INDEX

Explanations

references to data structures, particularly lists and collections

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

 pouce

-0.47

atangan

-0.40

">//

-0.39

евич

-0.39

 pont

-0.39

PILE

-0.38

 Jump

-0.37

Jump

-0.37

 presentar

-0.36

SKIP

-0.36

POSITIVE LOGITS

 bezeichneter

0.78

ILayout

0.76

 ivelany

0.75

IndentedString

0.75

ValueStyle

0.75

transQ

0.74

 myſelf

0.73

 AppModule

0.72

CompleteListener

0.72

 Baillargeon

0.72

Activations Density 0.006%

references to data structures, particularly lists and collections

No Comments

No Known Activations