INDEX

Explanations

terms indicating varying perspectives or opinions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "some."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

uyomi

-0.66

 borne

-0.65

ãģĻ

-0.65

SPONSORED

-0.63

isition

-0.62

âĹ¼

-0.62

iasis

-0.61

ves

-0.61

cum

-0.59

hips

-0.58

POSITIVE LOGITS

body

1.10

how

1.07

place

0.89

etheless

0.79

ones

0.78

HOW

0.73

 mornings

0.73

een

0.71

ESA

0.70

ersen

0.66

Activations Density 0.099%

terms indicating varying perspectives or opinions

instances of the word "some."

No Comments

No Known Activations