abliteratedmodels.org

KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED

KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED is a refusal-ablated variant of jakeroxs/KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED, published on Hugging Face by jakeroxs. It is 35B total parameters (mixture-of-experts), 48GB VRAM class class and apache-2.0 licence. It is distributed across 5 repositories in GGUF, GGUF (imatrix) and Transformers formats, totalling 6.3K downloads.

Abliteration (directional ablation)35B MoEOther25B – 50B6,3447

Specification

Model name
KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED
Publisher
jakeroxs
Parameters
35B (mixture-of-experts)
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, GGUF (imatrix), Transformers
First indexed
12 Aug 2026
Last updated
6 Sept 2026

Ablation technique

How KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED 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

A merged Hugging Face-format checkpoint combining the Philadelphia Class abliteration of KAT-Coder V2.5 Dev 35B-A3B with the compatible Multi-Token Prediction (MTP) layer from KAT-Coder V2.5 Dev.

Running it

Commands are templates — confirm the exact repository and quant file before use.

llama.cpp / GGUF
hf download jakeroxs/KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED-GGUF --local-dir ./kat-coder-v2.5-dev-35b-a3b-mtp-abliterated
llama-cli -m ./kat-coder-v2.5-dev-35b-a3b-mtp-abliterated/<file>.gguf -p "..."
Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "jakeroxs/KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, torch_dtype="auto", device_map="auto"
)
vLLM
vllm serve jakeroxs/KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED --trust-remote-code

Downloads and variants (4)

Every published repository of KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED, including quantised re-releases by other authors.

Security-team assessment

Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.

  • KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED 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 jakeroxs/KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED, not added by ablation — at 35B 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 jakeroxs/KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED’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 KAT-Coder-V2.5-Dev-35B-A3B-MTP-ABLITERATED is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with jakeroxs, and does not independently verify publisher claims. See responsible use.