INDEX

Explanations

phrases related to inquiry and understanding

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

interactions and questions

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

 yourselves

-0.47

 appetites

-0.43

 amateurs

-0.43

 tetrachloride

-0.42

 fellas

-0.41

 ourselves

-0.41

 budgets

-0.40

 locals

-0.39

 chemists

-0.39

our

-0.39

POSITIVE LOGITS

 himself

0.78

himself

0.71

 للمعارف

0.64

 BorderSide

0.60

 człowiek

0.54

ChildScrollView

0.53

 Himself

0.52

eseorang

0.51

his

0.51

 mwenye

0.50

Activations Density 0.191%

phrases related to inquiry and understanding

interactions and questions

No Comments

No Known Activations