INDEX

Explanations

elements related to high numerical values or statistical terms

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

special characters, code, or specific names

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

///<

-0.35

And

-0.30

 ilman

-0.28

jspb

-0.28

 olmayan

-0.28

 inférieur

-0.27

By

-0.27

 Näch

-0.27

 oczywiście

-0.27

 omkring

-0.27

POSITIVE LOGITS

ロウィン

0.84

 パンチラ

0.79

 tartalo

0.79

 vooz

0.79

 zwiſchen

0.78

 nahil

0.76

 wiſſen

0.76

 好文分享

0.75

 intptr

0.75

 Weiſe

0.75

Activations Density 0.037%

elements related to high numerical values or statistical terms

special characters, code, or specific names

No Comments

No Known Activations