Machine Learning RSS Feeds

Directory7 machine learning feeds · 218 articlesEnglish only →

Latest

Machine Learning articles

Feedback Distillation: A New Training Method for Improving LLM Reasoning in Theorem Proving

This paper introduces Feedback Distillation, a novel training method for reasoning models that improves upon standard GRPO (Group Relative Policy Optimization). The method trains a model to match its own token-level distribution conditioned on privileged feedback from a language

Insighttechnologysciencemachine learning

arxiv.org·Business by Flipboard·1h ago·1 min read

Why Stochastic Rounding Prevents Error Accumulation in Low-Precision Arithmetic

Insightsciencemachine learningprogramming

convergentthinking.sh2d ago

DeepSeek-V4: Hybrid Sparse-Attention Architecture Enables Efficient Million-Token Context Inference

Insighttechnologymachine learningprogramming

Shared by @ericalbers.bsky.social on Bluesky ↗

artgor.medium.com14h ago

Rotary GPU: Enabling Large Mixture-of-Experts Models on Consumer Laptop GPUs with Limited Memory

Insighttechnologymachine learningprogramming

arxiv.org1d ago

What pretraining on unlabeled text teaches large language models about language structure

Insighttechnologyeducationartificial intelligence

Shared by @ocramz.bsky.social on Bluesky ↗

sebastianraschka.com2d ago

LinkedIn cuts GPU training hours by 65% with Generative Recommender system optimizations

Newstechnologymachine learningprogramming

Shared by @startuphub.bsky.social on Bluesky ↗

startuphub.ai3d ago

Rank-Aware Decomposition Technique Reduces Computation in Recommender Systems by 87.5%

Insighttechnologymachine learningprogramming

Shared by @reachsumit.com on Bluesky ↗

arxiv.org4d ago

PromptEmbedder: A Dual-LLM Framework for Efficient, Architecture-Agnostic Text Embedding

The article presents PromptEmbedder, a novel dual-LLM framework for efficient and transferable text embedding. It addresses the bottleneck of current methods like LoRA that require costly retraining when new backbone architectures emerge. PromptEmbedder uses a Prompting LLM to ge

Insighttechnologysciencemachine learning

arxiv.org·Tech by Flipboard·4d ago·2 min read

Unified Framework for Variational Quantum Knowledge Graph Embeddings on NISQ Devices

This paper introduces a unified framework for variational quantum algorithms (VQAs) applied to knowledge graph embeddings on near-term NISQ quantum devices. It compares two existing approaches—one using n+1 qubits with a switch test on an ancillary qubit, and another using 2n+1 q

Insighttechnologysciencemachine learning

arxiv.org·Tech by Flipboard·4d ago·2 min read

Contextual Rollout Bandits: A Neural Scheduling Framework for Efficient Reinforcement Learning with Verifiable Rewards

This paper introduces Contextual Rollout Bandits, a novel framework for Reinforcement Learning with Verifiable Rewards (RLVR) that addresses inefficiencies in how rollouts are used during training of large language models. The authors formulate rollout scheduling as a contextual

Insighttechnologysciencemachine learning

arxiv.org·Tech by Flipboard·5d ago·2 min read

Eureka: An LLM-Driven Framework for Automated Feature Engineering in Enterprise AI

Insighttechnologysciencemachine learning

Shared by @ai-firehose.column.social on Bluesky ↗

arxiv.org5d ago

Sleep-Like Consolidation Mechanism Improves Long-Context Performance in Transformer Language Models

Insighttechnologysciencemachine learning

arxiv.org5d ago

PICO: A Practical Learned Image Codec Optimized for Human Visual Perception

Newstechnologysciencemachine learning

apple.github.io7d ago

Multi-Stream LLMs: A Parallel Architecture to Overcome Single-Stream Bottlenecks in Language Models

Insighttechnologysciencemachine learning

arxiv.org10d ago

Hands-on evaluation of MiniMax M2.7 via API on ML and coding workflows

Reviewtechnologymachine learningprogramming

andlukyane.com12d ago

Self-Distillation Fine-Tuning (SDFT): A Method for Continual Learning from Demonstrations

Insighttechnologysciencemachine learning

arxiv.org15d ago

Orthrus: A Dual-Architecture Framework for Fast, Lossless LLM Inference via Diffusion Decoding

Orthrus is a dual-architecture framework that combines autoregressive LLMs with diffusion models to enable fast, lossless parallel token generation. It uses a Qwen3 backbone and guarantees strictly lossless generation while achieving significant speedups (up to 4.25×). The projec

Codetechnologymachine learningprogramming

github.com·Hacker News: Front Page·16d ago·4 min read

Optimizing Matrix Multiplication in Swift for LLM Training on Apple Silicon

technologymachine learningprogramming

Cocoa with Love21d ago

buildfastwithai.com0

theaiinnovator.com0

Owl Posting0

Stack Abuse0

sander.ai0

Feedback Distillation: A New Training Method for Improving LLM Reasoning in Theorem Proving

Why Stochastic Rounding Prevents Error Accumulation in Low-Precision Arithmetic

DeepSeek-V4: Hybrid Sparse-Attention Architecture Enables Efficient Million-Token Context Inference

Rotary GPU: Enabling Large Mixture-of-Experts Models on Consumer Laptop GPUs with Limited Memory

What pretraining on unlabeled text teaches large language models about language structure

LinkedIn cuts GPU training hours by 65% with Generative Recommender system optimizations

Rank-Aware Decomposition Technique Reduces Computation in Recommender Systems by 87.5%

PromptEmbedder: A Dual-LLM Framework for Efficient, Architecture-Agnostic Text Embedding

Unified Framework for Variational Quantum Knowledge Graph Embeddings on NISQ Devices

Contextual Rollout Bandits: A Neural Scheduling Framework for Efficient Reinforcement Learning with Verifiable Rewards

Eureka: An LLM-Driven Framework for Automated Feature Engineering in Enterprise AI

Sleep-Like Consolidation Mechanism Improves Long-Context Performance in Transformer Language Models

PICO: A Practical Learned Image Codec Optimized for Human Visual Perception

Multi-Stream LLMs: A Parallel Architecture to Overcome Single-Stream Bottlenecks in Language Models

Hands-on evaluation of MiniMax M2.7 via API on ML and coding workflows

Self-Distillation Fine-Tuning (SDFT): A Method for Continual Learning from Demonstrations

Orthrus: A Dual-Architecture Framework for Fast, Lossless LLM Inference via Diffusion Decoding

Optimizing Matrix Multiplication in Swift for LLM Training on Apple Silicon