AkashicPulse EQVAE is a model based on AkashicPulse series, but fine-tuned to utilize <a target="_blank" rel="ugc" href="https://huggingface.co/Anzhc/MS-LC-EQ-D-VR_VAE">Anzhc's EQVAE</a> as its VAE. (In theory,) This improve the model generation color and accuracy.Recommended settings [UPDATED FOR V1.0.x]:<span style="background-color:rgb(26, 27, 30);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">Same like v0.2, the model is in between needing RescaleCFG and not needing it. Expect to turn on and off this a lot.<ul><li>The usage of artist tags IS HIGHLY RECOMMENDED. Put it at the FRONT of your prompt (and use BREAK if your generation platform support it).</li><li>Steps: 32-50</li><li>CFG: 7-10 with RescaleCFG on or off, or 4.5-6 if you don't wanna touch RescaleCFG</li><li>RescaleCFG: 0.7 OR OFF, inspect generated image quality if something is off.</li><li>Sampler: Euler a (or try <a target="_blank" rel="ugc" href="https://github.com/nawka12/adept-sampler">AkashicSolver available on Adept Sampler</a> with AkashicAOS).</li><li>Noise Scale: 0.9, using 1.0 will cause the model to over-generate noise.</li></ul>Recommended negative prompt:<pre><code>worst quality, low quality, bad anatomy, bad hands, text, signature, watermark, username, artist name, [things you want to remove]</code></pre>WARNING: Using excessive negative prompt might resulting the model to lose its quality.Model Specs:v1.0.2<ul><li>Based on: AkashicPulse EQVAE v1.0.1-Base</li><li>Dataset size: ~21k</li><li>Epoch posted: Epoch 10 out of 10</li><li>EQVAE used: <a target="_blank" rel="ugc" href="https://huggingface.co/Anzhc/MS-LC-EQ-D-VR_VAE/resolve/main/EQB7.safetensors">Anzhc's EQVAE B7</a></li></ul>v1.0-v1.0.1<ul><li>Based on: AkashicPulse EQVAE v0.2</li><li>Dataset size: ~4.6k</li><li>Epoch posted: Epoch 15 out of 15</li><li>EQVAE used: <a target="_blank" rel="ugc" href="https://huggingface.co/Anzhc/MS-LC-EQ-D-VR_VAE/resolve/main/EQB7.safetensors">Anzhc's EQVAE B7</a></li></ul>v0.2 [CHECK ABOUT THIS VERSION FOR DETAILS]<ul><li>Based on: AkashicPulse EQVAE v0.1</li><li>Dataset size: ~9.4k</li><li>Epoch posted: Epoch 7 out of 10.</li><li>EQVAE used: <a target="_blank" rel="ugc" href="https://huggingface.co/Anzhc/MS-LC-EQ-D-VR_VAE/resolve/main/EQB7.safetensors">Anzhc's EQVAE B7</a></li></ul>v0.1<ul><li>Based on: <a target="_blank" rel="ugc" href="https://civarchive.com/models/1055452?modelVersionId=2426269">AkashicPulse 4.1</a></li><li>Dataset size: 10.6k, repeated twice.</li><li>Epoch posted: Epoch 9 out of 10.</li><li>EQVAE used: <a target="_blank" rel="ugc" href="https://huggingface.co/Anzhc/MS-LC-EQ-D-VR_VAE/resolve/main/EQB7.safetensors">Anzhc's EQVAE B7</a></li></ul>

AkashicPulse EQVAE

First attempt at adding knowledge to the model, with 32 hours of wall clock training time.Using v1.0.1-Base as base.Dataset used:<ul><li>hololive, 21k images from May 1st, 2025 until March 12th, 2026</li></ul>Donation goal and 5-day early access is purely to support next batch of training. I hope you understand that GPU training time aren't free and I have limited budget for training.

Files with SHA256: 7ea5940747f51058b26f25070ee8dcde5c1ced02587c17b76fb973727e3b6e2e