INDEX

Explanations

terms related to educational programs and classes

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

arts, skills, and careers

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

apnews

-0.34

 rispondere

-0.33

VYMaps

-0.32

verifyException

-0.31

TRAILING

-0.30

 zure

-0.29

zeige

-0.29

 picioare

-0.29

 detik

-0.29

 atentos

-0.29

POSITIVE LOGITS

 ſte

0.67

 houſe

0.64

MLLoader

0.63

 faſt

0.62

 Houſe

0.60

 ſta

0.59

skills

0.58

ſelf

0.58

curricular

0.57

ffions

0.57

Activations Density 0.008%

terms related to educational programs and classes

arts, skills, and careers

No Comments

No Known Activations