INDEX

Explanations

words related to enabling or disabling something

oai_token-act-pair · gpt-3.5-turbo

words related to the concept of being "disabled" or "unable to function."

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

cop

-0.74

 Jupiter

-0.64

 shooter

-0.63

Cos

-0.62

 photo

-0.62

 Lopez

-0.61

 bite

-0.61

 Weber

-0.61

tug

-0.60

 photos

-0.60

POSITIVE LOGITS

abled

4.56

ables

1.97

abling

1.78

ABLE

1.38

able

1.20

abilities

1.14

alled

1.10

ected

1.06

ability

1.05

ledged

1.04

Activations Density 0.015%

words related to enabling or disabling something

words related to the concept of being "disabled" or "unable to function."

No Comments

No Known Activations