Qwen-3 8B

The 'Qwen-3-8B' dataset/benchmark is used to evaluate the performance of Large Language Model compression methods, specifically in terms of perplexity and zero-shot tasks.

Papers using Qwen-3 8B (9)

Output-Space Allocation Costs for Calibration-Guided LLM Compression: An Empirical Study2026A general tensor-structured compression scheme for efficient large language models2026Implicit Hierarchical GRPO: Decoupling Tool Invocation from Execution for Tool-Integrated Mathematical Reasoning2026Bridging Online and Offline RL: Contextual Bandit Learning for Multi-Turn Code Generation2026OSCAR: Offline Spectral Covariance-Aware Rotation for 2-bit KV Cache Quantization2026SigmaScale: LLM Compression with SVD-based Low-Rank Decomposition and Learned Scaling Matrices2026Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization2026LatentGuard: Controllable Latent Steering for Robust Refusal of Attacks and Reliable Response Generation2025Compressing Chain-of-Thought in LLMs via Step Entropy2025