INDEX

Explanations

phrases that involve communication about health-related issues and instructions

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

telling someone information

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

 bất

-0.39

UrlResolution

-0.35

ace

-0.35

Ace

-0.33

 published

-0.33

FlatList

-0.33

 Draper

-0.32

 player

-0.32

ultima

-0.32

ETTI

-0.32

POSITIVE LOGITS

 Infórmanos

0.75

 nakalista

0.65

XmlAccessorType

0.61

 الرياضيه

0.58

Билгалдахарш

0.57

 initState

0.56

tagHelperRunner

0.55

Referencoj

0.54

تقاوى

0.54

 Administrativna

0.53

Activations Density 0.021%

phrases that involve communication about health-related issues and instructions

telling someone information

No Comments

No Known Activations