Mrbizarro commited on
Commit
2d7abab
·
verified ·
1 Parent(s): c85bb46

README: Phosphene one-click banner + edit support clarified

Browse files
Files changed (1) hide show
  1. README.md +8 -0
README.md CHANGED
@@ -22,8 +22,16 @@ authors:
22
 
23
  > Ported by **[Mrbizarro](https://huggingface.co/Mrbizarro)** · MIT licensed · published to mlx-community
24
 
 
 
 
 
 
 
25
  A 6-bit quantized MLX port of [HiDream-ai/HiDream-O1-Image-Dev](https://huggingface.co/HiDream-ai/HiDream-O1-Image-Dev).
26
 
 
 
27
  ## Sibling repos
28
 
29
  - 🟢 **[BF16 (full precision)](https://huggingface.co/mlx-community/HiDream-O1-Image-Dev-mlx-bf16)** — 17.5 GB, ~16 GB RAM, clean across all dimensions. **Use this when in doubt.**
 
22
 
23
  > Ported by **[Mrbizarro](https://huggingface.co/Mrbizarro)** · MIT licensed · published to mlx-community
24
 
25
+ ## 🎛️ Run it one-click in **[Phosphene](https://github.com/mrbizarro/phosphene)**
26
+
27
+ Phosphene is a free local generative-video panel for Apple Silicon. HiDream is wired into its Image Studio. **[Install Pinokio](https://pinokio.computer)**, then in Pinokio install [Phosphene](https://github.com/mrbizarro/phosphene). Note: Phosphene's HiDream integration uses BF16 by default since **edit requires BF16** — this Q6 repo is for text-to-image-only workflows on RAM-constrained machines.
28
+
29
+ ---
30
+
31
  A 6-bit quantized MLX port of [HiDream-ai/HiDream-O1-Image-Dev](https://huggingface.co/HiDream-ai/HiDream-O1-Image-Dev).
32
 
33
+ **⚠ Q6 does NOT support edit / multi-ref.** Per-group dequantization noise compounds against reference-image features in attention and produces degenerate output. For edit / multi-reference workflows use the [BF16 sibling repo](https://huggingface.co/mlx-community/HiDream-O1-Image-Dev-mlx-bf16) instead.
34
+
35
  ## Sibling repos
36
 
37
  - 🟢 **[BF16 (full precision)](https://huggingface.co/mlx-community/HiDream-O1-Image-Dev-mlx-bf16)** — 17.5 GB, ~16 GB RAM, clean across all dimensions. **Use this when in doubt.**