Instructions to use ZhengmingYu/DMAD with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use ZhengmingYu/DMAD with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline from diffusers.utils import export_to_video # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("MiniMaxAI/MiniMax-H3", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("ZhengmingYu/DMAD") prompt = "A man with short gray hair plays a red electric guitar." output = pipe(prompt=prompt).frames[0] export_to_video(output, "output.mp4") - Inference
- Notebooks
- Google Colab
- Kaggle
Model card: sync the 8-step checkpoint row with the GitHub README
Browse files
README.md
CHANGED
|
@@ -76,7 +76,7 @@ Rank-128 LoRAs on the H3 transformer that turn the 50-step teacher into a 4-step
|
|
| 76 |
|------|------------|------|
|
| 77 |
| `minimax_h3/dmad_minimax_h3_4step_lora_critic.safetensors` | the checkpoint of the paper: EMA of the student at iteration 800 of the main run | 1.4 GB |
|
| 78 |
| `minimax_h3/dmad_minimax_h3_4step_full_critic.safetensors` | the student of a run whose critic backbone is fully trained (the paper's run keeps it frozen under a LoRA): iteration 1600, live weights; it scores higher on AVGen-Bench | 1.4 GB |
|
| 79 |
-
| `minimax_h3/dmad_minimax_h3_8step_full_critic.safetensors` | the 8-step student: the `full_critic` recipe trained with 8-step sampling from scratch
|
| 80 |
| `minimax_h3/dmad_minimax_h3_4step_lora_critic_comfyui.safetensors` | `lora_critic` in ComfyUI's MiniMax-H3 key layout (exact conversion) | 2.0 GB |
|
| 81 |
| `minimax_h3/dmad_minimax_h3_4step_full_critic_comfyui.safetensors` | `full_critic` in ComfyUI's MiniMax-H3 key layout (exact conversion) | 2.0 GB |
|
| 82 |
| `minimax_h3/dmad_minimax_h3_8step_full_critic_comfyui.safetensors` | `8step_full_critic` in ComfyUI's MiniMax-H3 key layout (exact conversion); 8 steps in the scheduler | 2.0 GB |
|
|
|
|
| 76 |
|------|------------|------|
|
| 77 |
| `minimax_h3/dmad_minimax_h3_4step_lora_critic.safetensors` | the checkpoint of the paper: EMA of the student at iteration 800 of the main run | 1.4 GB |
|
| 78 |
| `minimax_h3/dmad_minimax_h3_4step_full_critic.safetensors` | the student of a run whose critic backbone is fully trained (the paper's run keeps it frozen under a LoRA): iteration 1600, live weights; it scores higher on AVGen-Bench | 1.4 GB |
|
| 79 |
+
| `minimax_h3/dmad_minimax_h3_8step_full_critic.safetensors` | the 8-step student: the `full_critic` recipe trained with 8-step sampling from scratch. Sample it with 8 steps (`--steps 8`) | 1.4 GB |
|
| 80 |
| `minimax_h3/dmad_minimax_h3_4step_lora_critic_comfyui.safetensors` | `lora_critic` in ComfyUI's MiniMax-H3 key layout (exact conversion) | 2.0 GB |
|
| 81 |
| `minimax_h3/dmad_minimax_h3_4step_full_critic_comfyui.safetensors` | `full_critic` in ComfyUI's MiniMax-H3 key layout (exact conversion) | 2.0 GB |
|
| 82 |
| `minimax_h3/dmad_minimax_h3_8step_full_critic_comfyui.safetensors` | `8step_full_critic` in ComfyUI's MiniMax-H3 key layout (exact conversion); 8 steps in the scheduler | 2.0 GB |
|