Llama-3-8B-Instruct

mentions 3 type Organization feed RSS

// recent coverage 3 mentions

04:00

2026-06-19

arxiv.org

large-language-models

Pruning via Causal Attribution Preserves Reasoning Performance in Large Language Models

Researchers introduced Causal Attribution Pruning (CAP), a training-free method that identifies critical attention heads in large language models by measuring their causal impact on reasoning tasks. C…

04:00

2026-06-12

arxiv.org

large-language-models

Dual-Stance Evaluation of Sycophancy: The Structure of Agreement and the Limits of Intervention

Researchers at arXiv introduced a dual-stance evaluation method to test whether activation steering on Llama-3-8B-Instruct reduces sycophancy without suppressing agreement with factually correct state…

04:00

2026-06-04

arxiv.org

large-language-models

Discourse-Role Labels as Presentation-Time Variables for Context Use in Language Models

Researchers introduced a paired fixed-content probe over 500 MMLU-Pro items to test how discourse-role labels such as Instruction:, Reference:, and Example: affect language model adoption of misleadin…

// co-occurs with top 8 entities

GPT-5.5 1 DeepSeek V4 Pro 1 Qwen2.5-7B-Instruct 1 MMLU-Pro 1 Causal Attribution Pruning 1 Wanda 1 Mistral-7B-Instruct 1 GSM8K 1

// topics top 6 topics

large language models 3 ai research 3 natural language processing 2 ai safety 1 ai ethics 1 ai products 1