INDEX

Explanations

references to time-related terms and intervals

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

time periods and durations

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

 cerveja

-0.33

 quedo

-0.31

 possible

-0.30

 spéciaux

-0.30

 reaction

-0.29

SPATH

-0.29

 posibles

-0.29

 quick

-0.29

 répondit

-0.28

лючение

-0.28

POSITIVE LOGITS

 nahilalakip

0.79

 للاسماء

0.76

DAY

0.75

Day

0.74

day

0.74

 الحره

0.73

Хьажоргаш

0.73

 pinulongan

0.73

vernight

0.72

morning

0.71

Activations Density 0.065%

references to time-related terms and intervals

time periods and durations

No Comments

No Known Activations