INDEX

Explanations

numeric representations or sequences

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

numbers followed by units or counts

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

 féminin

-0.48

 osallist

-0.46

 تضيفلها

-0.45

 שוליים

-0.45

 actrices

-0.45

Poppins

-0.44

 høre

-0.44

 sorpresas

-0.44

 laporan

-0.43

 duradero

-0.43

POSITIVE LOGITS

AddTagHelper

0.63

 ویکی‌پدی

0.59

ModelAndView

0.58

 parentId

0.51

 projectId

0.51

 totalPrice

0.51

Географиясе

0.51

 Parkway

0.50

 Myron

0.48

 concor

0.48

Activations Density 0.097%

numeric representations or sequences

numbers followed by units or counts

No Comments

No Known Activations