INDEX

Explanations

phrases related to legal matters and individuals involved in political scandals

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

references to political aides and advisers within a context of investigations and potential misconduct

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Dialogue

-0.69

izoph

-0.67

 ç¥ŀ

-0.66

thing

-0.66

luster

-0.66

[|

-0.65

INAL

-0.64

 locality

-0.62

Japan

-0.61

 Insect

-0.61

POSITIVE LOGITS

 Huma

1.36

 briefed

1.15

 Manafort

1.05

 advising

1.04

 Valerie

1.04

 confid

1.03

 tasked

1.02

 Abedin

1.01

 Priebus

0.97

 Cheryl

0.95

Activations Density 0.204%

phrases related to legal matters and individuals involved in political scandals

references to political aides and advisers within a context of investigations and potential misconduct

No Comments

No Known Activations