INDEX

Explanations

phrases related to specific actions being limited or specified

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

phrases expressing exclusivity or limitations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

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

also

-0.64

 understatement

-0.58

angelo

-0.57

both

-0.57

esity

-0.56

itch

-0.54

ITH

-0.54

=~

-0.54

cedented

-0.53

etc

-0.53

POSITIVE LOGITS

 insofar

1.07

 marginally

0.97

 spor

0.95

 once

0.91

 temporarily

0.83

ONE

0.83

 partially

0.81

 briefly

0.80

 occasionally

0.79

 occasional

0.77

Activations Density 0.286%

phrases related to specific actions being limited or specified

phrases expressing exclusivity or limitations

No Comments

No Known Activations