlluSign: Illustrating Sign Language Videos by Leveraging the Attention Mechanism Open
Sign languages are dynamic visual languages that involve hand gestures, in combination with non manual elements such as facial expressions. While video recordings of sign language are commonly used for education and documentation, the dyna…
View article: A Meaningful Perturbation Metric for Evaluating Explainability Methods
A Meaningful Perturbation Metric for Evaluating Explainability Methods Open
Deep neural networks (DNNs) have demonstrated remarkable success, yet their wide adoption is often hindered by their opaque decision-making. To address this, attribution methods have been proposed to assign relevance values to each part of…
View article: Non-Invasive Quantification of Viability in Spheroids Using Deep Learning
Non-Invasive Quantification of Viability in Spheroids Using Deep Learning Open
1 Abstract In vitro viability assays are widely used in drug discovery, development, and pharmacovigilance. Traditional methods for evaluating cell viability often involve destructive processes, rendering the culture non-viable. As such, t…
View article: On the Expressivity of Selective State-Space Layers: A Multivariate Polynomial Approach
On the Expressivity of Selective State-Space Layers: A Multivariate Polynomial Approach Open
Recent advances in efficient sequence modeling have introduced selective state-space layers, a key component of the Mamba architecture, which have demonstrated remarkable success in a wide range of NLP and vision tasks. While Mamba's empir…
View article: Deep Active Speech Cancellation with Mamba-Masking Network
Deep Active Speech Cancellation with Mamba-Masking Network Open
We present a novel deep learning network for Active Speech Cancellation (ASC), advancing beyond Active Noise Cancellation (ANC) methods by effectively canceling both noise and speech signals. The proposed Mamba-Masking architecture introdu…
View article: Classifier-Guided Captioning Across Modalities
Classifier-Guided Captioning Across Modalities Open
Most current captioning systems use language models trained on data from specific settings, such as image-based captioning via Amazon Mechanical Turk, limiting their ability to generalize to other modality distributions and contexts. This …
View article: Reversed Attention: On The Gradient Descent Of Attention Layers In GPT
Reversed Attention: On The Gradient Descent Of Attention Layers In GPT Open
The success of Transformer-based Language Models (LMs) stems from their attention mechanism. While this mechanism has been extensively studied in explainability research, particularly through the attention values obtained during the forwar…
View article: Diffusion-Based Attention Warping for Consistent 3D Scene Editing
Diffusion-Based Attention Warping for Consistent 3D Scene Editing Open
We present a novel method for 3D scene editing using diffusion models, designed to ensure view consistency and realism across perspectives. Our approach leverages attention features extracted from a single reference image to define the int…
View article: SphereUFormer: A U-Shaped Transformer for Spherical 360 Perception
SphereUFormer: A U-Shaped Transformer for Spherical 360 Perception Open
This paper proposes a novel method for omnidirectional 360$\degree$ perception. Most common previous methods relied on equirectangular projection. This representation is easily applicable to 2D operation layers but introduces distortions i…
View article: Add-it: Training-Free Object Insertion in Images With Pretrained Diffusion Models
Add-it: Training-Free Object Insertion in Images With Pretrained Diffusion Models Open
Adding Object into images based on text instructions is a challenging task in semantic image editing, requiring a balance between preserving the original scene and seamlessly integrating the new object in a fitting location. Despite extens…
View article: Accelerating Error Correction Code Transformers
Accelerating Error Correction Code Transformers Open
Error correction codes (ECC) are crucial for ensuring reliable information transmission in communication systems. Choukroun & Wolf (2022b) recently introduced the Error Correction Code Transformer (ECCT), which has demonstrated promising p…
View article: Mitigating Copy Bias in In-Context Learning through Neuron Pruning
Mitigating Copy Bias in In-Context Learning through Neuron Pruning Open
Large language models (LLMs) have demonstrated impressive few-shot in-context learning (ICL) abilities. Still, we show that they are sometimes prone to a `copying bias', where they copy answers from provided examples instead of learning th…
View article: Detection-Driven Object Count Optimization for Text-to-Image Diffusion Models
Detection-Driven Object Count Optimization for Text-to-Image Diffusion Models Open
Accurately controlling object count in text-to-image generation remains a key challenge. Supervised methods often fail, as training data rarely covers all count variations. Methods that manipulate the denoising process to add or remove obj…
View article: Training-Free Consistent Text-to-Image Generation
Training-Free Consistent Text-to-Image Generation Open
Text-to-image models offer a new level of creative flexibility by allowing users to guide the image generation process through natural language. However, using these models to consistently portray the same subject across diverse prompts re…
View article: DeciMamba: Exploring the Length Extrapolation Potential of Mamba
DeciMamba: Exploring the Length Extrapolation Potential of Mamba Open
Long-range sequence processing poses a significant challenge for Transformers due to their quadratic complexity in input length. A promising alternative is Mamba, which demonstrates high performance and achieves Transformer-level capabilit…
View article: Knowledge Editing in Language Models via Adapted Direct Preference Optimization
Knowledge Editing in Language Models via Adapted Direct Preference Optimization Open
Large Language Models (LLMs) can become outdated over time as they may lack updated world knowledge, leading to factual knowledge errors and gaps. Knowledge Editing (KE) aims to overcome this challenge using weight updates that do not requ…
View article: Factor Graph Optimization of Error-Correcting Codes for Belief Propagation Decoding
Factor Graph Optimization of Error-Correcting Codes for Belief Propagation Decoding Open
The design of optimal linear block codes capable of being efficiently decoded is of major concern, especially for short block lengths. As near capacity-approaching codes, Low-Density Parity-Check (LDPC) codes possess several advantages ove…
View article: LLM Questionnaire Completion for Automatic Psychiatric Assessment
LLM Questionnaire Completion for Automatic Psychiatric Assessment Open
We employ a Large Language Model (LLM) to convert unstructured psychological interviews into structured questionnaires spanning various psychiatric and personality domains. The LLM is prompted to answer these questionnaires by impersonatin…
View article: Explaining Modern Gated-Linear RNNs via a Unified Implicit Attention Formulation
Explaining Modern Gated-Linear RNNs via a Unified Implicit Attention Formulation Open
Recent advances in efficient sequence modeling have led to attention-free layers, such as Mamba, RWKV, and various gated RNNs, all featuring sub-quadratic complexity in sequence length and excellent scaling properties, enabling the constru…
View article: Learning Linear Block Error Correction Codes
Learning Linear Block Error Correction Codes Open
Error correction codes are a crucial part of the physical communication layer, ensuring the reliable transfer of data over noisy channels. The design of optimal linear block codes capable of being efficiently decoded is of major concern, e…