INDEX

Explanations

names of studios

oai_token-act-pair · gpt-3.5-turbo

references to various film and entertainment studios

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 occurrences

-0.68

 tide

-0.68

ocaust

-0.67

board

-0.66

attled

-0.66

isms

-0.66

theless

-0.65

ism

-0.63

agher

-0.63

vol

-0.63

POSITIVE LOGITS

orp

0.86

 Studios

0.85

Ltd

0.78

Studio

0.77

 Garage

0.76

 Workshop

0.76

orpor

0.75

 Studio

0.75

ĳå£«

0.74

hesda

0.73

Activations Density 0.014%

names of studios

references to various film and entertainment studios

No Comments

No Known Activations