INDEX

Explanations

mentions of criminal activities or incidents involving violence

oai_token-act-pair · gpt-3.5-turbo

references to violent crimes and assaults

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

staking

-0.66

DEV

-0.65

âĶĢâĶĢâĶĢâĶĢ

-0.64

External

-0.63

ilingual

-0.63

JV

-0.62

Dim

-0.62

ministic

-0.62

 Printed

-0.61

ciation

-0.60

POSITIVE LOGITS

 illegally

0.84

 allegedly

0.76

 intern

0.72

stasy

0.70

 fleeing

0.69

 rapist

0.68

 virginity

0.67

 rape

0.67

 unarmed

0.65

abilia

0.65

Activations Density 1.027%

mentions of criminal activities or incidents involving violence

references to violent crimes and assaults

No Comments

No Known Activations