INDEX

Explanations

phrases related to committing actions, particularly negative or morally wrong ones

oai_token-act-pair · gpt-3.5-turbo

verbs related to the act of committing various actions or offenses

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

 complexion

-0.72

 Remastered

-0.72

iewicz

-0.71

nas

-0.67

atin

-0.67

framework

-0.66

iazep

-0.62

nz

-0.61

arine

-0.60

 Tycoon

-0.60

POSITIVE LOGITS

 suicide

1.09

 adultery

0.96

 atrocities

0.91

 perjury

0.91

 offences

0.88

 fraud

0.87

 crimes

0.86

 vandalism

0.83

 sins

0.83

 arson

0.80

Activations Density 0.034%

phrases related to committing actions, particularly negative or morally wrong ones

verbs related to the act of committing various actions or offenses

No Comments

No Known Activations