Downloads · 30 days
96
17% of all-time downloads
MorinoNushi/GLM-5.3-Flash-Heretic-LoRA-V1-GGUF
GLM-5.3-Flash-Heretic-LoRA-V1-GGUF is a machine learning model from MorinoNushi. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as mit.
⚠️ Content warning: This adapter has had the base model's refusal behavior surgically suppressed. The resulting model will comply with requests the base model refuses, including requests that are harmful, unethical, o…
Downloads · 30 days
96
17% of all-time downloads
All-time downloads
574
Public
Repo size
29.2 MB
Likes
7
Public
Click a slice to open those files.
.gguf29.2 MB · 100%
From the Hugging Face model README
⚠️ Content warning: This adapter has had the base model's refusal behavior surgically suppressed. The resulting model will comply with requests the base model refuses, including requests that are harmful, unethical, offensive, or illegal. It has reduced safety guardrails. See Responsible use below — you are solely responsible for what you do with it.
This is a rank-1 LoRA adapter that decensors / "abliterates" GLM-5.3-Flash (320B total / 18B active MoE, MIT license), produced with heretic-gguf — a GGUF-native port of Heretic's Optuna-optimized directional ablation, which runs the whole search directly on quantized GGUF weights via llama.cpp.
Early release. The Optuna study behind this adapter is still running. This first adapter (trial 8) already cuts the refusal rate from 95% to ~26% at very low drift, but it is not the finished product — expect stronger configurations. Additional LoRA adapters covering more points on the refusal/KL tradeoff, and fully merged, ready-to-run GGUFs, will be uploaded to this repository as the study matures.
This repository contains only the adapter. You need the base model separately — any GGUF quant of GLM-5.3-Flash works, since the adapter is applied in f32/f16 compute regardless of the base quant (it was tuned and evaluated against UD-IQ4_XS). The LoRA form is the lossless option: the base weights are never modified or requantized, and the download is ~28 MB instead of ~160 GB.
heretic-gguf is available at github.com/MoriNoNushi/heretic-gguf — the full tool, so the method can be applied to other GGUF models.
Measured on a held-out eval set of 140 harmful prompts (100 from
mlabonne/harmful_behaviors test + 40 custom) and 100 harmless prompts
(mlabonne/harmless_alpaca test), greedy decoding, 100-token responses,
against the UD-IQ4_XS base:
| Refusal rate (harmful) | KL divergence (harmless) | |
|---|---|---|
| Base model | 95.00% (133/140) | 0 (by definition) |
| glm-5.3-heretic-lora.gguf | 26.43% (37/140) | 0.0682 |
Refusals are counted by refusal-keyword matching (English + Chinese markers); KL divergence is measured on first-token logits on harmless prompts. The study is single-stage — every trial is scored on the full eval set, so these numbers are exactly what the optimizer saw.
Note on KL: the KL divergence above (and the optimization objective itself) was measured against the UD-IQ4_XS quant. KL is a baseline-relative metric, so if you run the adapter on a different quant, the effective drift from that quant's baseline may differ.
llama-server \
-m GLM-5.3-Flash-UD-IQ4_XS-00001-of-00005.gguf \
--lora glm-5.3-heretic-lora.gguf \
--jinja
Add your usual offload/context flags (-ngl 999, -c, tensor splits,
etc.) — nothing model-specific is required, and no special sampling
parameters are needed. Simply omitting --lora restores the base model
exactly.
llama.cpp requirements. GLM-5.3-Flash (glm5next) support had not
landed on llama.cpp master at the time of this release — use a build of
PR #27754 (or master,
once merged). In addition, the stock PR routes the KDA layers' attention
out-projection through a raw ggml_mul_mat, which silently ignores LoRA on
those tensors — 31 of the 45 layers. The adapter was tuned and evaluated
with a one-line patch that fixes this
(llama_cpp_glm5next_lora.patch, shipped in the
heretic-gguf repo); apply it
for the adapter's attention ablation to take full effect. Without the patch
the adapter still applies to the MLP experts and the MLA layers, but its
effect is weaker than the measured numbers above.
glm53flash.glm53flash, trial 8; per-layer direction
scope; per-expert strengths scaled by measured harmful/harmless routing
frequency; row_normalization = "pre"):
adapter.heretic.* GGUF metadata keys; inspect with
strings glm-5.3-heretic-lora.gguf | grep adapter.heretic.The base model is MIT-licensed (see the base repo); this adapter inherits those terms. The heretic-gguf tooling used to produce it is AGPL-3.0-or-later.