INDEX

Explanations

concepts related to identity and belonging within narratives

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

 Expert

-0.52

ModelSerializer

-0.52

 Scientific

-0.52

iſen

-0.49

 Wikimedijinoj

-0.49

 Neural

-0.49

 queſta

-0.48

+#+

-0.48

出版年

-0.47

 Hormone

-0.46

POSITIVE LOGITS

 Räume

0.47

 Gegenstand

0.43

 Kontexten

0.43

 explored

0.40

IsPostBack

0.38

 explorations

0.38

 Brücke

0.38

izante

0.38

 Verständnis

0.37

 discernible

0.37

Activations Density 0.037%

concepts related to identity and belonging within narratives

No Comments

No Known Activations