Ternary-Bonsai-2-27B-Uncensored-Heretic
27BOtherheretic
from prism-ml/Ternary-Bonsai-2-27B-gguf
↓ 131.1K♥ 1542 repos
Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0
Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0 is a refusal-ablated variant of prism-ml/Ternary-Bonsai-2-27B-gguf, published on Hugging Face by BoldingBuilds. It is 27B parameters, 48GB VRAM class class and apache-2.0 licence. It is distributed across 4 repositories in GGUF and MLX formats, totalling 49.5K downloads.
Specification
- Model name
- Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0
- Source repository
- BoldingBuilds/Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-GGUF
- Base model
- prism-ml/Ternary-Bonsai-2-27B-gguf
- Publisher
- BoldingBuilds
- Parameters
- 27B
- VRAM required
- ~17.2 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
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, MLX
- First indexed
- 26 Sept 2026
- Last updated
- 1 Nov 2026
Ablation technique
How Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0 was modified, and what that implies.
Abliteration (directional ablation)
The single residual-stream direction that mediates refusal is identified from harmful/harmless prompt pairs and projected out of the model weights.
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
“v2 is built to be run with thinking on. Abliterated Bonsai 2 builds, my own v1 included, tend to write the whole answer inside their reasoning and run out of budget before handing it over. v2 changes one row of the output layer so the model closes its reasoning once the answer is ready. With reasoningeffort set to...”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download BoldingBuilds/Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-GGUF --local-dir ./ternary-bonsai-2-27b-abliterated-v2-pq2-0
llama-cli -m ./ternary-bonsai-2-27b-abliterated-v2-pq2-0/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "BoldingBuilds/Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-GGUF"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve BoldingBuilds/Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-GGUF --trust-remote-codeDownloads and variants (3)
Every published repository of Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| BoldingBuilds/Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-GGUFsource | GGUF | 45.4K |
| 2001Y/Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-MLX | MLX | 2.6K |
| timfromhcs/Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-GGUF | GGUF | 1.6K |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0 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 prism-ml/Ternary-Bonsai-2-27B-gguf, not added by ablation — at 27B 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 prism-ml/Ternary-Bonsai-2-27B-gguf’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 Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0 is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with BoldingBuilds, and does not independently verify publisher claims. See responsible use.