INDEX

Explanations

header-like structures or markers in the text

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

section markers {#sec

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

qtype

-0.32

汰

-0.26

 başlad

-0.26

Toponymie

-0.24

 sumpay

-0.24

 geeignet

-0.24

Statistiche

-0.24

 Geschäfte

-0.23

 deshalb

-0.23

 Erhöhung

-0.23

POSITIVE LOGITS

 müſſen

0.76

 كومونز

0.76

Autoritní

0.76

 deſſen

0.75

 geweſen

0.71

 zuſammen

0.71

 verſch

0.70

 ſelbſt

0.70

rungsseite

0.70

IntoConstraints

0.70

Activations Density 0.109%

header-like structures or markers in the text

section markers {#sec

No Comments

No Known Activations