INDEX

Explanations

structured references and identifiers related to categories or modules

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

place names and categories

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

rungsseite

-1.05

 autorytatywna

-0.99

 يتيمه

-0.98

GEBURTSDATUM

-0.85

MigrationBuilder

-0.84

 kasarigan

-0.82

 betweenstory

-0.80

 AssemblyCulture

-0.79

ſſung

-0.78

 Administrativna

-0.77

POSITIVE LOGITS

<eos>

0.67

setOn

0.33

www

0.31

 hope

0.30

www

0.28

and

0.28

oden

0.28

Bool

0.28

 next

0.28

tabular

0.28

Activations Density 0.012%

structured references and identifiers related to categories or modules

place names and categories

No Comments

No Known Activations