INDEX

Explanations

high-frequency terms indicating numerals, counts, or significant qualifiers in a context related to technical or analytical writing

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

elaborates, rest of, eventually run

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

adaptiveStyles

-0.42

 snippetHide

-0.41

 kaarangay

-0.40

 Forder

-0.40

Externé

-0.35

curkan

-0.35

 CONC

-0.34

ienza

-0.34

 Económica

-0.34

 Wikiseite

-0.34

POSITIVE LOGITS

 AssemblyTitle

0.51

帖最后由

0.50

setVerticalGroup

0.46

 مرئيه

0.44

脚注の使い方

0.42

ArrowToggle

0.42

 Audiodateien

0.41

ConstraintMaker

0.41

detectChanges

0.40

 <=",

0.40

Activations Density 0.179%

high-frequency terms indicating numerals, counts, or significant qualifiers in a context related to technical or analytical writing

elaborates, rest of, eventually run

No Comments

No Known Activations