INDEX

Explanations

phrases related to technical and scientific processes

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

data, estimates, detection

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

 Infór

-0.43

 kolye

-0.42

 cairan

-0.40

 zijne

-0.40

 Harcourt

-0.40

 Arquitectura

-0.40

 território

-0.40

fjspx

-0.39

 kebijakan

-0.39

 dimaksud

-0.39

POSITIVE LOGITS

 صوتيه

0.48

 CanadaChoose

0.47

AndEndTag

0.46

DataSnapshot

0.44

0.43

 Савезне

0.41

-------------</

0.41

 acci

0.40

 OpenAPI

0.40

migrationBuilder

0.40

Activations Density 0.183%

phrases related to technical and scientific processes

data, estimates, detection

No Comments

No Known Activations