Files
zeekayandHanzo Dev b7fade4b5f fix(quant): the CUDA GPTQ layer knows about NVFP4
Adding the NVFP4 config variant left `gptq_cuda.rs` non-exhaustive, so
`--features cuda` has not compiled since. The arm is unreachable like its
neighbours: GPTQ never receives an NVFP4 config.

Co-authored-by: Hanzo Dev <[email protected]>
2026-09-15 02:04:06 -07:00
..