File size: 3,757 Bytes
648f1ca b299e7b 648f1ca b299e7b 648f1ca |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 |
---
license: mit
library_name: sd15-flow-trainer
tags:
- geometric-deep-learning
- stable-diffusion
- ksimplex
- pentachoron
- flow-matching
- cross-attention-prior
base_model: sd-legacy/stable-diffusion-v1-5
pipeline_tag: text-to-image
---
# KSimplex Geometric Attention Prior
Geometric cross-attention prior for SD1.5 using pentachoron (4-simplex) structures.
# Before and After
## Pretrain
## Final
## Architecture
| Component | Params |
|-----------|--------|
| SD1.5 UNet (frozen) | 859,520,964 |
| **Geo prior (trained)** | **4,845,725** |
The geometric prior modulates CLIP encoder hidden states through
4-layer stacked k-simplex attention before they reach
the 16 cross-attention blocks in the UNet.
## Simplex Configuration
| Parameter | Value |
|-----------|-------|
| k (simplex dim) | 4 |
| Embedding dim | 32 |
| Feature dim | 768 |
| Stacked layers | 4 |
| Attention heads | 8 |
| Base deformation | 0.25 |
| Residual blend | learnable |
| Timestep conditioned | True |
## Usage
```python
from sd15_trainer_geo.pipeline import load_pipeline, load_geo_from_hub
# Load base SD1.5 + fresh geo prior
pipe = load_pipeline()
# Load trained geo weights from this repo
load_geo_from_hub(pipe, "AbstractPhil/sd15-geoflow-characters")
# Or one-shot: load base + geo in one call
pipe = load_pipeline(geo_repo_id="AbstractPhil/sd15-geoflow-characters")
```
## Training Info
- **dataset**: AbstractPhil/synthetic-characters (schnell_simple_2)
- **samples**: 50000
- **epochs**: 1
- **steps**: 8333
- **shift**: 2.5
- **base_lr**: 5e-05
- **min_snr_gamma**: 5.0
- **cfg_dropout**: 0.1
- **batch_size**: 6
- **geo_loss_weight**: 0.01
- **loss_final**: 0.3177722838521004
# Assessment
## License
MIT — [AbstractPhil](https://huggingface.co/AbstractPhil)
|