INDEX

Explanations

structural elements and values in scientific data formats

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

greater than or less than

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

RefNanny

-0.37

ברס

-0.35

 whole

-0.35

 ści

-0.34

 tục

-0.33

ArrowToggle

-0.32

Уда

-0.32

데

-0.32

整

-0.32

ỗi

-0.31

POSITIVE LOGITS

beyond

0.66

Beyond

0.65

 beyond

0.64

YOND

0.63

withOpacity

0.62

 Beyond

0.60

 BEYOND

0.60

GEBURTSDATUM

0.59

 NSCoder

0.59

 exceeding

0.59

Activations Density 0.394%

structural elements and values in scientific data formats

greater than or less than

No Comments

No Known Activations