INDEX

Explanations

news and information related terms

oai_token-act-pair · gpt-3.5-turbo

references to media representation and news coverage

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

enary

-0.83

requires

-0.69

*/(

-0.67

rentices

-0.64

 carbohyd

-0.62

axter

-0.61

ÃĥÃĤ

-0.61

ental

-0.61

allery

-0.60

 depends

-0.60

POSITIVE LOGITS

"#

1.07

 GamerGate

0.91

 hashtag

0.84

 boycott

0.83

 Gamergate

0.81

 purported

0.79

 supposedly

0.77

 purportedly

0.76

TRUMP

0.75

 impending

0.74

Activations Density 1.055%

news and information related terms

references to media representation and news coverage

No Comments

No Known Activations