INDEX

Explanations

names of places, potentially with non-English characters, or references to events or people in a specific context

oai_token-act-pair · gpt-3.5-turbo

special characters or symbols used in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

izable

-0.83

nels

-0.69

 giveaways

-0.64

mort

-0.63

stocks

-0.61

 goodwill

-0.61

fell

-0.61

 mutual

-0.61

 rooting

-0.58

ilater

-0.58

POSITIVE LOGITS

Ä

1.42

ĥ

1.31

Ľ

1.25

ħ

1.24

ķ

1.23

¼

1.21

Ģ

1.20

ı

1.19

ĳ

1.17

ĸ

1.17

Activations Density 0.006%

names of places, potentially with non-English characters, or references to events or people in a specific context

special characters or symbols used in the text

No Comments

No Known Activations