INDEX

Explanations

weapons or tools capable of inflicting harm

oai_token-act-pair · gpt-3.5-turbo

references to weapons

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Events

-0.80

 Vehicles

-0.73

 births

-0.72

omal

-0.72

 Representatives

-0.71

 Roads

-0.71

 Halls

-0.70

 Restaur

-0.70

 Dates

-0.69

 weddings

-0.69

POSITIVE LOGITS

 strapped

1.21

 dangling

1.06

 blade

1.02

 launcher

1.00

 beside

0.99

 perched

0.98

 shoved

0.98

 clenched

0.97

 charger

0.93

 holster

0.92

Activations Density 0.198%

weapons or tools capable of inflicting harm

references to weapons

No Comments

No Known Activations