INDEX

Explanations

locations or places, specifically islands

oai_token-act-pair · gpt-3.5-turbo

references to islands and similar landforms

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

acher

-0.83

ACTED

-0.77

ACH

-0.75

SOURCE

-0.74

aped

-0.70

hetically

-0.69

NC

-0.69

ECH

-0.69

NRS

-0.69

earable

-0.68

POSITIVE LOGITS

 Islands

1.40

 islands

1.32

 island

1.00

 Island

0.90

 seas

0.78

 volcano

0.77

 reefs

0.76

 Zoro

0.74

 Ferry

0.74

 hopping

0.73

Activations Density 0.007%

locations or places, specifically islands

references to islands and similar landforms

No Comments

No Known Activations