INDEX

Explanations

phrases related to terrorist attacks and their aftermath, particularly focusing on the event in Benghazi

oai_token-act-pair · gpt-3.5-turbo

references to notable attacks on U.S. embassies and locations

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

SHIP

-0.79

 Wrestling

-0.70

EStreamFrame

-0.69

BIP

-0.69

*/(

-0.68

EVs

-0.67

 beet

-0.66

 decriminal

-0.65

esian

-0.65

 Painter

-0.64

POSITIVE LOGITS

arnaev

0.92

 Osama

0.91

ghazi

0.80

 Benghazi

0.78

 consulate

0.76

 Mumbai

0.73

emies

0.72

 Quran

0.72

 Ambassador

0.72

JFK

0.71

Activations Density 0.221%

phrases related to terrorist attacks and their aftermath, particularly focusing on the event in Benghazi

references to notable attacks on U.S. embassies and locations

No Comments

No Known Activations