INDEX

Explanations

punctuation marks and certain syntactic structures that suggest dialogue or questions

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

specific comment syntax

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

 zoude

-0.49

 dezelve

-0.47

 demikian

-0.45

 zdrowia

-0.44

 genieten

-0.43

correre

-0.43

 således

-0.42

 gedurende

-0.42

 totalidad

-0.41

 pertanto

-0.41

POSITIVE LOGITS

 autorytatywna

0.95

 PeEnEo

0.54

/*

0.54

Autoritní

0.54

ⓧ

0.53

WithIOException

0.52

supset

0.50

 defaultstate

0.50

 informée

0.50

 disambiguazione

0.50

Activations Density 0.103%

punctuation marks and certain syntactic structures that suggest dialogue or questions

specific comment syntax

No Comments

No Known Activations