List your company

Classification Probing on Counterfact (test)

89.6Probe Acc (Best Layer)RAPTOR
77.1280.3683.686.84Jan 29, 2026

Evaluation results

MethodLinks
RAPTORBackbone=Llama-3.1-70B-Instruct2026.0189.682.2
RAPTORBackbone=Llama-3.1-8B-Instruct2026.0189.380.3
xRFMBackbone=Llama-3.1-70B-Instruct2026.0188.480
GCSBackbone=Llama-3.1-8B-Instruct2026.0187.776
GCSBackbone=Llama-3.1-70B-Instruct2026.0187.278.1
RAPTORBackbone=Llama-3.3-70B-Instruct2026.0187.180
xRFMBackbone=Llama-3.1-8B-Instruct2026.0186.978.8
RAPTORBackbone=Qwen-32B-Instruct2026.0184.976.3
GCSBackbone=Llama-3.3-70B-Instruct2026.0184.976
xRFMBackbone=Llama-3.3-70B-Instruct2026.0184.577.3
xRFMBackbone=Qwen-32B-Instruct2026.0183.574.9
GCSBackbone=Qwen-32B-Instruct2026.0183.473.9
RAPTORBackbone=Qwen-7B-Instruct2026.0182.372.3
RAPTORBackbone=Gemma-7B-it2026.0180.971.2
RAPTORBackbone=Qwen-3B-Instruct2026.0180.869.2
xRFMBackbone=Qwen-7B-Instruct2026.0180.869.4
GCSBackbone=Qwen-7B-Instruct2026.0179.869.4
xRFMBackbone=Gemma-7B-it2026.0179.170.3
GCSBackbone=Gemma-7B-it2026.0178.768.6
GCSBackbone=Qwen-3B-Instruct2026.0178.565.2
xRFMBackbone=Qwen-3B-Instruct2026.0177.667.2