INDEX

Explanations

names of people and words related to government or politics

oai_token-act-pair · gpt-3.5-turbo

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

halla

-0.88

poon

-0.78

etheless

-0.74

 surely

-0.72

fortunately

-0.71

ancy

-0.70

Ļ½

-0.69

 juven

-0.67

imei

-0.67

 awfully

-0.66

POSITIVE LOGITS

window

1.40

FILE

1.28

 Protesters

1.02

Buy

0.91

 Supporters

0.91

Supporters

0.90

Members

0.89

 Members

0.89

 Employees

0.88

 Photographer

0.86

Activations Density 0.120%

names of people and words related to government or politics

No Comments

No Known Activations