INDEX

Explanations

patterns related to formatting or structural elements in the text

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

section separators

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

Negative Logits

 المعيارى

-0.97

WriteTagHelper

-0.82

MLLoader

-0.82

出版年

-0.79

 createSprite

-0.79

mpagne

-0.78

 ویکی‌پدی

-0.77

 Мексичка

-0.77

 يتيمه

-0.75

iſten

-0.74

POSITIVE LOGITS

0.43

0.36

↵

0.35

メント

0.35

0.34

↵↵

0.34

0.33

<strong>

0.33

Activations Density 0.062%

patterns related to formatting or structural elements in the text

section separators

No Known Activations