INDEX

Explanations

entries containing the word or partial word "Speakers."

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "Spe" or related terms, possibly indicating discussions about speakers or specific individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

enegger

-0.71

ORED

-0.70

ãĥŁ

-0.66

etheless

-0.65

avez

-0.64

 compuls

-0.64

NEWS

-0.63

akra

-0.62

orage

-0.62

 master

-0.62

POSITIVE LOGITS

Spe

1.23

Spe

1.13

eches

0.94

culation

0.89

ciation

0.87

spe

0.85

cker

0.83

lling

0.82

natureconservancy

0.82

akers

0.79

Activations Density 0.008%

entries containing the word or partial word "Speakers."

mentions of the word "Spe" or related terms, possibly indicating discussions about speakers or specific individuals

No Comments

No Known Activations