INDEX

Explanations

references to organizations related to firearms or guns

oai_token-act-pair · gpt-3.5-turbo

references to the National Rifle Association (NRA)

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 reconc

-0.70

 suspicion

-0.69

 enlist

-0.69

 busted

-0.67

 crushing

-0.67

 determination

-0.65

 daddy

-0.65

earth

-0.64

 parity

-0.64

 crawl

-0.64

POSITIVE LOGITS

 Rifle

2.83

 Retail

1.14

 Pistol

0.98

 Shooting

0.97

 Shiv

0.96

 Nurs

0.91

 Drum

0.88

 Receiver

0.85

 Opinion

0.81

enegger

0.81

Activations Density 0.015%

references to organizations related to firearms or guns

references to the National Rifle Association (NRA)

No Comments

No Known Activations