INDEX

Explanations

personal names, such as "Bush" and "Trump" along with related words

oai_token-act-pair · gpt-3.5-turbo

references to prominent political figures and their roles

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Translation

-0.63

 guiName

-0.62

âĶģ

-0.62

farious

-0.61

Ò

-0.61

igious

-0.60

actionDate

-0.59

soType

-0.58

%%%%

-0.58

 Include

-0.57

POSITIVE LOGITS

 cheated

1.10

 stole

1.06

 survived

1.03

 forgot

1.01

 loves

1.01

 died

1.00

 hates

1.00

 blew

0.99

 screwed

0.98

 overcame

0.97

Activations Density 0.625%

personal names, such as "Bush" and "Trump" along with related words

references to prominent political figures and their roles

No Comments

No Known Activations