INDEX

Explanations

sequences of structured data or coding syntax

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

CUDA device 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

Negative Logits

AddTagHelper

-0.52

KURZBESCHREIBUNG

-0.52

SequentialGroup

-0.51

AndroidJUnit

-0.48

usercontent

-0.47

Tikang

-0.44

 CreateTagHelper

-0.44

 nonUne

-0.44

Datuak

-0.43

__'

-0.43

POSITIVE LOGITS

 الرياضيه

0.46

 ivelany

0.42

 matelas

0.42

Jeografia

0.41

 渋谷

0.41

Specs

0.40

 Silverman

0.40



0.38

("

0.38

 saites

0.38

Activations Density 0.074%

sequences of structured data or coding syntax

CUDA device information

No Known Activations