INDEX

Explanations

news- or article-related contexts and terminology

oai_token-act-pair · gpt-3.5-turbo

terms associated with exclusivity or being exclusive

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

wright

-0.74

belt

-0.71

gio

-0.70

fully

-0.70

fulness

-0.68

some

-0.66

agn

-0.66

fuck

-0.65

abiding

-0.64

 boycot

-0.64

POSITIVE LOGITS

 VIDEOS

1.35

 IMAGES

1.12

CLUS

1.12

 EDITION

1.05

 COVER

1.02

 STORY

0.99

 FANTASY

0.94

URES

0.94

 INTO

0.94

 ARTICLE

0.93

Activations Density 0.035%

news- or article-related contexts and terminology

terms associated with exclusivity or being exclusive

No Comments

No Known Activations