Downloads · 30 days
0
sensenova-u1-8b-infographic-v3-fast-mlx
SceneWorks/sensenova-u1-8b-infographic-v3-fast-mlx
sensenova-u1-8b-infographic-v3-fast-mlx is a text-to-image model from SceneWorks. Use it when you need an image from a text prompt. It is set up for mlx. The card lists the license as apache-2.0.
The 8-step distilled variant of sensenova/SenseNova-U1-8B-MoT-Infographic-V3, hosted for the SceneWorks app. The V1 8-step distill LoRA (sensenova/SenseNova-U1-8B-MoT-LoRAs, SenseNova-U1-8B-MoT-LoRA-8step-V1.0.safeten…
Access
Public
Updated Jul 19, 2026
Repo size
66.8 GB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.safetensors66.8 GB · 100%
At a glance
- Task
- Text-to-Image
- Library
- mlx
- License
- apache-2.0
- Access
- Public
- Created
- Jul 18, 2026
- Updated
- Jul 19, 2026
- SHA
- 660506e9
Base models
Model card
From the Hugging Face model README
SenseNova-U1 8B Infographic V3 Fast — MLX quant-matrix (SceneWorks re-host)
The 8-step distilled variant of sensenova/SenseNova-U1-8B-MoT-Infographic-V3,
hosted for the SceneWorks app. The V1 8-step distill LoRA (sensenova/SenseNova-U1-8B-MoT-LoRAs,
SenseNova-U1-8B-MoT-LoRA-8step-V1.0.safetensors) is pre-merged into the generation path before packing (296 gen-path targets merged),
so the loader never re-merges (a distill_merged.json marker signals it is baked in).
Tiers (each subdir is a self-contained load root, distill-merged)
q4/(default) — packed Q4 backbone (~11.8 GB)q8/— packed Q8 backbone (~19.9 GB)bf16/— merged dense checkpoint (~35 GB)
The loader packed-detects ({base}.scales) — no in-app re-quantize.
Provenance
- Source:
sensenova/SenseNova-U1-8B-MoT-Infographic-V3(Apache-2.0) + distill LoRAsensenova/SenseNova-U1-8B-MoT-LoRAs - Converter:
mlx-gen-sensenova::convert::prequantize_fast_turnkey(SceneWorks/inference) - Default inference: 8 steps, CFG 1.0, timestep-shift 3.0
License: Apache-2.0 (inherits the source license).
- Task
- Text-to-Image
- Library
- mlx
- License
- apache-2.0
- Created
- Jul 18, 2026
- Updated
- Jul 19, 2026