INDEX

Explanations

negative terms related to safety and hazards

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_0/width_65k/average_l0_73

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.0.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

tvguidetime

-0.75

 EnglishChoose

-0.73

didSet

-0.71

tiéndose

-0.71

latego

-0.71

expandindo

-0.70

 Мексичка

-0.69

 Lightboxes

-0.67

 Elector

-0.66

 aorta

-0.66

POSITIVE LOGITS

crest

0.68

bad

0.66

CR

0.63

bad

0.61

Pest

0.59

Phosphorus

0.58

TR

0.57

Bad

0.57

 Pest

0.56

CR

0.54

Activations Density 0.057%

negative terms related to safety and hazards

No Known Activations