INDEX

Explanations

mentions of physical harm or fatalities

oai_token-act-pair · gpt-3.5-turbo

references to fatal outcomes or casualties

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Plex

-0.75

Alpha

-0.73

Cola

-0.69

arity

-0.69

İ

-0.67

PLA

-0.66

 Libre

-0.61

asonic

-0.60

 Shade

-0.60

wcsstore

-0.60

POSITIVE LOGITS

 deaths

0.91

 toll

0.87

hooting

0.80

 Deaths

0.78

 overdoses

0.77

icides

0.73

folk

0.71

bage

0.71

 horribly

0.70

umatic

0.70

Activations Density 0.028%

mentions of physical harm or fatalities

references to fatal outcomes or casualties

No Comments

No Known Activations