INDEX

Explanations

actions and instructions related to communication and privacy

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

RegressionTest

-0.40

farwyddwr

-0.38

 Administrativna

-0.37

offsetof

-0.36

 crossorigin

-0.36

archiviato

-0.36

 >=",

-0.35

xically

-0.34

bebasan

-0.34

 ONCE

-0.33

POSITIVE LOGITS

httphttps

0.83

expandindo

0.71

ReusableCell

0.54

хьтан

0.52

EndGlobal

0.49

RTEE

0.49

 estekak

0.47

OGND

0.45

 cherchés

0.45

 onCreateView

0.45

Activations Density 0.039%

actions and instructions related to communication and privacy

No Comments

No Known Activations