INDEX

Explanations

terms related to adjustability and fitting

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

adjustable parts

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

 surla

-0.51

 المعيارى

-0.50

ScopeManager

-0.48

 LiveData

-0.47

Vidite

-0.45

ьаж

-0.44

räger

-0.44

 censiti

-0.44

VersionUID

-0.43

 للمعارف

-0.42

POSITIVE LOGITS

 adjusts

0.50

 adjustable

0.49

adjustable

0.48

Adjust

0.48

 adjust

0.47

 adjustment

0.47

IntoConstraints

0.46

 adjustments

0.43

 Adjust

0.41

 adjusting

0.41

Activations Density 0.009%

terms related to adjustability and fitting

adjustable parts

No Comments

No Known Activations