vllm-omni/tests/diffusion/quantization/artifacts/fp8_z_image_quality

FP8 Z-Image Quality Artifact

Generated by:

CUDA_VISIBLE_DEVICES=0 \
VLLM_OMNI_QUALITY_OUTPUT_DIR=$PWD/tests/diffusion/quantization/artifacts/fp8_z_image_quality \
.venv/bin/python -m pytest \
  tests/diffusion/quantization/test_quantization_quality.py::test_quantization_quality[fp8_z_image] \
  -s -v -m ""

Quantized config:

{
    "method": "fp8",
    "ignored_layers": [
        "img_mlp",
        "layers.15..29.{attention.to_qkv,attention.to_out.0,feed_forward.w13,feed_forward.w2}",
        "model.layers.28..35.{self_attn.q_proj,self_attn.k_proj,self_attn.v_proj,self_attn.o_proj,"
        "mlp.gate_proj,mlp.up_proj,mlp.down_proj}",
    ],
}

Result: passed with max_lpips=0.15.

Metric	Value
LPIPS	0.128773
PSNR	20.125577 dB
MAE	0.047712

This uses regular online FP8 quantization for the text encoder, with only the final 8 text-encoder blocks routed through ignored_layers for BF16 fallback.

Comparison

Individual Outputs