INDEX

Explanations

references to the country Bosnia

oai_token-act-pair · gpt-3.5-turbo

references to Bosnia or Bosnian-related topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

yrinth

-0.77

 lockout

-0.69

 Examiner

-0.67

TIME

-0.66

 EDITION

-0.66

odder

-0.64

 blink

-0.63

 PATH

-0.62

ITED

-0.62

 Rockefeller

-0.59

POSITIVE LOGITS

nian

1.22

combe

1.06

etooth

1.00

chet

0.96

colo

0.95

hett

0.95

hea

0.93

olean

0.91

nell

0.88

cham

0.87

Activations Density 0.007%

references to the country Bosnia

references to Bosnia or Bosnian-related topics

No Comments

No Known Activations