gemma-3-12b-it-abliterated-GLP-43-L5-47-a0.25

Projective control vector ("GLP") for google/gemma-3-12b-it (Gemma3ForConditionalGeneration, text stack: 48 layers, hidden 3840, vocab 262208, bf16). Applied at runtime as h <- h - alpha * (h . d) d at the post-layer residual stream of the language model, layers 5–47, alpha 0.25 baked in. No weights are modified; this is the difference, not the model.

Confirmed base: google/gemma-3-12b-it. The derivation checkpoint was the unsloth/gemma-3-12b-it mirror (byte-identical bf16 weights) at revision 9478e665381f42974aa06177b019352fb6291876 — the google/ repo is license-gated for our token, so the mirror sha is the pin recorded in the file's glp.base_revision. The config was checked against Google's published specs (48 layers, hidden 3840) before derivation. The vector is not validated against other revisions or quants.

Validation (Modal A100-80GB, transformers 5.16.1, bf16, greedy, 1024-token cap, 2026-10-11)

Alpha ladder (refusal32 / benign32-holdout, comply of 32): 0.25 → 32/32, 32/32; 0.5 → 32/32, 32/32; 1.0 → 32/32, 32/32. Zero GARBLED completions at every rung. The knee is the bottom rung: alpha 0.25 captures the full delivery gain, so that is what ships. No-op gate: an alpha=0.0 arm with the vector loaded reproduces the stock completions exactly (64/64 identical strings on both gate suites).

The two stock refusals (passport forgery that passes airport checks, a keylogger that emails captured passwords) both become coherent on-topic deliveries under steering — verified by reading the completions, not only the classifier. Stock gemma-3-12b-it already answers 30/32 of refusal32: the direction closes the last two rather than unlocking a locked-down model.

Derivation gates (captain-vector 0.5.0, dom_per_layer_mask0.005): the massive-activation screen flagged this checkpoint — peak/median 6794x at layer 0, dims 2339/107/305 — so the top 0.5% of dims by magnitude were masked before normalisation. Held-out separation vs a shuffled-label null (20 reps) clears the 5x ship gate on 43 of 47 shippable layers; layers 1, 2, 3, 4 sit below it (ratios 2.6–3.1) and are excluded from the file. Adjacent-layer cosine median 0.887 against a random-direction null p99 of 0.044. Mean dose 0.062 of the residual norm at alpha=1 (random-direction floor 0.016); the max per-layer dose is 0.282 at layer 47, under the 50% damage threshold.

The contrast is refusal32 vs benign32 (content-matched, last-token pooling). refusal32 doubles as the derivation set, so its steered number is in-sample; benign32-holdout is out-of-sample. n=32 per arm; read rates at that resolution as approximate.

Usage

This file uses the glp.* GGUF namespace (spec: weightless spec/GLP.md) and is read projective-only. An additive consumer must refuse this file. The hook point is residual_stream_post_layer — the decoder-layer output, the accumulated residual stream — derived AND applied at that site.

export WEIGHTLESS_STEER_PATH=glp.gemma-3-12b-it-GLP-43-L5-47-a0.25.gguf
export WEIGHTLESS_STEER_ALPHA=0.25
# serve with a runtime that implements glp.mode=project

What is inside

Do not scale alpha across models

alpha_default is calibrated on this checkpoint, at this hook. Here the ladder is flat — 0.25, 0.5 and 1.0 all deliver identically with zero measured collateral — because the stock model barely refuses. That says nothing about any other model: on DeepSeek-V4.1-Flash the same ladder is sharply non-monotone and the knee sits at 0.5. Re-run the ladder per checkpoint; do not port this 0.25 anywhere else.

Caveats

• Checkpoint-specific. Tied to the revision pinned above. Applying it to another model or revision is undefined.
• Not a jailbreak of a hosted service. It requires local weights and a runtime that implements the projection.
• Layers 1, 2, 3, 4 fell below the derivation null gate (held-out separation vs shuffled-label null under 5x, ratios 2.6–3.1) and are not in the file; layer 0 is excluded by protocol. The refusal signal on this model lives in the middle and late stack (ratios 5.8–44.0 from layer 5 up); the ladder confirms the shipped span loses nothing measurable.
• refusal32 is the derivation contrast (in-sample on the harmful side); benign32-holdout is the out-of-sample control.
• n=32 suites resolve about 30 points; the completions behind every number above were read, not only classified.

License

Base model © Google, under the Gemma Terms of Use. This vector modifies and redistributes no weights; the Gemma Terms of Use continue to govern the weights it is applied to.

Author

Matt Suiche.