{ "recipe": { "experts": { "default": "native", "layers": { "20": "gptq:4:64", "21": "gptq:4:64", "22": "gptq:4:64", "23": "gptq:4:64" } }, "dense": { "attn": [ 8, 64 ], "embed": [ 8, 64 ], "head": [ 8, 64 ], "vision": [ 8, 64 ] } }, "calibration_tokens": 49465, "method": "GPTQ (error-compensated rounding), affine, group size 64; scales and offsets optimised at float16 precision and stored as bfloat16" }