WiktorMatuszek/MOSS-Image-SFT-FP8-W8A8-Row-Max

MOSS-Image-SFT — FP8-W8A8-Row-Max

Release status

Experimental checkpoint — further testing required. Packed W8A8 weights have passed serialization checks, but native original SFT50 image generation and visual quality for this variant remain unverified in the reported environment. GPU and software kernel support must be tested separately.

Benchmark interpretation: The image gallery and measurements are reference examples for visual inspection, not a required reproducibility package or guarantee of quality, speed, or VRAM usage on another setup.

Original MOSS-Image-SFT DiT with genuinely packed FP8 W8A8, dynamic row weights. The model is quantized from the original SFT checkpoint; there is no extra training, LoRA, distillation or architecture change.

Property	This checkpoint
Quantization	FP8 E4M3 weights and dynamic per-row activation scaling
Coverage	269 / 269 transformer Linear projections (all selected projections quantized)
Stored DiT weights	4.739 GiB, 49.9% smaller than original 9.467 GiB BF16 DiT
Checkpoint files	Three TorchAO/PyTorch .bin weight shards; not generic safetensors
Upstream revision	OpenMOSS-Team/MOSS-Image-SFT, fc5de0d56fc910cbca6f6625a70f5f582b26deb1
Integrity	All 463 original state keys matched; shards reloaded and SHA-256 verified

This repo contains the transformer only. The original Qwen text encoder, tokenizer, FP32 VAE, scheduler and pipeline code are loaded from the pinned upstream revision as needed. The stored DiT size is not total RAM/VRAM or whole-pipeline model size.

Evidence and measured limits

Weights: all 269 quantized Linear weight matrices compared with the original BF16 tensors: parameter-weighted MSE 1.52e-06, SQNR 31.54 dB, cosine 0.999649. These are weight-space errors, not generated-image correctness. Detailed per-layer measurements, quantization receipt, checkpoint integrity.

Original SFT50 image study: Not measured with the original 50-step pipeline and native runtime in the recorded SM80 benchmark. Checkpoint weight verification does not imply that image quality or GPU execution passed.

Runtime qualification: the recorded SM80 setup had no supported native FP8 W8A8 execution for the original SFT50 test. A different runtime or short student-adapter control cannot certify this checkpoint's native performance or original-model image quality. See the measured archive.

Use with original SFT inference

Install a CUDA-compatible PyTorch for your GPU, followed by the repository requirements. The conversion environment was PyTorch 2.7.0+cu128 / TorchAO 0.12.0+cu128; compatibility with packed tensor subclasses and native kernels is device dependent.

hf download WiktorMatuszek/MOSS-Image-SFT-FP8-W8A8-Row-Max --local-dir ./moss-quant
cd moss-quant
python -m pip install -r requirements.txt

# Check all real saved shards before using the model
python generate.py --verify-only

# Original MOSS-SFT: 50 denoising steps, CFG 3, 1024px
python generate.py --prompt-file example_layout.json --output image.png

The launcher downloads the unchanged missing upstream components, not duplicate original BF16 DiT weights. Optional Qwen FP8 encoder / Flux2 VAE FP16 can be selected with --encoder-repo and --vae-repo; that combination needs its own end-to-end quality validation.

Files and limitations

transformer/ contains the quantized checkpoint; generate.py, run_quantized_sft.py, compact_components.py contain the strict pinned loader; QUANTIZATION_RECEIPT.json and COMPLETE.json record checksums. Frozen prompt suites are archived once in the research dataset, not copied into this model repo.

Pixel similarity does not prove correct Polish spelling, photorealism, aesthetic quality or native-kernel support. Test speed and GPU memory on your intended stack. The full cross-format gallery, including explicitly missing measurements, is maintained once in the benchmark dataset.

Attribution: independently quantized from OpenMOSS-Team/MOSS-Image-SFT at fc5de0d56fc910cbca6f6625a70f5f582b26deb1; original Apache-2.0 license in LICENSE. Not an official OpenMOSS-Team release.