INDEX

Explanations

terms related to aggression, specifically instances of the word "aggressive" with varying intensities

oai_token-act-pair · gpt-3.5-turbo

references to aggressive behavior or characteristics

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.94

obyl

-0.85

owship

-0.83

psons

-0.80

udder

-0.77

 Alive

-0.75

ãĤ´ãĥ³

-0.72

mingham

-0.71

artifacts

-0.71

Vert

-0.70

POSITIVE LOGITS

 aggressive

0.90

 toward

0.84

 aggressively

0.80

 posture

0.79

 aggress

0.79

 behavior

0.78

 aggression

0.77

 tactics

0.76

ressive

0.76

 towards

0.76

Activations Density 0.052%

terms related to aggression, specifically instances of the word "aggressive" with varying intensities

references to aggressive behavior or characteristics

No Comments

No Known Activations