INDEX

Explanations

indices related to multi-dimensional arrays or matrices

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

mathematical indices `i`

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

 berkom

-0.44

 चीज़ों

-0.40

 Photocase

-0.38

GRAN

-0.38

">—

-0.37

).__

-0.36

numerusform

-0.35

 STEV

-0.35

magitan

-0.35

气

-0.34

POSITIVE LOGITS

CloseOperation

0.52

 الاطلاع

0.49

DockStyle

0.47

setVerticalGroup

0.46

Välislingid

0.46

 GenerationType

0.45

 المعيارى

0.45

énario

0.43

 informée

0.42

ParallelGroup

0.42

Activations Density 0.075%

indices related to multi-dimensional arrays or matrices

mathematical indices `i`

No Comments

No Known Activations