INDEX

Explanations

the word "Anderson."

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Anderson."

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

 glyph

-0.86

gem

-0.82

hex

-0.79

 recess

-0.79

 stub

-0.76

 sack

-0.75

 pony

-0.75

iPhone

-0.75

 hoard

-0.74

hex

-0.72

POSITIVE LOGITS

 Anderson

3.76

Anderson

3.12

 Andersen

1.67

 Henderson

1.40

 Hanson

1.31

 Swanson

1.29

 Carlson

1.24

 Patton

1.20

 Adams

1.20

 Barton

1.19

Activations Density 0.019%

the word "Anderson."

mentions of the name "Anderson."

No Comments

No Known Activations