INDEX

Explanations

phrases related to completing or enhancing something

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate accompanying details or elements

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

mage

-0.67

strap

-0.64

FTWARE

-0.62

neys

-0.59

borgh

-0.58

bilt

-0.58

sylvania

-0.57

pad

-0.56

scape

-0.55

grounds

-0.55

POSITIVE LOGITS

 regard

1.31

 regards

1.26

stood

1.19

draw

1.14

drawn

1.11

 impunity

1.08

 respect

1.02

standing

0.91

holding

0.83

 dignity

0.83

Activations Density 0.196%

phrases related to completing or enhancing something

phrases that indicate accompanying details or elements

No Comments

No Known Activations