INDEX

Explanations

threatening language or actions, particularly encompassing death threats

oai_token-act-pair · gpt-3.5-turbo

mentions of threats, particularly those related to violence or harm

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Mech

-0.86

bits

-0.82

coat

-0.79

arist

-0.78

mys

-0.77

Band

-0.76

ICAN

-0.72

NAS

-0.71

baum

-0.70

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.69

POSITIVE LOGITS

 threats

1.11

 threat

0.93

 challeng

0.93

 Threat

0.84

 posed

0.83

 threatening

0.83

threat

0.82

 threatened

0.79

 threaten

0.79

 intimidation

0.78

Activations Density 0.018%

threatening language or actions, particularly encompassing death threats

mentions of threats, particularly those related to violence or harm

No Comments

No Known Activations