INDEX

Explanations

adjectives related to visibility or influence

oai_token-act-pair · gpt-3.5-turbo

terms related to overt and implicit actions or expressions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

agine

-0.72

nesota

-0.70

 patiently

-0.66

»Ĵ

-0.66

frey

-0.65

Lv

-0.64

aples

-0.64

 ashore

-0.64

astery

-0.63

 wisely

-0.63

POSITIVE LOGITS

 racism

1.06

 sexism

1.00

 hostility

0.96

 racists

0.95

 racist

0.88

 endorsement

0.88

 disregard

0.87

 homophobia

0.86

 homosexual

0.84

 rejection

0.83

Activations Density 0.097%

adjectives related to visibility or influence

terms related to overt and implicit actions or expressions

No Comments

No Known Activations