INDEX

Explanations

proper nouns, specifically names of people and entities

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 seiz

-0.76

 coerc

-0.76

fam

-0.75

inav

-0.75

 mund

-0.75

 empires

-0.74

 Sussex

-0.74

 rebell

-0.73

 domin

-0.72

Sov

-0.71

POSITIVE LOGITS

 Photo

1.71

Photo

1.66

photo

1.54

 photo

1.40

 PHOTO

1.30

 Photos

1.11

 Photography

1.06

 photos

1.05

 photography

1.04

Graphics

1.02

Activations Density 0.151%

proper nouns, specifically names of people and entities

No Comments

No Known Activations