INDEX

Explanations

references to time periods, specifically those indicating years, months, days, and weeks

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

time periods ago

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

RectangleBorder

-0.63

OrNil

-0.54

 snippetHide

-0.53

aarrggbb

-0.53

évaluateur

-0.49

 defaultstate

-0.48

StoryboardSegue

-0.48

 referenties

-0.48

NameInMap

-0.46

帖最后由

-0.46

POSITIVE LOGITS

ago

0.68

 earlier

0.55

 yester

0.53

Hace

0.52

 Wart

0.50

joba

0.50

Earlier

0.49

 recent

0.49

 recently

0.48

 Earlier

0.48

Activations Density 0.015%

references to time periods, specifically those indicating years, months, days, and weeks

time periods ago

No Comments

No Known Activations