sgl-project--sglang/scripts/ci/utils/diffusion/comparison_configs.json

{
    "_comment": "Per-model comparison config. Sampling params omitted where model defaults are correct — only override resolution, seed, and params that differ from defaults.",
    "test_image_url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png",
    "cases": [
        {
            "id": "flux1_dev_t2i_1024",
            "model": "black-forest-labs/FLUX.1-dev",
            "task": "text-to-image",
            "prompt": "A futuristic cyberpunk city at night, neon lights reflecting on wet streets",
            "width": 1024,
            "height": 1024,
            "seed": 42,
            "num_gpus": 2,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --dit-layerwise-offload false --tp-size 2",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "flux2_dev_t2i_1024",
            "model": "black-forest-labs/FLUX.2-dev",
            "task": "text-to-image",
            "prompt": "A futuristic cyberpunk city at night, neon lights reflecting on wet streets",
            "width": 1024,
            "height": 1024,
            "seed": 42,
            "num_gpus": 2,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --dit-layerwise-offload false --tp-size 2",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "qwen_image_2512_t2i_1024",
            "model": "Qwen/Qwen-Image-2512",
            "task": "text-to-image",
            "prompt": "A futuristic cyberpunk city at night, neon lights reflecting on wet streets",
            "width": 1024,
            "height": 1024,
            "seed": 42,
            "num_gpus": 2,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --tp-size 2",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "qwen_image_edit_2511",
            "model": "Qwen/Qwen-Image-Edit-2511",
            "task": "image-edit",
            "prompt": "Make the cat wear a red hat",
            "reference_image": true,
            "width": 1024,
            "height": 1024,
            "seed": 42,
            "num_gpus": 2,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --tp-size 2",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "zimage_turbo_t2i_1024",
            "model": "Tongyi-MAI/Z-Image-Turbo",
            "task": "text-to-image",
            "prompt": "A futuristic cyberpunk city at night, neon lights reflecting on wet streets",
            "width": 1024,
            "height": 1024,
            "seed": 42,
            "num_gpus": 2,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --tp-size 2",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "wan22_t2v_a14b_720p",
            "model": "Wan-AI/Wan2.2-T2V-A14B-Diffusers",
            "task": "text-to-video",
            "prompt": "A cat and a dog baking a cake together in a kitchen.",
            "width": 1280,
            "height": 720,
            "num_frames": 81,
            "seed": 42,
            "num_gpus": 4,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --enable-cfg-parallel --ulysses-degree 2 --text-encoder-cpu-offload --pin-cpu-memory",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "wan22_ti2v_5b_720p",
            "model": "Wan-AI/Wan2.2-TI2V-5B-Diffusers",
            "task": "text-image-to-video",
            "prompt": "The cat starts walking slowly towards the camera.",
            "reference_image": true,
            "width": 1280,
            "height": 720,
            "num_frames": 81,
            "seed": 42,
            "num_gpus": 1,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "ltx2.3_twostage_ti2v_2gpus",
            "model": "Lightricks/LTX-2.3",
            "task": "text-image-to-video",
            "prompt": "The cat starts walking slowly towards the camera.",
            "reference_image": true,
            "width": 768,
            "height": 512,
            "num_frames": 121,
            "seed": 42,
            "num_gpus": 2,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --pipeline-class-name LTX2TwoStagePipeline --cfg-parallel-size 2",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "ideogram4_fp8_t2i_2gpu",
            "model": "ideogram-ai/ideogram-4-fp8",
            "task": "text-to-image",
            "prompt": "A futuristic cyberpunk city at night, neon lights reflecting on wet streets",
            "width": 1024,
            "height": 1024,
            "seed": 42,
            "num_gpus": 2,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --tp-size 2 --attention-backend fa",
                    "extra_env": {}
                }
            }
        },
        {
            "id": "cosmos3_super_t2v_2gpu",
            "model": "nvidia/Cosmos3-Super",
            "task": "text-to-video",
            "prompt": "A cat and a dog baking a cake together in a kitchen.",
            "width": 1280,
            "height": 720,
            "num_frames": 81,
            "seed": 42,
            "num_gpus": 2,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --tp-size 2",
                    "extra_env": {"SGLANG_DISABLE_COSMOS3_GUARDRAILS": "1"}
                }
            }
        },
        {
            "id": "wan22_i2v_a14b_720p",
            "model": "Wan-AI/Wan2.2-I2V-A14B-Diffusers",
            "task": "image-to-video",
            "prompt": "The cat starts walking slowly towards the camera.",
            "reference_image": true,
            "width": 1280,
            "height": 720,
            "num_frames": 81,
            "seed": 42,
            "num_gpus": 4,
            "frameworks": {
                "sglang": {
                    "serve_args": "--warmup --enable-cfg-parallel --ulysses-degree 2 --text-encoder-cpu-offload --pin-cpu-memory",
                    "extra_env": {}
                }
            }
        }
    ]
}