INDEX

Explanations

proper nouns or names related to various subjects

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

specific capitalized names

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

 betweenstory

-0.46

 сцен

-0.42

mybatisplus

-0.41

 ***!

-0.41

Sklici

-0.41

 للمعارف

-0.38

Externé

-0.37

 Messieurs

-0.37

FromNib

-0.35

 pères

-0.35

POSITIVE LOGITS

RenderAtEndOf

0.65

Laure

0.52

++++++++++++++++

0.51

✢

0.50

nachron

0.50

 czę

0.49

orszá

0.49

resentation

0.48

erçe

0.47

 wyda

0.47

Activations Density 0.047%

proper nouns or names related to various subjects

specific capitalized names

No Comments

No Known Activations