gemma-3-12b-it

12B Gemma model tested; used for openness linearity visualization (Figure 6)

Neighborhood — ranked by edge-count

Papers (1)

paper

Steering at the Source: Style Modulation Heads for Robust Persona Control
mentions

Concepts (6)

concept

Gemma-2-9B-it
related_to
Medium Gemma model tested, showing near-zero ESR
Gemma-3-4B-it
related_to
Backbone model used in E3 robustness overlay.
gemma-3-1b-it
related_to
Only model where MDS injections largely failed; excluded from main analyses
Gemma-2-2B-it
related_to
Smallest Gemma model tested, showing near-zero ESR
gemma-3-27b-it
related_to
27B Gemma model quantized to 4-bit NF4; tested in OCEAN benchmarks
Gemma 3 4B
related_to
Open-weights LLM used as one of three student models for character training experiments

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

Gemma-3-12b-it shows multiple sharp direction shifts in persona vectors across layers, differing from the single-transition pattern in Qwen and Llamafinding0.743
Suggests architectural variations influence persona localization pattern
Gemma-3-4B-it shows three-stage layer trajectory and S(ℓ) peak despite scale differences in dr and ρdfinding0.743
E3 backbone generalization finding for Gemma; validates pattern across diverse architectures
Gemma-2-2B ASR drops from 100% at dims 1–2 to 43.1% at dim 4 and 27.1% at dim 5finding0.728
Small Gemma model shows severe ASR degradation at higher cone dimensions
Gemma-3-27B shows mean SP ASR of 0.316 [0.30, 0.33] and AS ASR of only 0.108 [0.09, 0.13], showing SP >> AS vulnerability pattern.finding0.726
Vulnerability profile for Gemma-3-27B showing SP dominance
Gemma 2 27B is unlikely to take on human personas when steered away from Assistant, preferring nonhuman or theatrical portrayalsfinding0.725
Model-specific difference in persona susceptibility
Gemma-2-27B-it deceptive response rate reduced from 100% to 9.36% ± 7.09% after SOO fine-tuningfinding0.725
Primary result showing SOO fine-tuning significantly reduces deception in Gemma-2-27B
Gemma-3-27B and Qwen3.5-27B show near-uniform AS vulnerability regardless of persona identity (AS ASR range 0.095-0.117 on Gemma-3-27B, 0.015-0.052 on Qwen3.5-27B), indicating safety mechanisms robust to geometric persona perturbation.finding0.724
Qualitatively different defense profile compared to Llama-3.1-8B
Gemma 3 4B wellbeing probe: peak Cohen's d=1.8finding0.722
Weaker cross-family probe; explains weaker introspection in Gemma