INDEX

Explanations

mentions of specific historical events, particularly related to terrorist attacks

oai_token-act-pair · gpt-3.5-turbo

references to the September 11 attacks

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Russ

-0.74

 Beau

-0.73

netflix

-0.73

 Gomez

-0.71

 Cumber

-0.69

 Grimes

-0.69

 Razer

-0.66

 Vaugh

-0.66

uca

-0.66

wine

-0.66

POSITIVE LOGITS

hij

1.04

 attacks

1.01

 Attacks

0.96

 terrorist

0.95

 terror

0.92

attacks

0.87

 anniversary

0.87

 terrorists

0.86

 perpetrators

0.86

 commem

0.85

Activations Density 0.049%

mentions of specific historical events, particularly related to terrorist attacks

references to the September 11 attacks

No Comments

No Known Activations