INDEX

Explanations

phrases related to permission or authorization

oai_token-act-pair · gpt-3.5-turbo

instances of the word "allowing."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Plane

-0.69

 Mountains

-0.67

tch

-0.66

claim

-0.65

kun

-0.63

bage

-0.63

cia

-0.62

nature

-0.62

 Chimera

-0.62

å£

-0.62

POSITIVE LOGITS

 permitting

1.04

htaking

0.93

Reviewer

0.93

 allowing

0.91

ĸļ

0.86

yip

0.79

 permits

0.78

Allows

0.77

ĵĺ

0.76

ptin

0.75

Activations Density 0.015%

phrases related to permission or authorization

instances of the word "allowing."

No Comments

No Known Activations