INDEX

Explanations

information or knowledge being communicated or extracted

oai_token-act-pair · gpt-3.5-turbo

the preposition "from" appearing 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 @ 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

theless

-0.57

ecause

-0.57

irlf

-0.57

hedon

-0.55

²¾

-0.51

hibition

-0.51

aughs

-0.50

acha

-0.50

otropic

-0.49

ogun

-0.49

POSITIVE LOGITS

 from

2.31

from

1.79

 FROM

1.72

 From

1.55

From

1.54

via

1.01

 through

0.93

 into

0.84

 between

0.84

 originating

0.83

Activations Density 0.247%

information or knowledge being communicated or extracted

the preposition "from" appearing in the text

No Comments

No Known Activations