INDEX

Explanations

numerical or coding references related to data processing or validation tasks

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_18/width_65k/average_l0_88

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.18.hook_mlp_out

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

uſ

-0.52

tonsoft

-0.52

 fair

-0.51

AddRef

-0.50

 lote

-0.50

IELD

-0.50

 raiſ

-0.50

StoredProcedure

-0.49

 Sert

-0.49

 lamella

-0.48

POSITIVE LOGITS

memoized

0.68

AndroidJUnit

0.63

حياته

0.58

]$}

0.56

__*/

0.54

 Италијани

0.53

WebServlet

0.53

锈钢

0.51

setVerticalGroup

0.51

 téléphonique

0.50

Activations Density 0.214%

numerical or coding references related to data processing or validation tasks

No Comments

No Known Activations