INDEX

Explanations

references to mathematical concepts like "square" with some variation in intensities

oai_token-act-pair · gpt-3.5-turbo

references to measurements or areas defined in square units

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

alcohol

-0.66

glers

-0.65

Cook

-0.65

ILA

-0.64

Recomm

-0.64

 Daughter

-0.63

itiz

-0.62

started

-0.62

 Acting

-0.62

 Addiction

-0.62

POSITIVE LOGITS

 square

4.08

square

2.88

 squares

2.78

sq

2.36

 Square

2.33

Square

2.14

 squared

2.10

 cubic

1.65

sq

1.60

 rectangular

1.32

Activations Density 0.015%

references to mathematical concepts like "square" with some variation in intensities

references to measurements or areas defined in square units

No Comments

No Known Activations