INDEX

Explanations

phrases indicating completed actions or achievements

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

past participles

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

 patrulla

-0.38

 cuarzo

-0.37

 alberi

-0.35

 diferite

-0.35

 spese

-0.35

vė

-0.34

føring

-0.34

 imprese

-0.34

 avanti

-0.33

 égale

-0.33

POSITIVE LOGITS

 Autorizaciones

0.58

 محفوظة

0.57

HasBeenSet

0.57

migrationBuilder

0.56

 ویکی‌آمباردا

0.56

EnumMember

0.55

providedIn

0.55

modelBuilder

0.54

 EconPapers

0.53

ngOnDestroy

0.53

Activations Density 0.054%

phrases indicating completed actions or achievements

past participles

No Comments

No Known Activations