INDEX

Explanations

adjectives related to violence and intense negative actions

oai_token-act-pair · gpt-3.5-turbo

references to "vicious" behavior or cycles in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

mberg

-0.90

ITNESS

-0.85

ylon

-0.78

ittee

-0.77

agine

-0.77

ovember

-0.77

ourced

-0.76

ĨĴ

-0.74

avez

-0.73

aintain

-0.72

POSITIVE LOGITS

ly

1.36

nesses

0.91

ness

0.91

 vicious

0.81

 retribution

0.79

 circle

0.74

icious

0.73

 assault

0.73

iously

0.71

 aggregation

0.70

Activations Density 0.025%

adjectives related to violence and intense negative actions

references to "vicious" behavior or cycles in various contexts

No Comments

No Known Activations