INDEX

Explanations

programming constructs related to enumerations and game mechanics

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

creature instances and data

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

 dear

-0.34

Erase

-0.32

 iterations

-0.32

liato

-0.31

 sloth

-0.30

accede

-0.30

zij

-0.30

 Krieges

-0.30

 gazelle

-0.30

 catast

-0.30

POSITIVE LOGITS

 lenker

0.59

Datuak

0.59

SourceChecksum

0.57

 informée

0.56

RTGC

0.55

rungsseite

0.55

таратура

0.54

 kasarigan

0.54

-------------</

0.54

Nuorodos

0.53

Activations Density 0.021%

programming constructs related to enumerations and game mechanics

creature instances and data

No Comments

No Known Activations