INDEX

Explanations

mentions of weekly news and updates within a community or publication context

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

news and trends

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

principalColumn

-0.57

UnusedPrivate

-0.52

 Normdatei

-0.52

 ComVisible

-0.50

OGND

-0.50

Autoritní

-0.48

 ***!

-0.47

elemField

-0.47

GenerationType

-0.45

 jsPsych

-0.45

POSITIVE LOGITS

trends

0.54

 trends

0.54

 happenings

0.53

omation

0.51

 zaken

0.49

 affari

0.47

 news

0.46

 brief

0.46

adan

0.45

comings

0.44

Activations Density 0.021%

mentions of weekly news and updates within a community or publication context

news and trends

No Comments

No Known Activations