Downloads · 30 days
29
49% of all-time downloads
ygugyg/Minimax-H3-Latent-Upscaler-BF16-MAXQUALITY
Minimax-H3-Latent-Upscaler-BF16-MAXQUALITY is a machine learning model from ygugyg. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
MAXQUALITY re-quantization from FP32 source with conservative mixed-precision fallback.
Downloads · 30 days
29
49% of all-time downloads
All-time downloads
59
Public
Repo size
2.8 GB
Likes
3
Public
Click a slice to open those files.
.pth1.4 GB · 50%
From the Hugging Face model README
MAXQUALITY re-quantization from FP32 source with conservative mixed-precision fallback.
| File | Size | Format |
|---|---|---|
minimax_h3_bf16_CONSERVATIVE_v5.safetensors | ~658 MB | BF16 + FP16 fallback |
| Original BF16 (author) | This v5 | |
|---|---|---|
| Source | FP16 (naive conversion) | FP32 (full precision) |
| Strategy | .to(bfloat16) | Conservative mixed-precision |
| Outlier correction | ❌ None | ❌ Not needed (clean BF16) |
| Mixed precision | ❌ 0% FP16 | ✅ 150 tensors FP16 (norm/head/tail/embed) |
| Conv layers | Distorted | Pure BF16 |
| Artifacts | Flickering, "soap" | None |
Place in: ComfyUI/models/latent_upscale_models/
Node settings:
precision: bf16device: cudatile_size: 256temporal_overlap: 2-4minimax_h3_latent_upscaler_3d_fp32.pth)This is a derivative work of Minimax H3 Latent Upscaler by LBH-123-AI.
License: Apache-2.0
You must comply with the original license terms. No additional restrictions applied.