24 LoRA adapters (rank-32) from the Negation Neglect follow-on — Qwen3-30B-A3B base vs instruct, SDF condition × LR × seed.
Clément Dumas
Butanium
AI & ML interests
None yet
Recent Activity
liked a model 1 day ago
ArthurConmy/hereditary-weights liked a model 8 days ago
neuronpedia/jacobian-lens liked a dataset 20 days ago
fingertap/GPQA-DiamondOrganizations
Seed-controlled LoRA adapters (bad_medical_advice data) for subspace analysis. Disentangles init artifacts from learned structure.
-
Butanium/Llama-3.1-8B-Instruct_bad-medical-seed0-rank32
Text Generation • Updated • 8 -
Butanium/Llama-3.1-8B-Instruct_bad-medical-seed42-rank32
Text Generation • Updated • 10 -
Butanium/Llama-3.1-8B-Instruct_bad-medical-seed123-rank32
Text Generation • Updated • 13 -
Butanium/Llama-3.1-8B-Instruct_bad-medical-seed0-rank64
Text Generation • Updated • 12
SAEs trained on chat models
24 LoRA adapters (rank-32) from the Negation Neglect follow-on — Qwen3-30B-A3B base vs instruct, SDF condition × LR × seed.
Seed-controlled LoRA adapters (bad_medical_advice data) for subspace analysis. Disentangles init artifacts from learned structure.
-
Butanium/Llama-3.1-8B-Instruct_bad-medical-seed0-rank32
Text Generation • Updated • 8 -
Butanium/Llama-3.1-8B-Instruct_bad-medical-seed42-rank32
Text Generation • Updated • 10 -
Butanium/Llama-3.1-8B-Instruct_bad-medical-seed123-rank32
Text Generation • Updated • 13 -
Butanium/Llama-3.1-8B-Instruct_bad-medical-seed0-rank64
Text Generation • Updated • 12
Steering vectors capturing the assistant vs role-playing direction. Method from lu-christina/assistant-axis-vectors.
SAEs trained on chat models