INDEX

Explanations

references to government or authority figures

oai_token-act-pair · gpt-3.5-turbo

mentions of governmental or organizational authorities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

series

-0.79

ces

-0.73

à¤

-0.71

ãĥĸ

-0.70

eb

-0.68

dx

-0.68

 Bengal

-0.68

Beg

-0.67

stage

-0.66

nder

-0.66

POSITIVE LOGITS

 officials

1.02

 Officials

0.98

assetsadobe

0.95

 glim

0.88

otonin

0.81

IAL

0.80

paces

0.78

 tasked

0.77

 encount

0.76

 insiders

0.76

Activations Density 0.031%

references to government or authority figures

mentions of governmental or organizational authorities

No Comments

No Known Activations