Sage-A1.5-Ablitrex-abliterated
Other
from InternScience/Sage-A1.5
↓ 6.4K♥ 23 repos
Sage-A1.5-Heretic-Abliterated-Uncensored
Sage-A1.5-Heretic-Abliterated-Uncensored is a Heretic-ablated variant of InternScience/Sage-A1.5, published on Hugging Face by sam33339999. It is apache-2.0 licence. It is distributed across 5 repositories in GGUF (imatrix), GGUF, Transformers and EXL3 formats, totalling 14.1K downloads.
Specification
- Model name
- Sage-A1.5-Heretic-Abliterated-Uncensored
- Source repository
- sam33339999/Sage-A1.5-Heretic-Abliterated-Uncensored
- Base model
- InternScience/Sage-A1.5
- Publisher
- sam33339999
- Parameters
- Not determinable from name
- VRAM required
- Unknown — no published parameter count
- Hardware class
- Unknown
- Licence
- apache-2.0
- Task
- Image text to text
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF (imatrix), GGUF, Transformers, EXL3
- First indexed
- 22 Oct 2026
- Last updated
- 23 Oct 2026
Ablation technique
How Sage-A1.5-Heretic-Abliterated-Uncensored 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
“Heretic 由 Philipp Emanuel Weidmann 開發。它把 Arditi 等人（2024）的 abliteration，和 Optuna 的 TPE 參數搜尋合在一起：同時壓低拒答次數，以及相對原模型的 KL 散度，盡量保留原模型能力。工具說明見 Heretic README。”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Sage-A1.5-Heretic-Abliterated-Uncensored-i1-GGUF --local-dir ./sage-a1.5-heretic-abliterated-uncensored
llama-cli -m ./sage-a1.5-heretic-abliterated-uncensored/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "sam33339999/Sage-A1.5-Heretic-Abliterated-Uncensored"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve sam33339999/Sage-A1.5-Heretic-Abliterated-Uncensored --trust-remote-codeDownloads and variants (4)
Every published repository of Sage-A1.5-Heretic-Abliterated-Uncensored, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/Sage-A1.5-Heretic-Abliterated-Uncensored-i1-GGUF | GGUF (imatrix) | 8.7K |
| mradermacher/Sage-A1.5-Heretic-Abliterated-Uncensored-GGUF | GGUF | 5.2K |
| sam33339999/Sage-A1.5-Heretic-Abliterated-Uncensoredsource | Transformers | 153 |
| sam33339999/Sage-A1.5-Heretic-Abliterated-Uncensored-exl3-4.0bpw | EXL3 | 62 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Sage-A1.5-Heretic-Abliterated-Uncensored 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.
- 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 InternScience/Sage-A1.5’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 Sage-A1.5-Heretic-Abliterated-Uncensored is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with sam33339999, and does not independently verify publisher claims. See responsible use.