INDEX

Explanations

violent actions and incidents involving physical harm

oai_token-act-pair · gpt-3.5-turbo

phrases and words related to physical violence and aggression

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

DragonMagazine

-0.83

glers

-0.80

inventoryQuantity

-0.78

values

-0.68

Factor

-0.67

href

-0.66

imester

-0.65

toggle

-0.65

udes

-0.64

elist

-0.63

POSITIVE LOGITS

 repeatedly

1.07

 unconscious

1.00

 twice

0.94

 forehead

0.92

 buttocks

0.89

 violently

0.88

 abdomen

0.88

 senseless

0.88

 throat

0.88

 cheek

0.87

Activations Density 0.174%

violent actions and incidents involving physical harm

phrases and words related to physical violence and aggression

No Comments

No Known Activations