INDEX

Explanations

references to formal statements or claims in mathematical and theoretical contexts

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

mathematical propositions

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

 deudas

-0.48

 Turquía

-0.48

-0.47

Berikut

-0.47

Incluye

-0.46

 hindurch

-0.45

AnimationsModule

-0.45

 Bélgica

-0.44

Leak

-0.44

Knife

-0.43

POSITIVE LOGITS

 Proposition

1.02

 proposition

1.02

proposition

0.97

Proposition

0.90

 propositions

0.78

Disposition

0.71

 Disposition

0.68

 Position

0.68

 POSITION

0.66

 Exposition

0.65

Activations Density 0.011%

references to formal statements or claims in mathematical and theoretical contexts

mathematical propositions

No Comments

No Known Activations