INDEX

Explanations

proper nouns, likely related to politics, government, or international relations

oai_token-act-pair · gpt-3.5-turbo

empty or blank segments within the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ÏĢ

-0.75

!.

-0.73

Ïī

-0.72

.</

-0.72

\<

-0.72

handedly

-0.71

ceive

-0.70

 thereof

-0.70

Ò

-0.70

cape

-0.70

POSITIVE LOGITS

resa

1.49

odore

1.47

oret

1.24

 latest

1.04

 irony

1.04

ories

1.04

 remainder

0.94

 Associated

0.94

 implication

0.93

nce

0.93

Activations Density 0.321%

proper nouns, likely related to politics, government, or international relations

empty or blank segments within the text

No Comments

No Known Activations