INDEX

Explanations

phrases or terms related to software and programming constructs

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

code and programming identifiers

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

ſcher

-0.60

linawan

-0.56

lany

-0.53

hilangan

-0.52

ðsíða

-0.52

 auroit

-0.50

iſche

-0.49

ſelben

-0.49

holding

-0.48

ſſung

-0.47

POSITIVE LOGITS

EnableWeb

0.48

 jsPsych

0.41

saraba

0.40

GEBURTSDATUM

0.40

postsleuth

0.35

ArgumentParser

0.35

 nonUne

0.35

Carriera

0.35

TagMode

0.35

 AssemblyProduct

0.34

Activations Density 0.171%

phrases or terms related to software and programming constructs

code and programming identifiers

No Comments

No Known Activations