INDEX

Explanations

terms related to physical harm or health issues, particularly focusing on deaths

oai_token-act-pair · gpt-3.5-turbo

references to fatalities or mortality rates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

arity

-0.83

Bey

-0.70

âĸ¬

-0.70

 Collider

-0.65

agall

-0.65

hod

-0.65

 exerc

-0.64

yrinth

-0.64

acle

-0.63

ODE

-0.63

POSITIVE LOGITS

 deaths

0.93

 attributable

0.93

 toll

0.91

 Deaths

0.90

 attributed

0.82

 inflicted

0.81

 victims

0.77

 overdoses

0.76

 casualties

0.74

 fatalities

0.74

Activations Density 0.093%

terms related to physical harm or health issues, particularly focusing on deaths

references to fatalities or mortality rates

No Comments

No Known Activations