Llama

The 'Llama' dataset/benchmark is used to evaluate the performance of weight-space model merging techniques in large language models by assessing expert access and I/O efficiency.

Papers using Llama (17)

TransMLA: Multi-head Latent Attention Is All You Need2025 · 6 citesAsk the Right Comparison:Bias-Aware Bayesian Active Top-$k$ Ranking with LLM Judges2026EPTS: Elastic Post-Training Sparsity for Efficient Large Language Model Compression2026RAS: Measuring LLM Safety Through Refusal Alignment2026Reasoning-preserved Efficient Distillation of Large Language Models via Activation-aware Initialization2026GAMMA: Global Bit Allocation for Mixed-Precision Models under Arbitrary Budgets2026Access Sets Matter: Budgeting Expert Reads for Scalable Weight-Space Model Merging20261+1>2: A Synergistic Sparse and Low-Rank Compression Method for Large Language Models2025Chunks as Arms: Multi-Armed Bandit-Guided Sampling for Long-Context LLM Preference Optimization2025PhantomHunter: Detecting Unseen Privately-Tuned LLM-Generated Text via Family-Aware Learning2025Quartet: Native FP4 Training Can Be Optimal for Large Language Models2025ARMOR: High-Performance Semi-Structured Pruning via Adaptive Matrix Factorization2025Spectral Scaling Laws in Language Models: How Effectively Do Feed-Forward Networks Use Their Latent Space?2025Accelerating Large Language Model Reasoning via Speculative Search2025SlimLLM: Accurate Structured Pruning for Large Language Models2025SlimGPT: Layer-wise Structured Pruning for Large Language Models2024 · 1 citesDAQ: Density-Aware Post-Training Weight-Only Quantization For LLMs2024