tmlr-2026

[TMLR 26] EdgeCrafter: Compact ViTs for Edge Dense Prediction via Task-Specialized Distillation

[TMLR] Offical Implementation of ToMoE: ToMoE: Converting Dense Large Language Models to Mixture-of-Experts through Dynamic Structural Pruning

Official implementation of "Semantic-Drive": a local-first neuro-symbolic framework for long-tail AV data curation. We utilize open-vocabulary grounding (YOLOE) and multi-VLM consensus to mine safety-critical "Dark Data" within a consumer hardware compute budget.

Reciprocal space-aware long-range modeling for crystalline property prediction (TMLR 2026)

[TMLR'26] PriSM: Prior-Guided Search Methods for Query Efficient Black-Box Adversarial Attacks

TMLR 2026 | Mechanistic interpretability: attention-head binding (EB*) as a marker of concept emergence. 7 models, 5 architectures (Pythia 160M–2.8B, OLMo-1B, CRFM GPT-2, SmolLM3-3B, Qwen2.5-1.5B), 41 terms.

Reproducible benchmark for cold-start adaptability: measures cumulative errors while learning from scratch, not just final accuracy. Published in TMLR 2026.

Code for TMLR manuscript 'Paradoxical noise preference in RNNs'

[TMLR 2026] Website for "Towards Near-Real-Time Telemetry-Aware Routing with Neural Routing Algorithms"

[TMLR 2026] Code for "The Scissors Effect: When Resize-Based Input Diversity Helps or Hurts Transfer Attacks".

LLM as a Judge evaluation benchmark for measuring bias and testing debiasing strategies across GPT, Claude, Gemini, and Llama, with reproducible code and data.