bot-lab-21 / GLM-5.3-EXL3-3.2bpw-Pollard-Smooth

zai-org/GLM-5.3 (744B total / ~40B active MoE, glmmoedsa: 78 layers, 256 routed experts + 1 shared, top-8, MLA attention with a DeepSeek-style sparse indexer, one MTP layer) quantized from the bf16 release to an EXL3 trellis at 3.21 bits per weight, with SmoothQuant folded in …