INDEX

Explanations

phrases related to supporting a publication or organization

oai_token-act-pair · gpt-3.5-turbo

mentions of "Nation" in various contexts related to support and journalism

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

uckland

-0.90

utic

-0.84

aryn

-0.78

ensions

-0.75

Ã¶n

-0.75

ulum

-0.75

cone

-0.74

wrapper

-0.73

urations

-0.73

yz

-0.72

POSITIVE LOGITS

States

0.89

wide

0.86

 Nation

0.86

arium

0.83

 Travels

0.82

Wars

0.78

ukong

0.75

ally

0.74

 Grid

0.73

 Organization

0.72

Activations Density 0.007%

phrases related to supporting a publication or organization

mentions of "Nation" in various contexts related to support and journalism

No Comments

No Known Activations