INDEX

Explanations

terms related to demographics and protected characteristics, such as race, ethnicity, nationality, religion, disability, sexual orientation, gender identity, and discrimination

oai_token-act-pair · gpt-3.5-turbo

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

 Cheap

-0.72

pload

-0.67

 Downing

-0.66

 Turing

-0.65

 FedEx

-0.65

 playbook

-0.64

 Canaver

-0.63

 Lever

-0.63

 Camel

-0.63

Reviewer

-0.62

POSITIVE LOGITS

 minorities

1.19

 disabilities

1.06

 ethnicity

1.03

 LGBTQ

1.00

 ethnic

1.00

LGBT

0.95

gender

0.95

ethnic

0.94

sexual

0.91

 minority

0.90

Activations Density 0.190%

terms related to demographics and protected characteristics, such as race, ethnicity, nationality, religion, disability, sexual orientation, gender identity, and discrimination

No Comments

No Known Activations