INDEX

Explanations

instances of rules or guidelines for proper behavior and etiquette

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

use feedback

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

 guidance

-0.35

 vPvB

-0.34

 participants

-0.31

setDescription

-0.31

 participantes

-0.31

basicConfig

-0.29

 Treff

-0.29

 $(\%)$

-0.29

Résumé

-0.29

 acesta

-0.29

POSITIVE LOGITS

 disambiguazione

0.81

 виправивши

0.77

يكب

0.75

ArrowToggle

0.65

 Италијани

0.64

 يتيمه

0.62

点此举报

0.62

 informée

0.59

RegressionTest

0.57

djangoproject

0.56

Activations Density 0.016%

instances of rules or guidelines for proper behavior and etiquette

use feedback

No Comments

No Known Activations