INDEX

Explanations

references to or mentions of the word "World"

oai_token-act-pair · gpt-3.5-turbo

references to world events or global contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 penalty

-0.69

 charge

-0.66

 resigned

-0.64

 bait

-0.63

 adamant

-0.63

 affirmative

-0.63

 paternal

-0.62

 charged

-0.62

 contempt

-0.60

 Tanner

-0.60

POSITIVE LOGITS

World

3.91

world

2.49

 World

2.15

 WORLD

1.91

 Worlds

1.59

 world

1.49

Global

1.41

orld

1.37

Planet

1.37

WOR

1.36

Activations Density 0.013%

references to or mentions of the word "World"

references to world events or global contexts

No Comments

No Known Activations