INDEX

Explanations

numerals and terms associated with rankings or ratings

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

betting, odds, favourite, lowest, tickets

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



-0.78

RegressionTest

-0.77

IUrlHelper

-0.74

 estekak

-0.68

GEBURTSDATUM

-0.65

 AssemblyTitle

-0.63

 يتيمه

-0.60

ValueStyle

-0.59

 surla

-0.59

########.

-0.58

POSITIVE LOGITS

omin

0.36

 Tages

0.35

 ticket

0.35

 hubiese

0.35

 Rood

0.34

 impro

0.33

veter

0.33

 mérit

0.32

 Abends

0.32

 tickets

0.31

Activations Density 0.002%

numerals and terms associated with rankings or ratings

betting, odds, favourite, lowest, tickets

No Comments

No Known Activations