INDEX

Explanations

the formatting and syntactic structure of code or JSON-like data

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

unsigned long declarations

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.24.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

 ब्रेकडाउन

-0.57

ftagPool

-0.56

 IBOutlet

-0.56

 चीज़ों

-0.56

ViewImports

-0.55

MigrationBuilder

-0.55

Aktualisiert

-0.54

 AssemblyCulture

-0.54

AndEndTag

-0.53

KommentareTeilen

-0.52

POSITIVE LOGITS

でも

0.37

り

0.36

 dAtA

0.35

ULT

0.35

Gas

0.33

Fact

0.33

Sho

0.32

<!--

0.32

Activations Density 0.126%

the formatting and syntactic structure of code or JSON-like data

unsigned long declarations

No Comments

No Known Activations