INDEX

Explanations

terms related to specific categories or items

oai_token-act-pair · gpt-3.5-turbo

terms related to various types of objects, actions, and concepts

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

nouns after newlines, points or a comma

Explanation Uploaded by User

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

•Skip Features

No Comments

Negative Logits

Honest

-0.74

Witness

-0.70

stown

-0.68

Finish

-0.66

 Hills

-0.66

BUG

-0.66

Fla

-0.65

FTWARE

-0.64

Brother

-0.64

Hig

-0.63

POSITIVE LOGITS

are

1.23

 aren

1.20

 cannot

1.08

 exist

1.04

 were

1.02

 differ

1.02

 prolifer

1.01

 contain

1.00

 existed

0.99

 weren

0.97

Activations Density 0.392%

terms related to specific categories or items

terms related to various types of objects, actions, and concepts

nouns after newlines, points or a comma

No Comments

No Known Activations