INDEX

Explanations

phrases that include references to the reader or the speaker, indicating direct communication

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

pronouns (you, us, me)

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

 colorear

-0.45

 MonoBehaviour

-0.44

 esfuer

-0.43

 demás

-0.39

 ruban

-0.37

 nadie

-0.37

 muere

-0.37

 lleg

-0.37

CheckedChanged

-0.35

 ferons

-0.35

POSITIVE LOGITS

Panamoan

0.67

0.66

botto

0.61

MemoryWarning

0.59

an

0.59

 Wikimedijinoj

0.57

 ddelweddau

0.57

ophthal

0.57

ſſung

0.57

concel

0.56

Activations Density 0.015%

phrases that include references to the reader or the speaker, indicating direct communication

pronouns (you, us, me)

No Comments

No Known Activations