dev

GGitHubFeat: W8A16 Dequant implemented in torch (#24)

5a1e3c5b创建于 2025年5月14日历史提交

文件	最后提交记录	最后更新时间
qwen3-l0e1.result.safetensorsLFS	test: add test resources	1 年前
qwen3-l0e1.weight.safetensorsLFS	test: add test resources	1 年前
w8a16active-l0q_a_proj.safetensorsLFS	Feat: W8A16 Dequant implemented in torch (#24) * feat: add torch install role * feat: w8a16 dequant in CPU * Revert "feat: add torch install role" This reverts commit e62592825e9a9549eabb66c5a72371d036f52554. * fix: padding case * dev: load scale from weight server	1 年前