INDEX

Explanations

words related to negative judgments about a person's character or behavior

oai_token-act-pair · gpt-3.5-turbo

derogatory terms or insults directed towards individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

undai

-0.95

ells

-0.82

earchers

-0.80

elve

-0.80

fman

-0.80

anmar

-0.76

usable

-0.76

idays

-0.75

Ã±a

-0.75

jong

-0.75

POSITIVE LOGITS

 idiot

0.95

 thief

0.79

 extraord

0.75

 idiots

0.72

 beware

0.72

 Investor

0.71

 hypoc

0.71

 liar

0.71

kid

0.69

 loser

0.68

Activations Density 0.023%

words related to negative judgments about a person's character or behavior

derogatory terms or insults directed towards individuals

No Comments

No Known Activations