INDEX

Explanations

numerical values or identifiers related to programming or coding contexts

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

 antaranya

-0.52

ppuden

-0.52

 scattata

-0.46

demás

-0.42

Sekolah

-0.41

 SEDS

-0.40

 Gitar

-0.40

gorro

-0.39

 podéis

-0.39

 saurait

-0.38

POSITIVE LOGITS

 '\\;'

0.71

__":

0.58

0.56

__*/

0.55

 CreateTagHelper

0.50

TagMode

0.50

__':

0.50

]]);

0.49

Autoritní

0.49

الإنجليزية

0.47

Activations Density 0.002%

numerical values or identifiers related to programming or coding contexts

No Comments

No Known Activations