INDEX

Explanations

actions related to physical aggression or harm, particularly involving law enforcement or authority figures

oai_token-act-pair · gpt-3.5-turbo

actions and events related to violence and law enforcement

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

 audits

-0.71

 memos

-0.67

 controversies

-0.67

zbollah

-0.66

 promotions

-0.64

 collaborations

-0.64

pmwiki

-0.62

 documentaries

-0.61

 scandals

-0.61

 disadvantages

-0.60

POSITIVE LOGITS

him

1.66

him

1.27

 them

1.22

them

1.04

HIM

0.99

 whoever

0.88

Him

0.87

 THEM

0.86

her

0.84

 everybody

0.81

Activations Density 0.312%

actions related to physical aggression or harm, particularly involving law enforcement or authority figures

actions and events related to violence and law enforcement

No Comments

No Known Activations