Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -10,19 +10,57 @@ library_name: diffusers
|
|
| 10 |
|
| 11 |
# MiniMax-H3
|
| 12 |
|
| 13 |
-
Quantised media packages of
|
| 14 |
|
| 15 |
## MECHA Forge
|
| 16 |
|
| 17 |
-
MECHA
|
| 18 |
|
| 19 |
These media packages have been forged and optimized in the MECHA Forge.
|
| 20 |
Each one holds the Denoiser, Text Encoder, and VAE components packaged together for easier deployment.
|
| 21 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
## Available quantizations
|
| 23 |
|
| 24 |
| Quant | Denoiser | Text encoder | VAE |
|
| 25 |
|---|---|---|---|
|
|
|
|
|
|
|
| 26 |
| FP4 | FP4 | FP4 | FP8 |
|
|
|
|
| 27 |
|
| 28 |
The VAE is shared between formats: FP4 reuses the FP8 file, INT4 reuses the INT8 file.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
# MiniMax-H3
|
| 12 |
|
| 13 |
+
Quantised media packages of [MiniMaxAI/MiniMax-H3](https://huggingface.co/MiniMaxAI/MiniMax-H3).
|
| 14 |
|
| 15 |
## MECHA Forge
|
| 16 |
|
| 17 |
+
MECHA \u2014 Model Engineering & Contextual Hearth Assembly
|
| 18 |
|
| 19 |
These media packages have been forged and optimized in the MECHA Forge.
|
| 20 |
Each one holds the Denoiser, Text Encoder, and VAE components packaged together for easier deployment.
|
| 21 |
|
| 22 |
+
## Description
|
| 23 |
+
|
| 24 |
+
MiniMax H3: multimodal (audio+video) generation. 25 steps, CFG 7.0.
|
| 25 |
+
|
| 26 |
## Available quantizations
|
| 27 |
|
| 28 |
| Quant | Denoiser | Text encoder | VAE |
|
| 29 |
|---|---|---|---|
|
| 30 |
+
| FP8 | FP8 | FP8 | FP8 |
|
| 31 |
+
| INT8 | INT8 | INT8 | INT8 |
|
| 32 |
| FP4 | FP4 | FP4 | FP8 |
|
| 33 |
+
| INT4 | INT4 | INT4 | INT8 |
|
| 34 |
|
| 35 |
The VAE is shared between formats: FP4 reuses the FP8 file, INT4 reuses the INT8 file.
|
| 36 |
+
|
| 37 |
+
## Recommended settings
|
| 38 |
+
|
| 39 |
+
From the source model card (adjust to taste):
|
| 40 |
+
|
| 41 |
+
| Parameter | Value |
|
| 42 |
+
|---|---|
|
| 43 |
+
| Steps | 25 |
|
| 44 |
+
| CFG / Guidance scale | CFG-distilled (no CFG) |
|
| 45 |
+
| Resolution | 768p default (2K via regenerate) |
|
| 46 |
+
| Notes | Audio+video; prompt writing guides linked |
|
| 47 |
+
|
| 48 |
+
## Example prompts
|
| 49 |
+
|
| 50 |
+
Caught in the moment (candid, slice of life):
|
| 51 |
+
|
| 52 |
+
`
|
| 53 |
+
masterpiece, best quality, highres, absurdres, very aesthetic, safe, 1woman, solo, candid moment, natural lighting. A recording studio, warm lights, a microphone on a stand, soft acoustic panels. A 29-year-old woman with sharp cheekbones and dark eyes. She has a slender build and a focused expression. She wears a casual sweater and jeans, holding headphones, mid-smile, candid and relaxed.
|
| 54 |
+
`
|
| 55 |
+
|
| 56 |
+
Fashion photography (magazine cover style):
|
| 57 |
+
|
| 58 |
+
`
|
| 59 |
+
Fashion video still: a 29-year-old woman with sharp cheekbones, dark eyes, and full lips; wearing a sequined mini dress, statement earrings; stage lights, dynamic pose, cinematic, 24fps, 8K.
|
| 60 |
+
`
|
| 61 |
+
|
| 62 |
+
Scenery:
|
| 63 |
+
|
| 64 |
+
`
|
| 65 |
+
Scenery: a futuristic cityscape at dusk, neon lights, flying vehicles, holographic billboards, cinematic, 8K.
|
| 66 |
+
`
|