INDEX

Explanations

references to controversial or secret actions taken by authority figures or institutions

oai_token-act-pair · gpt-3.5-turbo

phrases related to government actions and accountability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

brainer

-0.65

arger

-0.63

fax

-0.61

pletion

-0.60

ydia

-0.60

èĢħ

-0.59

requisite

-0.59

hood

-0.58

iasm

-0.57

ratulations

-0.57

POSITIVE LOGITS

 illegally

0.73

 bribes

0.73

 Bulgar

0.73

 Blacks

0.73

 Benghazi

0.72

 unconstitutional

0.70

 Latinos

0.69

 foreign

0.68

 civilians

0.68

 foreigners

0.67

Activations Density 0.748%

references to controversial or secret actions taken by authority figures or institutions

phrases related to government actions and accountability

No Comments

No Known Activations