FrogNano-4B-heretic
4BOtherheretic
from microsoft/FrogNano-4B-2609
↓ 8.1K♥ 13 repos
FrogNano-4B-2609-heretic
FrogNano-4B-2609-heretic is a Heretic-ablated variant of microsoft/FrogNano-4B-2609, published on Hugging Face by richardyoung. It is 4B parameters, single consumer GPU class and mit licence. It is distributed across 5 repositories in Transformers, GGUF and GGUF (imatrix) formats, totalling 5.9K downloads.
Specification
- Model name
- FrogNano-4B-2609-heretic
- Source repository
- richardyoung/FrogNano-4B-2609-heretic
- Base model
- microsoft/FrogNano-4B-2609
- Publisher
- richardyoung
- Parameters
- 4B
- VRAM required
- ~3.2 GB at Q4_K_M, 4k context (estimated) — runs on 4 GB cards
- Hardware class
- 4B – 10B — single consumer GPU
- Licence
- mit
- Task
- Image text to text
- Context window
- 131K tokens (as documented)
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, GGUF (imatrix)
- First indexed
- 15 Oct 2026
- Last updated
- 20 Oct 2026
Ablation technique
How FrogNano-4B-2609-heretic 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
“Abliterated build of microsoft/FrogNano-4B-2609, Microsoft's 4B repository-level coding agent built on Qwen3.5-4B and post-trained only with reinforcement learning on synthetic software-engineering tasks. GGUF quantizations: richardyoung/FrogNano-4B-2609-heretic-GGUF · Ollama: richardyoung/frognano-4b-2609-heretic”
Running it
Commands are templates — confirm the exact repository and quant file before use.
Ollama (publisher-documented)
ollama run richardyoung/frognano-4b-2609-hereticllama.cpp / GGUF
hf download mradermacher/FrogNano-4B-2609-heretic-i1-GGUF --local-dir ./frognano-4b-2609-heretic
llama-cli -m ./frognano-4b-2609-heretic/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "richardyoung/FrogNano-4B-2609-heretic"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve richardyoung/FrogNano-4B-2609-heretic --trust-remote-codeDownloads and variants (4)
Every published repository of FrogNano-4B-2609-heretic, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/FrogNano-4B-2609-heretic-i1-GGUF | GGUF (imatrix) | 3.3K |
| richardyoung/FrogNano-4B-2609-heretic-GGUF | GGUF | 1.6K |
| mradermacher/FrogNano-4B-2609-heretic-GGUF | GGUF | 924 |
| richardyoung/FrogNano-4B-2609-hereticsource | Transformers | 88 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- FrogNano-4B-2609-heretic 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 microsoft/FrogNano-4B-2609, not added by ablation — at 4B parameters, expect a small model’s reasoning and a high error rate on exploit detail.
- The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
- Declared licence is mit, but the terms that bind you are microsoft/FrogNano-4B-2609’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 FrogNano-4B-2609-heretic is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with richardyoung, and does not independently verify publisher claims. See responsible use.