INDEX

Explanations

copyright-related terms and formats

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

copyright notices

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

 surla

-0.84

RuleContext

-0.60

σιμοποι

-0.60

extAlignment

-0.60

ніципалі

-0.59

iſche

-0.59

majánló

-0.59

ésultats

-0.59

 незавершена

-0.58

]")]

-0.58

POSITIVE LOGITS

ViewImports

0.48

 насељу

0.35

 editorial

0.29

 newOwner

0.29

Bibliograf

0.28

 pré

0.28

Насе

0.28

 final

0.27

期刊论文

0.27

ignées

0.25

Activations Density 0.002%

copyright-related terms and formats

copyright notices

No Comments

No Known Activations