INDEX

Explanations

elements related to programming constructs and array manipulations

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

order statistics of permutations

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

 ujednoznacz

-0.56

verwijspagina

-0.55

Rüyada

-0.50

rungsseite

-0.48

ImageContext

-0.47

 miniaturka

-0.45

GIVEREF

-0.44

sizeCache

-0.43

 TestBed

-0.42

 senhora

-0.41

POSITIVE LOGITS

 CAPIT

0.47

 '\\;'

0.45

oto

0.45

kehr

0.43

Esco

0.43

UnitOfWork

0.42

 Ains

0.42

 SUPER

0.42

rhe

0.41

ky

0.41

Activations Density 0.007%

elements related to programming constructs and array manipulations

order statistics of permutations

No Comments

No Known Activations