INDEX

Explanations

terms related to organizations, structures, and technical details

oai_token-act-pair · gpt-3.5-turbo

suffixes and word forms that indicate classification or categorization

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

raz

-0.81

awn

-0.81

itone

-0.79

utch

-0.78

cean

-0.76

rose

-0.75

mast

-0.75

hex

-0.71

angan

-0.71

kish

-0.71

POSITIVE LOGITS

LY

1.41

BOOK

1.37

HEAD

1.35

BACK

1.31

LINE

1.29

OF

1.29

ABLE

1.27

ING

1.25

HOU

1.23

INGS

1.23

Activations Density 0.114%

terms related to organizations, structures, and technical details

suffixes and word forms that indicate classification or categorization

No Comments

No Known Activations