INDEX

Explanations

references to academic citations or document identifiers

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

URLs or identifiers

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

apimachinery

-0.53

WebElementEntity

-0.51

 nahilalakip

-0.50

MIDDLEWARE

-0.47

 ویکی‌پدیا

-0.44

StandardCharsets

-0.43

ReusableCell

-0.42

 otomatig

-0.41

 محفوظة

-0.41

SBATCH

-0.40

POSITIVE LOGITS

 jspb

0.52

RegressionTest

0.50

 '\\;'

0.44

 CHtml

0.42

 jScrollPane

0.40

 ExecuteAsync

0.40

 jadx

0.39

ViewFeatures

0.39

 wireType

0.38

 WaitForSeconds

0.38

Activations Density 0.103%

references to academic citations or document identifiers

URLs or identifiers

No Known Activations