INDEX

Explanations

elements of code, particularly comments and syntax

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

programming code

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Predecesor

-0.42

下载附件

-0.37

 चीज़ों

-0.36

 paid

-0.35

ApiModelProperty

-0.34

dolu

-0.34

soort

-0.33

ímetros

-0.33

 appealing

-0.33

 onData

-0.33

POSITIVE LOGITS

parsedMessage

0.56

 AssemblyCulture

0.54

WithIOException

0.53

ViewFeatures

0.53

DoubleQuotes

0.50

RefNanny

0.50

FieldBuilder

0.49

 المعيارى

0.48

 engraçadas

0.47

BeginInit

0.46

Activations Density 0.003%

elements of code, particularly comments and syntax

programming code

No Comments

No Known Activations