Cyber-Tiel-Coder-35B-A3B
35BOtheruncensored
from huihui-ai/Huihui-Ornith-1.5-35B-A3B-abliterated
↓ 1M♥ 3433 repos
Rumeni-26B-A4B
Rumeni-26B-A4B is a Heretic-ablated variant of SubMaroon/Rumeni-26B-A4B, published on Hugging Face by SubMaroon. It is 26B total parameters (mixture-of-experts), 48GB VRAM class class and apache-2.0 licence. It is distributed across 4 repositories in Transformers, GGUF and GGUF (imatrix) formats, totalling 6.5K downloads.
Specification
- Model name
- Rumeni-26B-A4B
- Source repository
- SubMaroon/Rumeni-26B-A4B
- Base model
- SubMaroon/Rumeni-26B-A4B
- Publisher
- SubMaroon
- Parameters
- 26B (mixture-of-experts)
- VRAM required
- ~16.6 GB at Q4_K_M, 4k context (estimated) — runs on 24 GB cards
- Hardware class
- 25B – 50B — 48GB VRAM class
- Licence
- apache-2.0
- Task
- Unspecified
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, GGUF (imatrix)
- First indexed
- 17 Oct 2026
- Last updated
- 27 Oct 2026
Ablation technique
How Rumeni-26B-A4B was modified, and what that implies.
Heretic
Automated directional ablation via the Heretic toolchain, which searches for the refusal direction and applies it with a KL-divergence budget so general capability is preserved.
Reported scope: Not stated by the publisher. Publishers frequently omit this, so verify behaviour empirically rather than assuming full coverage.
From the publisher’s model card
“A model for roleplay in Russian language. A LoRA on attention OV and the dense MLP, trained on a Russian RP dataset and baked into my base StyleTune-QK-Heretic.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Rumeni-26B-A4B-i1-GGUF --local-dir ./rumeni-26b-a4b
llama-cli -m ./rumeni-26b-a4b/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "SubMaroon/Rumeni-26B-A4B"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve SubMaroon/Rumeni-26B-A4B --trust-remote-codeDownloads and variants (3)
Every published repository of Rumeni-26B-A4B, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/Rumeni-26B-A4B-i1-GGUF | GGUF (imatrix) | 5.1K |
| mradermacher/Rumeni-26B-A4B-GGUF | GGUF | 1.3K |
| SubMaroon/Rumeni-26B-A4Bsource | Transformers | 82 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Rumeni-26B-A4B will attempt offensive-security prompts that a hosted commercial model declines, which is what makes it usable for red-team corpus generation and for measuring what an unaligned model of this class produces.
- Capability is inherited from SubMaroon/Rumeni-26B-A4B, not added by ablation — at 26B parameters, expect roughly the base model’s competence, minus some instruction-following fidelity.
- The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
- Declared licence is apache-2.0, but the terms that bind you are SubMaroon/Rumeni-26B-A4B’s — a re-release cannot grant more than its parent.
Handling: run it isolated, put moderation in front of it if anyone outside your team can reach it, and verify the weights before loading — these are third-party artifacts. Full handling and licence guidance.
Metadata for Rumeni-26B-A4B is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with SubMaroon, and does not independently verify publisher claims. See responsible use.