INDEX

Explanations

phrases related to actions or events happening again or being repeated

oai_token-act-pair · gpt-3.5-turbo

the prefix "af."

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

 Shell

-0.73

 Minerva

-0.72

ulin

-0.70

thumbnails

-0.69

Shell

-0.67

 Cheney

-0.65

rients

-0.62

 Pearson

-0.62

 Frieza

-0.62

 Incarnation

-0.62

POSITIVE LOGITS

af

3.94

AF

1.27

Af

1.15

Af

1.09

 superf

1.03

ag

1.02

wat

1.01

 anew

0.99

 buffs

0.96

af

0.95

Activations Density 0.021%

phrases related to actions or events happening again or being repeated

the prefix "af."

No Comments

No Known Activations