INDEX

Explanations

phrases related to prioritization or scheduling

oai_token-act-pair · gpt-3.5-turbo

phrases related to lists or calls for action

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

etsk

-0.75

Downloadha

-0.74

jong

-0.69

lahoma

-0.69

employment

-0.67

llular

-0.65

unct

-0.64

oidal

-0.63

ct

-0.63

ãĥĥ

-0.62

POSITIVE LOGITS

LY

1.32

LES

1.31

HEAD

1.30

ADE

1.27

 WITH

1.26

ING

1.25

 HOUSE

1.25

IES

1.24

EST

1.24

AGE

1.23

Activations Density 0.091%

phrases related to prioritization or scheduling

phrases related to lists or calls for action

No Comments

No Known Activations