
Industry Pro28d ago
ExLlamaV3 v1.4.7 Speeds CPU MoE, Adds DRY Sampler
The ExLlamaV3 project released version 1.4.7, introducing improved CPU mixture-of-experts performance across all tiers and a DRY sampler . The update also fixes recurrent slot leaks and accelerates quantization for 6-, 7-, and 8-bit trellis…
SINGLE SOURCE