---
title: AkashicPulse-EQVAE
canonical_url: "https://www.modelscope.ai/models/kayfahaarukku/AkashicPulse-EQVAE"
md_url: "https://www.modelscope.ai/models/kayfahaarukku/AkashicPulse-EQVAE.md"
repository: kayfahaarukku/AkashicPulse-EQVAE
last_updated: 2026-03-23
license: other
pipeline_tag: text-to-image-synthesis
tasks:
  - text-to-image-synthesis
parameters: 13.9B
tensor_type:
  - F16
  - F32
  - I64
library_name:
  - safetensors
  - pytorch
frameworks:
  - Pytorch
downloads: 192
stars: 1
tags:
  - fine-tuned
  - illustrious
  - stable-diffusion
  - anime
  - checkpoint
---

# AkashicPulse-EQVAE

> AkashicPulse-EQVAE - An open-source model by kayfahaarukku on ModelScope. AkashicPulse EQVAE

kayfahaarukku/AkashicPulse-EQVAE is a 13.9B-parameter text-to-image-synthesis model on ModelScope. licensed under other.

- **Repository**: kayfahaarukku/AkashicPulse-EQVAE
- **License**: other
- **Tasks**: text-to-image-synthesis
- **Parameters**: 13.9B
- **Tags**: fine-tuned, illustrious, stable-diffusion, anime, checkpoint
- **Downloads**: 192
- **Stars**: 1
- **Last updated**: 2026-03-23

Source: https://www.modelscope.ai/models/kayfahaarukku/AkashicPulse-EQVAE

---

# AkashicPulse EQVAE

A fine-tuned Illustrious-based checkpoint that integrates Anzhc's EQVAE as its VAE to enhance color rendering and accuracy in generated images.

**Creator:** kayfahaarukku | **License:** Illustrious License | **Base Model:** Illustrious

## Files

| File | Version | Description |
|------|---------|-------------|
| `AkashicPulseEQVAE-1.0.1.safetensors` | v1.0.1 | Current release, fixes white background and washed-out generation issues |
| `AkashicPulseEQVAE-1.0.safetensors` | v1.0 | Initial public release |

## Recommended Settings

- **Steps:** 32-50
- **CFG:** 7-10 (with RescaleCFG) or 4.5-6 (without RescaleCFG)
- **RescaleCFG:** 0.7 or OFF (adjust based on output quality)
- **Sampler:** Euler a (or AkashicSolver with AkashicAOS)
- **Noise Scale:** 0.9 (avoid 1.0 to prevent over-generation)

## Prompt Guidance

- Place artist tags at the front of your prompt
- Use `BREAK` separator if your platform supports it
- Apply negative prompts cautiously to maintain quality

**Suggested negative prompt:**
`worst quality, low quality, bad anatomy, bad hands, text, signature, watermark, username, artist name`

## Version History

| Version | Dataset | Epochs | Base |
|---------|---------|--------|------|
| v1.0.1 | ~4.6k | 15/15 | AkashicPulse EQVAE v0.2 |
| v1.0 | ~4.6k | 15/15 | AkashicPulse EQVAE v0.2 |
| v0.2 | ~9.4k | 7/10 | AkashicPulse EQVAE v0.1 |
| v0.1 | 10.6k (2x) | 9/10 | AkashicPulse 4.1 |

All versions use Anzhc's EQVAE B7 VAE.

## Download

With ModelScope Python SDK:
```bash
pip install modelscope
```
```python
from modelscope import snapshot_download
model_dir = snapshot_download('kayfahaarukku/AkashicPulse-EQVAE')
```

With Git:
```
git clone https://www.modelscope.ai/kayfahaarukku/AkashicPulse-EQVAE.git
```

Also available on [CivitAI](https://civitai.com/models/2239459).