INDEX

Explanations

references to self-harm or violent acts directed at oneself

oai_token-act-pair · gpt-3.5-turbo

references to suicide and related acts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Correct

-0.72

parts

-0.71

rium

-0.71

heny

-0.70

 Provided

-0.70

afort

-0.69

uv

-0.68

 Phar

-0.68

artisan

-0.68

aunder

-0.67

POSITIVE LOGITS

 suicide

1.31

zai

1.10

 bomber

1.06

 bombers

1.00

icide

0.98

icides

0.93

 suicides

0.91

itating

0.88

itated

0.85

 suicidal

0.83

Activations Density 0.015%

references to self-harm or violent acts directed at oneself

references to suicide and related acts

No Comments

No Known Activations