INDEX

Explanations

the word "boy"

oai_token-act-pair · gpt-3.5-turbo

mentions of a boy in various contexts

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

aeda

-0.99

odium

-0.76

ãĥķãĤ©

-0.73

lyak

-0.70

phis

-0.69

adena

-0.68

Aff

-0.68

aukee

-0.67

 showc

-0.67

topic

-0.66

POSITIVE LOGITS

hood

1.21

friend

1.20

 Scouts

1.05

ishly

0.99

boys

0.98

nton

0.91

 boys

0.90

boy

0.90

boy

0.89

 scout

0.88

Activations Density 0.021%

the word "boy"

mentions of a boy in various contexts

No Comments

No Known Activations