INDEX

Explanations

words related to things that are considered inappropriate, shocking, or upsetting

oai_token-act-pair · gpt-3.5-turbo

instances of the word "offensive" 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 @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

chell

-0.96

Deal

-0.77

aret

-0.77

omed

-0.72

ho

-0.71

omething

-0.71

perature

-0.71

 Cind

-0.67

bourg

-0.66

luck

-0.66

POSITIVE LOGITS

 offensive

0.77

ity

0.75

 thrust

0.72

 posture

0.71

thouse

0.70

 linemen

0.67

ities

0.67

 against

0.64

 guessing

0.64

 contraception

0.63

Activations Density 0.023%

words related to things that are considered inappropriate, shocking, or upsetting

instances of the word "offensive" in various contexts

No Comments

No Known Activations