GSQ: Highly-Accurate Low-Precision Scalar Quantization for LLMs via Gumbel-Softmax Sampling Paper • 2604.18556 • Published Apr 20 • 7
bartowski/deepseek-r1-qwen-2.5-32B-ablated-GGUF Text Generation • 33B • Updated Jan 24, 2025 • 4.94k • 86