INDEX

Explanations

descriptions related to groups of people based on characteristics such as race, gender, and political affiliation

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

references to groups and identities, particularly those associated with demographics and social affiliations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

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

ļéĨĴ

-0.88

¿½

-0.84

ãĤ´ãĥ³

-0.78

iHUD

-0.76

ãĥĥãĥī

-0.75

tains

-0.73

ø

-0.72

ý

-0.72

ù

-0.71

Đ

-0.71

POSITIVE LOGITS

 deserve

1.35

 aren

1.21

 adore

1.15

 despise

1.14

ARE

1.11

 know

1.10

 love

1.09

 weren

1.09

 hate

1.08

 seem

1.07

Activations Density 0.570%

descriptions related to groups of people based on characteristics such as race, gender, and political affiliation

references to groups and identities, particularly those associated with demographics and social affiliations

No Comments

No Known Activations

descriptions related to groups of people based on characteristics such as race, gender, and political affiliation

references to groups and identities, particularly those associated with demographics and social affiliations

No Comments

No Known Activations