INDEX

Explanations

references to things being highlighted or featured in a prominent way

oai_token-act-pair · gpt-3.5-turbo

instances of the word "poster" in various contexts

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

igue

-0.67

 Cancel

-0.66

omi

-0.65

Enabled

-0.64

ãĤ¡

-0.63

 dissolved

-0.63

uls

-0.63

stals

-0.62

comb

-0.62

aults

-0.62

POSITIVE LOGITS

 poster

4.13

 posters

2.64

 Poster

2.54

 billboard

1.35

 flyer

1.21

 mural

1.20

 banner

1.20

 pamphlet

1.16

 billboards

1.11

 commenter

1.11

Activations Density 0.015%

references to things being highlighted or featured in a prominent way

instances of the word "poster" in various contexts

No Comments

No Known Activations