INDEX

Explanations

categories and hierarchical structures in data

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

archiviato

-0.47

CrossRef

-0.46

 hearing

-0.44

*/].

-0.42

AndEndTag

-0.42

gucigalpa

-0.42

toHaveBeen

-0.42

CheckBreak

-0.41

 inds

-0.41

########.

-0.40

POSITIVE LOGITS

LookAnd

0.60

SharedDtor

0.45

 Exacts

0.43

 createSlice

0.41

 diferite

0.39

 ontstaan

0.39

 requestCode

0.38

rouvez

0.38

(;;

0.38

 endregion

0.37

Activations Density 0.193%

categories and hierarchical structures in data

No Comments

No Known Activations