INDEX

Explanations

references to specific terms and categories within a structured context

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

French correct answers

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.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

featureID

-0.60



-0.54

OGND

-0.47

 localObject

-0.47

 aDecoder

-0.43

AnchorStyles

-0.43

 sqlSession

-0.41

 defaultstate

-0.41

AddTagHelper

-0.40

 noDo

-0.40

POSITIVE LOGITS

دانشنامهٔ

0.46

قيقي

0.44

әрмәләр

0.43

esgue

0.43

oredCriteria

0.41

haustible

0.40

withIdentifier

0.40

 Taktlose

0.39

 חיצוניים

0.39

 encontraban

0.39

Activations Density 0.130%

references to specific terms and categories within a structured context

French correct answers

No Comments

No Known Activations