INDEX

Explanations

symbols and characters that may denote various types of data or formatting elements

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

mathematical and code fragments

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

saraba

-0.85

PreferredItem

-0.74

ſammen

-0.72

 فريبيس

-0.71

ſelben

-0.71

 Bewußt

-0.68

 ddelwed

-0.66

帖最后由

-0.66

 queſta

-0.65

 AssemblyCompany

-0.65

POSITIVE LOGITS

 Already

0.28

zień

0.27

0.26

das

0.26

">//

0.25

˜

0.25

ynb

0.25

0.24

Sau

0.23

Activations Density 0.001%

symbols and characters that may denote various types of data or formatting elements

mathematical and code fragments

No Comments

No Known Activations