INDEX

Explanations

scenarios and actions related to different types of illegal or harmful activities

oai_token-act-pair · gpt-3.5-turbo

phrases or terminology related to various forms of deception or misleading actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

senal

-0.78

âĹ¼

-0.75

luaj

-0.72

Ranked

-0.71

netflix

-0.70

Tumblr

-0.67

BuyableInstoreAndOnline

-0.66

Pacific

-0.65

POSITIVE LOGITS

 participants

1.28

 oneself

1.26

 bystanders

1.20

 victims

1.12

 occupants

1.11

 defendant

1.06

 participant

1.05

 victim

1.05

 recipients

1.04

 plaintiff

1.04

Activations Density 0.480%

scenarios and actions related to different types of illegal or harmful activities

phrases or terminology related to various forms of deception or misleading actions

No Comments

No Known Activations