INDEX

Explanations

keywords associated with programming constructs and definitions in code

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

code definitions and structures

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

createStatement

-0.40

 weiterer

-0.36

 tambien

-0.34

 autres

-0.34

Greetings

-0.33

 otros

-0.32

disposing

-0.32

สึก

-0.32

 préféré

-0.31

DeleteBehavior

-0.31

POSITIVE LOGITS

 autorytatywna

0.65

WithIOException

0.56

RTGC

0.56

StoryboardSegue

0.55

 Coher

0.54

sfw

0.54

EDEFAULT

0.54

ulose

0.52

 bureau

0.52

 للمعارف

0.52

Activations Density 0.002%

keywords associated with programming constructs and definitions in code

code definitions and structures

No Comments

No Known Activations