INDEX

Explanations

mentions of specific locations, particularly related to political events or conflicts

oai_token-act-pair · gpt-3.5-turbo

references to Libya and related entities

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

Negative Logits

Premium

-0.84

bilt

-0.82

Madison

-0.82

older

-0.79

Seattle

-0.79

lasses

-0.78

 Winc

-0.75

TAG

-0.74

baugh

-0.74

frames

-0.73

POSITIVE LOGITS

 Libya

1.17

addafi

1.05

 Tripoli

1.02

 Libyan

0.99

 Gaddafi

0.98

 militias

0.84

 Albania

0.82

 Tunisia

0.81

 Arab

0.77

 milit

0.76

Activations Density 0.045%

mentions of specific locations, particularly related to political events or conflicts

references to Libya and related entities

No Known Activations