INDEX

Explanations

elements of code documentation, particularly file and software information annotations

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

code comments and special characters

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

astéro

-0.38

angliski

-0.37

EndContext

-0.36

 numerus

-0.35

期刊论文

-0.34

gewiesen

-0.34

 ویکی‌آمباردا

-0.34

 kohdetta

-0.34

libft

-0.34

 señal

-0.33

POSITIVE LOGITS

VersionUID

0.68

KURZBESCHREIBUNG

0.66

WriteTagHelper

0.56

rungsseite

0.55

addContainerGap

0.52

GEBURTS

0.49

 ujednoznacz

0.49

sidemargin

0.49

indd

0.47

0.45

Activations Density 0.006%

elements of code documentation, particularly file and software information annotations

code comments and special characters

No Comments

No Known Activations