INDEX

Explanations

proper nouns, specifically names of individuals and organizations

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

names followed by titles or events

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

oredCriteria

-0.64

 protoimpl

-0.56

RTSC

-0.44

ruptedException

-0.44

ErrUnexpectedEOF

-0.43

ᅠ

-0.42

Хьажоргаш

-0.42

HandlerContext

-0.40

 мәкал

-0.40

 outputStream

-0.39

POSITIVE LOGITS

AddTagHelper

0.54

 sekali

0.49

脚注の使い方

0.45

transQ

0.40

الدراسه

0.39

Erstellt

0.38

 للمعارف

0.37

colhead

0.37

AxisAlignment

0.35

cèse

0.35

Activations Density 0.019%

proper nouns, specifically names of individuals and organizations

names followed by titles or events

No Comments

No Known Activations