INDEX

Explanations

specific character names and titles in a narrative context

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

numbers and technical terms

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

RegressionTest

-0.45

цездатний

-0.44

RectangleBorder

-0.43

mybatisplus

-0.41

 extra

-0.37

windowFixed

-0.37

netbeans

-0.35

TextBlock

-0.34

 patched

-0.33

 ILogger

-0.33

POSITIVE LOGITS

 Normdatei

0.68

 surla

0.54

__*/

0.50

الحياه

0.50

NewUrlParser

0.50

accompag

0.49

Климат

0.47

 ProtoMessage

0.47

 Meksiku

0.47

operazione

0.47

Activations Density 0.202%

specific character names and titles in a narrative context

numbers and technical terms

No Comments

No Known Activations