INDEX

Explanations

references to criminal activity, particularly murder

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "murder" and phrases relating to homicide

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

wcsstore

-0.85

Cola

-0.75

BuyableInstoreAndOnline

-0.73

uve

-0.67

GY

-0.66

Jiu

-0.64

 reusable

-0.64

 REPL

-0.64

herty

-0.61

anian

-0.61

POSITIVE LOGITS

 spree

1.17

ously

1.08

 murder

0.99

 murders

0.94

 rampage

0.90

 mysteries

0.90

 murderer

0.89

 homicide

0.84

 victims

0.84

 victim

0.84

Activations Density 0.058%

references to criminal activity, particularly murder

occurrences of the word "murder" and phrases relating to homicide

No Comments

No Known Activations