johndpope commited on
Commit
aeef4ad
·
verified ·
1 Parent(s): 867ee34

Card: region head coming, LoRA run stopped at step 500

Browse files
Files changed (1) hide show
  1. README.md +3 -2
README.md CHANGED
@@ -14,7 +14,7 @@ tags:
14
  # MiniMax H3 two-track mixed resolution (MPEG-Diffusion)
15
 
16
  > [!WARNING]
17
- > **Research weights, help wanted.** The first training run is in progress. There is no quality result yet. Faces from the untrained sampler stay at half-resolution quality and can come out distorted. Training is meant to fix that. Help and results: [scrya-com/mpeg-diffusion issue #2](https://github.com/scrya-com/mpeg-diffusion/issues/2).
18
 
19
  > [!CAUTION]
20
  > **License and territory.** Derived from MiniMax H3 under the [MiniMax H3 Community License Agreement](LICENSE) (see [`NOTICE`](NOTICE)). Use and redistribution are allowed **only outside the European Union, the United Kingdom, the Republic of Korea and the United States of America**. Commercial use above US$20M yearly revenue needs MiniMax's written authorization. Do not use these weights to impersonate anyone, and disclose machine-generated output.
@@ -25,7 +25,8 @@ How it works: the unmodified H3 renders the whole frame at half resolution. A re
25
 
26
  | Folder | What |
27
  |---|---|
28
- | [`run1/`](run1) | First run: 1,951 Pexels photos of people and 403 isometric stills at 1344×768, 2,000 steps on one RTX 5090. Its card has the held-out loss. Appears when training finishes. |
 
29
 
30
  - Code, sampler and training pipeline: [scrya-com/mpeg-diffusion](https://github.com/scrya-com/mpeg-diffusion) (`lot/two_track_h3.py`, `lot/train_tt.py`, one command: `lot/train_tt_all.sh`).
31
  - Precomputed caption embeddings for the Pexels photos, which skip the 32B text encoder: [johndpope/MiniMax-H3-pexels-people-text](https://huggingface.co/datasets/johndpope/MiniMax-H3-pexels-people-text).
 
14
  # MiniMax H3 two-track mixed resolution (MPEG-Diffusion)
15
 
16
  > [!WARNING]
17
+ > **Research weights, help wanted.** The region head is published once its current run finishes. The LoRA is not published yet: its first run was stopped at step 500 to free the GPU for the region head. There is no image-quality result yet. Faces from the untrained sampler stay at half-resolution quality and can come out distorted. Training is meant to fix that. Help and results: [scrya-com/mpeg-diffusion issue #2](https://github.com/scrya-com/mpeg-diffusion/issues/2).
18
 
19
  > [!CAUTION]
20
  > **License and territory.** Derived from MiniMax H3 under the [MiniMax H3 Community License Agreement](LICENSE) (see [`NOTICE`](NOTICE)). Use and redistribution are allowed **only outside the European Union, the United Kingdom, the Republic of Korea and the United States of America**. Commercial use above US$20M yearly revenue needs MiniMax's written authorization. Do not use these weights to impersonate anyone, and disclose machine-generated output.
 
25
 
26
  | Folder | What |
27
  |---|---|
28
+ | [`region_head/`](region_head) | The self-selected region head: one linear layer on block 25 of the frozen DiT that marks the subject partway through sampling, and so sets the full-resolution region and the Level-of-Token layout. Trained on 2,354 pictures with YOLOE subject labels. Its card has the held-out mIoU and the limits. Uploaded when training finishes. |
29
+ | LoRA (not published) | Lets the full-resolution track start earlier. First run (2,346 pairs, RTX 5090) stopped at step 500 with held-out loss down 21% (half resolution) and 26% (full resolution). |
30
 
31
  - Code, sampler and training pipeline: [scrya-com/mpeg-diffusion](https://github.com/scrya-com/mpeg-diffusion) (`lot/two_track_h3.py`, `lot/train_tt.py`, one command: `lot/train_tt_all.sh`).
32
  - Precomputed caption embeddings for the Pexels photos, which skip the 32B text encoder: [johndpope/MiniMax-H3-pexels-people-text](https://huggingface.co/datasets/johndpope/MiniMax-H3-pexels-people-text).