Downloads · 30 days
40
23% of all-time downloads
Schattenblume-31B
Nimbz/Schattenblume-31B
Schattenblume-31B is a machine learning model from Nimbz. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
<style .schatten-wrapper { background-color: 06050a; background-image: radial-gradient(circle at 12% 6%, rgba(167, 139, 250, 0.13) 0%, transparent 42%), radial-gradient(circle at 88% 94%, rgba(134, 17, 74, 0.15) 0%, t…
All-time downloads
175
Public
Parameters
31.3B
62.6 GB on disk
Likes
25
Public
Repo makeup
Click a slice to open those files.
.safetensors62.5 GB · 100%
At a glance
- License
- apache-2.0
- Model type
- gemma4
- Access
- Public
- Created
- Aug 16, 2026
- Updated
- Aug 16, 2026
- SHA
- 9d1a417e
Base models
Model card
From the Hugging Face model README
<style>
.schatten-wrapper {
background-color: #06050a;
background-image: radial-gradient(circle at 12% 6%, rgba(167, 139, 250, 0.13) 0%, transparent 42%), radial-gradient(circle at 88% 94%, rgba(134, 17, 74, 0.15) 0%, transparent 46%), radial-gradient(circle at 50% 52%, rgba(21, 61, 46, 0.10) 0%, transparent 62%);
color: #cfc9dd;
font-family: 'Avenir Next', 'Segoe UI', Roboto, Helvetica, Arial, sans-serif;
padding: 42px;
border-radius: 6px;
border: 1px solid #2a1b3d;
line-height: 1.75;
position: relative;
overflow: hidden;
box-shadow: inset 0 0 130px rgba(0,0,0,0.90), 0 20px 55px rgba(0,0,0,0.70);
}
.schatten-wrapper::before {
content: "❀";
position: absolute;
top: -78px; right: -34px;
font-size: 260px;
opacity: 0.07;
color: #e4dcf7;
filter: drop-shadow(0 0 46px rgba(167, 139, 250, 0.6));
pointer-events: none;
z-index: 0;
}
.schatten-wrapper::after {
content: "";
position: absolute;
top: 0; left: 0; right: 0;
height: 3px;
background: linear-gradient(90deg, transparent 0%, #e4dcf7 22%, #a78bfa 52%, #86114a 82%, transparent 100%);
opacity: 0.75;
pointer-events: none;
z-index: 1;
}
.schatten-header { text-align: center; padding-bottom: 24px; position: relative; z-index: 10; }
.schatten-title {
font-size: 3.1rem;
font-weight: 300;
margin: 22px 0 16px 0;
letter-spacing: 12px;
color: #ece7f7;
text-shadow: 0 0 26px rgba(167, 139, 250, 0.55), 0 0 60px rgba(167, 139, 250, 0.22);
}
.schatten-sub {
display: inline-block;
background: transparent;
color: #a78bfa;
border: 1px solid #4c2d72;
text-transform: uppercase;
letter-spacing: 5px;
font-size: 0.78rem;
font-weight: 600;
padding: 5px 16px;
border-radius: 2px;
}
.schatten-tagline { color: #9a8fb3; font-style: italic; font-size: 1.15rem; letter-spacing: 1px; margin-top: 18px; }
.stat-grid { display: grid; grid-template-columns: repeat(3, minmax(0, 1fr)); gap: 18px; margin: 30px 0 6px; position: relative; z-index: 5; }
.stat-item {
text-align: center;
padding: 18px 10px;
background: linear-gradient(180deg, #120c1d 0%, #08060e 100%);
border: 1px solid #3b2757;
border-radius: 2px;
transition: all 0.25s ease;
}
.stat-item:hover { border-color: #a78bfa; box-shadow: 0 0 24px rgba(167, 139, 250, 0.22); }
.stat-value { display: block; font-size: 1.0rem; font-weight: 600; letter-spacing: 2px; color: #ded6f0; }
.schatten-card {
background: linear-gradient(145deg, rgba(18, 12, 29, 0.92), rgba(6, 5, 10, 0.97));
border: 1px solid #291a3c;
border-left: 4px solid #a78bfa;
padding: 26px 30px;
margin: 24px 0;
border-radius: 2px;
box-shadow: 0 10px 30px rgba(0,0,0,0.60);
position: relative;
z-index: 5;
}
.schatten-card h2, .schatten-card h3 { color: #ece7f7; margin-top: 0; letter-spacing: 2px; font-weight: 500; }
.schatten-berry-card {
border-left-color: #86114a;
border-color: #4a0f2c;
background: linear-gradient(145deg, rgba(34, 8, 22, 0.90), rgba(7, 4, 10, 0.97));
box-shadow: inset 0 -2px 14px rgba(134, 17, 74, 0.14), 0 0 26px rgba(134, 17, 74, 0.16);
}
.schatten-berry-card h3 { color: #f0a8c4; }
.schatten-h2 {
color: #ece7f7;
letter-spacing: 4px;
font-weight: 400;
border-bottom: 1px solid #2a1b3d;
padding-bottom: 10px;
position: relative;
z-index: 5;
margin-top: 40px;
}
.schatten-h2 .glyph {
display: inline-block;
color: #a78bfa;
margin-right: 10px;
filter: drop-shadow(0 0 8px rgba(167, 139, 250, 0.6));
}
hr.schatten-rule {
border: 0; height: 1px;
background-image: linear-gradient(90deg, rgba(167, 139, 250, 0.45) 0%, rgba(167, 139, 250, 0.10) 55%, transparent 100%);
margin: 18px 0;
}
.schatten-btns { text-align: center; margin: 10px 0 4px; position: relative; z-index: 5; }
.schatten-btn {
display: inline-block;
font-size: 0.95rem; letter-spacing: 2px; font-weight: 600;
color: #ded6f0 !important;
text-decoration: none !important;
background: #120c1d;
border: 1px solid #4c2d72;
padding: 13px 30px; margin: 8px;
border-radius: 2px;
transition: all 0.25s ease;
}
.schatten-btn:hover { border-color: #a78bfa; color: #ece7f7 !important; box-shadow: 0 0 26px rgba(167, 139, 250, 0.28); }
.schatten-btn.special { border-color: #6d1139; }
.schatten-btn.special:hover { border-color: #c2185b; box-shadow: 0 0 26px rgba(194, 24, 91, 0.28); }
.schatten-table { width: 100%; border-collapse: collapse; background: rgba(0,0,0,0.30); border-radius: 2px; overflow: hidden; border: 1px solid #291a3c; }
.schatten-table td, .schatten-table th { padding: 12px 18px; border-bottom: 1px solid rgba(167, 139, 250, 0.10); font-size: 0.92rem; vertical-align: top; }
.schatten-table th { text-align: left; background: #1c1130; color: #ded6f0; border-bottom: 1px solid #a78bfa; letter-spacing: 2px; font-weight: 500; }
.schatten-table tr:last-child td { border-bottom: none; }
.schatten-table td:first-child { color: #d6ccec; font-weight: 600; white-space: nowrap; }
.schatten-table td:last-child { color: #9a8fb3; font-family: ui-monospace, 'SF Mono', Menlo, Consolas, monospace; font-size: 0.82rem; }
.schatten-table tr:hover { background: rgba(167, 139, 250, 0.05); }
.schatten-gtable { display: grid; grid-template-columns: max-content 1fr; width: 100%; background: rgba(0,0,0,0.30); border: 1px solid #291a3c; border-radius: 2px; overflow: hidden; }
.schatten-gtable > div { padding: 11px 18px; border-bottom: 1px solid rgba(167, 139, 250, 0.10); font-size: 0.92rem; }
.schatten-gtable .vg-head { background: #1c1130; color: #ded6f0; letter-spacing: 2px; font-weight: 500; border-bottom: 1px solid #a78bfa; }
.schatten-gtable .vg-cell { color: #9a8fb3; font-weight: 500; }
.schatten-gtable .vg-val { text-align: right; color: #ded6f0; font-family: ui-monospace, 'SF Mono', Menlo, Consolas, monospace; font-size: 1.02em; }
.schatten-gtable > div:nth-last-child(-n+2) { border-bottom: none; }
.note-box { background: rgba(0,0,0,0.45); padding: 15px 18px; border-radius: 2px; border-left: 3px solid #86114a; margin-top: 16px; color: #cfc9dd; font-size: 0.92rem; }
.note-box strong { color: #f0a8c4; }
.schatten-credits ul { list-style: none; padding-left: 0; }
.schatten-credits li { margin-bottom: 13px; position: relative; padding-left: 28px; color: #cfc9dd; }
.schatten-credits li::before {
content: "❀";
position: absolute; left: 0; top: 0;
color: #a78bfa;
filter: drop-shadow(0 0 6px rgba(167, 139, 250, 0.7));
}
.schatten-credits a, .schatten-wrapper a { color: #c4b5fd; text-decoration: none; border-bottom: 1px dotted #6d4aa8; }
.schatten-credits a:hover, .schatten-wrapper a:hover { color: #f0a8c4; border-color: #f0a8c4; }
.schatten-footer-rule { text-align: center; color: #4c2d72; letter-spacing: 12px; margin: 28px 0 4px; font-size: 1.1rem; }
.summary-style summary { color: #c4b5fd; cursor: pointer; font-weight: 600; letter-spacing: 1px; padding: 8px 0; }
@media (max-width: 640px) { .stat-grid { grid-template-columns: repeat(2, minmax(0, 1fr)); } .schatten-wrapper { padding: 24px; } }
</style>
<div class="schatten-wrapper">
<div class="schatten-header">
<video src="https://huggingface.co/Nimbz/Schattenblume-31B/resolve/main/schattenblume-31b-header.webm" poster="https://huggingface.co/Nimbz/Schattenblume-31B/resolve/main/schattenblume-31b-header.png" width="100%" autoplay loop muted playsinline style="display: block; border-radius: 4px; border: 1px solid #4c2d72; box-shadow: 0 0 50px rgba(167, 139, 250, 0.18); margin: auto; margin-bottom: 6px;">
<img src="https://huggingface.co/Nimbz/Schattenblume-31B/resolve/main/schattenblume-31b-header.png" width="100%" style="border-radius: 4px; border: 1px solid #4c2d72; box-shadow: 0 0 50px rgba(167, 139, 250, 0.18); margin: auto; margin-bottom: 6px;">
</video>
<h1 class="schatten-title">SCHATTENBLUME</h1>
<div class="schatten-sub">31B · Gemma 4 · Merge</div>
<div class="schatten-tagline">"It grows where the light doesn't."</div>
</div>
<div class="stat-grid">
<div class="stat-item"><span class="stat-value">THREE MODELS</span></div>
<div class="stat-item"><span class="stat-value">LESS GEMMAISM</span></div>
<div class="stat-item"><span class="stat-value">SWIPE VARIETY</span></div>
<div class="stat-item"><span class="stat-value">IN-CHARACTER ADHERENCE</span></div>
<div class="stat-item"><span class="stat-value">ROLEPLAY MERGE</span></div>
<div class="stat-item"><span class="stat-value">LOW REFUSALS</span></div>
</div>
<div class="schatten-card">
<h3>❀ About The Model</h3>
<div>Meet Schattenblume-31B, the shadow flower. Leaning on the nightshade, which is <i>Nachtschatten</i> in German. Nightshade is the kind of pretty that doesn't warn you about anything, because it isn't hiding anything either. The berries are glossy and they taste sweet. That's not a trap, that's just what they are. I guess your parents warned you not to eat them.</div>
<hr class="schatten-rule">
<div>Right at the beginning it should be pointed out that this one is made purely because I was asked to and it seemed like more than just one person wanted it, so here we are - initially not my idea, I just made it work. I can only guess that they tested these three models - scotoma-2, Giftige-Blume-v1, MeroMero-v2 - and said they were the good ones for prose and creativity. So the brief was simple: take scotoma-2, Giftige-Blume-v1 and MeroMero-v2, and play all three of them to their strengths. Basically, that's the whole thing - three ingredients this time, plus the base, and one single della_linear phase.</div>
<hr class="schatten-rule">
<div>Before any confusion comes up: This one is roleplay first, actually just like Froopert could be used for RP of *any* kind, but let's just make it blunt this time. I think it's not just the system prompt I'm using; you should be able to expect a model that plays characters closer to how they're written than to how you'd like them to be mid-chat, thanks to the Giftige-Blume, and it will push back if it's fitting. If you want a model that says yes to everything, this isn't it (or you better prompt for it).</div>
</div>
<h2 class="schatten-h2"><span class="glyph">❀</span> GGUF Quants</h2>
<div style="font-size: 0.9rem; color: #9a8fb3; margin: 12px 0 4px;">Thank you to anyone providing quants! Links land here as they show up.</div>
<div class="schatten-btns">
<a class="schatten-btn" href="https://huggingface.co/Nimbz/Schattenblume-31B-GGUF">⬛ STATIC</a>
<a class="schatten-btn special" href="#">◆ IMATRIX (soon)</a>
</div>
<h2 class="schatten-h2"><span class="glyph">☾</span> The Merge</h2>
<div style="font-size: 0.95rem; color: #9a8fb3; margin: 12px 0;">One <b>della_linear</b> pass over the <b>gemma-4-31B-it</b> base, run on <a href="https://github.com/zerofata/mergekit/tree/gemma-4-support">zerofata's gemma-4-support fork of mergekit</a>.</div>
<table class="schatten-table">
<tr><th>Model</th><th>Purpose</th><th>Impacted Layers</th></tr>
<tr><td>scotoma-2</td><td>kills the Gemma tics, holds adherence and format at the output end</td><td>0-59 · 0.30 → 0.62 by depth<br>dips to 0.20 mid on purpose<br>spine (every 6th) 0.80</td></tr>
<tr><td>Giftige-Blume v1</td><td>character portrayal, scene, imagery, the actual prose</td><td>0-59 minus spine · peak 0.56 mid</td></tr>
<tr><td>MeroMero v2</td><td>swipe variety, entity tracking, action-forward writing</td><td>0-59 minus spine · peak 0.30 front-mid<br>down to 0.08 by 59</td></tr>
</table>
<div class="schatten-card schatten-berry-card" style="margin-top: 24px;">
<h3>◆ Some things I paid attention to in this merge:</h3>
<div><b style="color:#ece7f7;">I tried not to blend blindly, but to hit the specific layers where each model shines.</b> Every merge I've done before spread the donors evenly and let the weights sort it out. This time the three curves cross each other on purpose. Blume goes up where the character work happens and scotoma-2 goes down in the same place, then they swap over the back half so the corrector owns the output. MeroMero sits high early and steps out at the end, because the variety belongs where all the character tracking happens. Ultimately it was a design decision, for better or worse, I didn't measure it, just tested it and rolled with it because it seemed to work.</div>
<hr class="schatten-rule">
<div><b style="color:#ece7f7;">It was a bit of a fight to get variety & adherence right.</b> I've found that two of three had reports of being a bit loose on instruction following in their threats, so scotoma-2 is the beeg strong one here. So scotoma-2 comes up for the last layers and the spine, and the two flavour models are pinned to zero on every sixth layer, like usual. If you crunch the numbers you'll see that it's cooked slightly hotter than Froopert, but embeddings, output head, final norm, layer scalars and the vision tower are all filtered out, so everything should be intact.</div>
</div>
<details class="summary-style"><summary>View recipe - della_linear</summary>
merge_method: della_linear
base_model: ./google/gemma-4-31B-it
dtype: bfloat16
tokenizer_source: base
parameters:
normalize: false
epsilon: 0.05
lambda: 1.0
int8_mask: true
models:
- model: ./ReadyArt/gemma-4-31B-it-scotoma-2
parameters:
density: 0.85
weight:
- { filter: embed_tokens, value: 0.0 }
- { filter: lm_head, value: 0.0 }
- { filter: embed_vision, value: 0.0 }
- { filter: vision_tower, value: 0.0 }
- { filter: multi_modal_projector, value: 0.0 }
- { filter: model.language_model.norm., value: 0.0 }
- { filter: layer_scalar, value: 0.0 }
- { filter: model.language_model.layers.5., value: 0.80 }
- { filter: model.language_model.layers.11., value: 0.80 }
- { filter: model.language_model.layers.17., value: 0.80 }
- { filter: model.language_model.layers.23., value: 0.80 }
- { filter: model.language_model.layers.29., value: 0.80 }
- { filter: model.language_model.layers.35., value: 0.80 }
- { filter: model.language_model.layers.41., value: 0.80 }
- { filter: model.language_model.layers.47., value: 0.80 }
- { filter: model.language_model.layers.53., value: 0.80 }
- { filter: model.language_model.layers.59., value: 0.80 }
- value: [0.30, 0.26, 0.20, 0.38, 0.62]
- model: ./Blazed-Forge/Gemma-4-Giftige-Blume-31B-v1
parameters:
density: 0.55
weight:
- { filter: embed_tokens, value: 0.0 }
- { filter: lm_head, value: 0.0 }
- { filter: embed_vision, value: 0.0 }
- { filter: vision_tower, value: 0.0 }
- { filter: multi_modal_projector, value: 0.0 }
- { filter: model.language_model.norm., value: 0.0 }
- { filter: layer_scalar, value: 0.0 }
- { filter: model.language_model.layers.5., value: 0.0 }
- { filter: model.language_model.layers.11., value: 0.0 }
- { filter: model.language_model.layers.17., value: 0.0 }
- { filter: model.language_model.layers.23., value: 0.0 }
- { filter: model.language_model.layers.29., value: 0.0 }
- { filter: model.language_model.layers.35., value: 0.0 }
- { filter: model.language_model.layers.41., value: 0.0 }
- { filter: model.language_model.layers.47., value: 0.0 }
- { filter: model.language_model.layers.53., value: 0.0 }
- { filter: model.language_model.layers.59., value: 0.0 }
- value: [0.30, 0.42, 0.56, 0.42, 0.30]
- model: ./zerofata/G4-MeroMero-v2-31B
parameters:
density: 0.60
weight:
- { filter: embed_tokens, value: 0.0 }
- { filter: lm_head, value: 0.0 }
- { filter: embed_vision, value: 0.0 }
- { filter: vision_tower, value: 0.0 }
- { filter: multi_modal_projector, value: 0.0 }
- { filter: model.language_model.norm., value: 0.0 }
- { filter: layer_scalar, value: 0.0 }
- { filter: model.language_model.layers.5., value: 0.0 }
- { filter: model.language_model.layers.11., value: 0.0 }
- { filter: model.language_model.layers.17., value: 0.0 }
- { filter: model.language_model.layers.23., value: 0.0 }
- { filter: model.language_model.layers.29., value: 0.0 }
- { filter: model.language_model.layers.35., value: 0.0 }
- { filter: model.language_model.layers.41., value: 0.0 }
- { filter: model.language_model.layers.47., value: 0.0 }
- { filter: model.language_model.layers.53., value: 0.0 }
- { filter: model.language_model.layers.59., value: 0.0 }
- value: [0.24, 0.30, 0.30, 0.22, 0.08]
- Type
- gemma4
- License
- apache-2.0
- Created
- Aug 16, 2026
- Updated
- Aug 16, 2026