INDEX

Explanations

instances of high numerical values

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

Category classification

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

ioutil

-0.57

 saveiro

-0.57

MethodManager

-0.57

 calendriers

-0.56

 indígen

-0.54

 ویکی‌پدی

-0.54

 vlasy

-0.52

genodigd

-0.51

 gehouden

-0.51

 hilsen

-0.51

POSITIVE LOGITS

↵↵↵

0.78

↵↵↵↵

0.68

↵↵↵↵↵

0.64

 Stahl

0.56

↵↵↵↵↵↵↵

0.55

↵↵↵↵↵↵↵↵

0.54

↵↵↵↵↵↵↵↵↵

0.53

↵↵↵↵↵↵

0.52

 McCle

0.50

mstyle

0.48

Activations Density 0.018%

instances of high numerical values

Category classification

No Comments

No Known Activations