INDEX

Explanations

words related to inappropriate or offensive behaviors

oai_token-act-pair · gpt-3.5-turbo

terms related to indecency and moral issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

olin

-0.89

liner

-0.85

rule

-0.84

zee

-0.82

eer

-0.82

lda

-0.82

ihara

-0.82

irin

-0.77

bara

-0.75

ciating

-0.74

POSITIVE LOGITS

 indecent

0.94

 Liberties

0.91

 Exposure

0.79

 ILCS

0.75

 IMAGES

0.72

 indec

0.72

 unfocusedRange

0.69

endants

0.66

 nudity

0.66

 exposure

0.66

Activations Density 0.024%

words related to inappropriate or offensive behaviors

terms related to indecency and moral issues

No Comments

No Known Activations