作者 | SHA1 備註 | 提交日期 |
---|---|---|
|
a20e2ce155 fix: pass cutlass_fp8_supported correctly for fbgemm_fp8 | 5 月之前 |
|
598afb63dd chore: add ignored layers for fp8 quant | 5 月之前 |
|
32acf54c63 fix: fbgemm_fp8 when modules_to_not_convert=None | 5 月之前 |
|
408ca43d2e feat: support fbgemm_fp8 quant on ampere | 5 月之前 |
|
9be43994fe feat: fbgemm quantization support (#601) | 5 月之前 |