INDEX

Explanations

references to programming constructs related to classes and objects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 1-gemmascope-mlp-16k

Configuration

google/gemma-scope-2b-pt-mlp/layer_1/width_16k/average_l0_105

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.1.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

PYX

-0.59

源码

-0.48

 היש

-0.47

 FORTRAN

-0.47

Injectable

-0.46

 imageNamed

-0.45

 Fortran

-0.45

жалуйста

-0.44

ca

-0.43

消毒

-0.42

POSITIVE LOGITS

脚注の使い方

0.83

 Вікі

0.79

 صوتيه

0.66

Välislingid

0.65

TypedDataSet

0.63

 ویکی‌پدیای

0.63

Története

0.62



0.60

はじめに

0.60

جستارهای

0.59

Activations Density 0.750%

references to programming constructs related to classes and objects

No Comments

No Known Activations