在无FP8核心的消费级显卡上，实现93亿参数模型的高质量8比特量化。
Holding the FP8 Quality Ceiling at 8-Bit Weights and Activations: INT8 and GGUF Post-Training Quantization of Ideogram 4.0 for Consumer GPUs
- 采用分通道权重、动态激活与平滑量化，实现接近FP8精度的8比特推理。
- INT8方案在图像质量与保真度上显著优于4比特基线，仅差0.004人类偏好分。
- 发布可落地的GGUF Q4_K版本，兼顾性能与存储，适合本地部署使用。
本文研究了93亿参数流匹配扩散Transformer（DiT）Ideogram 4.0在无FP8张量核心的安培架构RTX 3090 GPU上的后训练量化（PTQ）。该模型基于双权重副本实现无分类器引导，并由Qwen3-VL文本编码器驱动。为验证性能，所有变体均在由大模型生成的符合结构化JSON规范的提示下评估，采用涵盖人类偏好（HPSv2）、CLIP与PickScore的图像质量评分，以及PP-OCR精确匹配与编辑距离的文本准确性，还有与最高精度公开检查点（FP8）对比的PSNR/SSIM/LPIPS保真度指标。在包含300个提示的基准测试中，采用分通道权重、逐标记动态激活、平滑量化及小高敏感层bf16保护的INT8 W8A8方案，在CLIP与PickScore上与FP8统计无差异（配对置信区间含零），人类偏好分差距仅约0.004；其8比特大小下对FP8输出的保真度最优（LPIPS 0.243，优于半尺寸4比特基线的0.277/0.306，且与INT8-Q4_K差距不包含零）。而同体积的GGUF Q4_K方案达到与原生NF4基线相当的独立图像质量，是质量-内存权衡中的帕累托最优选择。进一步发现：在结构化提示下，四类量化方案在整体图像质量上达一致，差异集中于保真度与文本渲染；文本可读性从原始字符串下的近零提升至55%的OCR精确匹配。所提INT8 W8A8与GGUF Q4_K模型已通过非商业授权发布于Hugging Face。
原文摘要 · Abstract (English)
We study post-training quantization (PTQ) of Ideogram 4.0, a 9.3B flow-matching diffusion transformer (DiT) that realizes classifier-free guidance with two separate-weight copies of a single-stream backbone and is conditioned by a Qwen3-VL text encoder, targeting Ampere RTX~3090 GPUs, which lack FP8 tensor cores. Because Ideogram~4.0 is trained on structured JSON captions, we evaluate every variant under schema-valid JSON prompts produced by an LLM expander built to Ideogram's published caption specification, and score them with a battery spanning human-preference (HPSv2), CLIP, and PickScore for standalone quality; PP-OCR exact-match and edit distance for text; and PSNR/SSIM/LPIPS for fidelity to the FP8 reference (the highest-precision public checkpoint) output. On a 300-prompt benchmark with paired bootstrap confidence intervals, an INT8 W8A8 recipe (per-channel weights, per-token dynamic activations, SmoothQuant, and bf16 protection of a small high-fragility layer set) is statistically indistinguishable from FP8 on CLIP and PickScore (paired CIs include zero) and within ~0.004 HPSv2, and, at its 8-bit size, is the most faithful reproduction of the FP8 output (LPIPS 0.243 vs 0.277/0.306 for the half-size 4-bit baselines; the INT8-Q4_K gap excludes zero). A GGUF Q4_K quantization reaches the same standalone quality as the published NF4 baseline at the same on-disk size, making it the Pareto choice on the quality-memory frontier. We further show that under JSON prompts all four variants reach parity on standalone quality, the variants separate on fidelity and text rendering, not on aggregate image-quality scores, and that text legibility, near-zero when the model is prompted with raw strings, reaches 55% OCR exact-match under the JSON captions it expects. We release the INT8 W8A8 and GGUF Q4_K quantized weights on Hugging Face under a gated, non-commercial license.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。