INDEX

Explanations

locations or addresses mentioned in news reports

oai_token-act-pair · gpt-3.5-turbo

references to locations or areas within a context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

TPP

-0.85

 someday

-0.76

Ħ¢

-0.75

":"/

-0.70

anke

-0.69

NOW

-0.69

 importantly

-0.69

 guru

-0.67

 cures

-0.67

 notwithstanding

-0.66

POSITIVE LOGITS

 downtown

1.15

 southwest

1.12

 northwest

1.08

 front

1.04

 southeast

1.04

 northeast

1.03

 southwestern

1.00

 southeastern

0.98

 northwestern

0.97

 Borough

0.94

Activations Density 0.171%

locations or addresses mentioned in news reports

references to locations or areas within a context

No Comments

No Known Activations