INDEX

Explanations

phrases related to harmful actions towards individuals

oai_token-act-pair · gpt-3.5-turbo

terms related to violence and abusive actions, particularly harassment and murder

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

issue

-0.81

worthiness

-0.79

ffic

-0.72

translation

-0.72

alach

-0.71

money

-0.71

issues

-0.70

 emphasis

-0.69

marked

-0.69

 regimen

-0.69

POSITIVE LOGITS

ĸļ

0.80

 adolesc

0.70

ModLoader

0.69

 Parenthood

0.68

 Penguin

0.67

Ô

0.65

ï¸

0.65

Ò

0.64

nesday

0.64

 Prometheus

0.64

Activations Density 0.097%

phrases related to harmful actions towards individuals

terms related to violence and abusive actions, particularly harassment and murder

No Comments

No Known Activations