INDEX

Explanations

negations or phrases indicating invalidity or errors

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

is or is not

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

principalTable

-0.62

-0.59

Hozzáférés

-0.51

hyrchwyd

-0.49

 autorytatywna

-0.48

ArgsConstructor

-0.47

########.

-0.47

SourceChecksum

-0.46

 defStyleAttr

-0.45

BeginInit

-0.44

POSITIVE LOGITS

tvguidetime

0.46

 tóxic

0.45

 weak

0.41

 scarce

0.40

DidLoad

0.40

condition

0.40

 linear

0.39

ésult

0.39

 condition

0.39

belongs

0.39

Activations Density 0.018%

negations or phrases indicating invalidity or errors

is or is not

No Comments

No Known Activations