INDEX

Explanations

hyphenated or negative numerical values

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

symbols and code identifiers

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

uxxxx

-0.65

transQ

-0.62

 Verſ

-0.59

expandindo

-0.56

 ſou

-0.56

 ſche

-0.54

 Inſ

-0.53

toMatchSnapshot

-0.52

zul

-0.52

 faſt

-0.52

POSITIVE LOGITS

mappedBy

0.51

 يتيمه

0.42

WriteTagHelper

0.39

StoryboardSegue

0.38

脚注の使い方

0.38

 فريبيس

0.38

enumi

0.36

 nahilalakip

0.34

četně

0.34

 cherchés

0.33

Activations Density 0.029%

hyphenated or negative numerical values

symbols and code identifiers

No Comments

No Known Activations