INDEX

Explanations

references to legal terminology and processes

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

afterwards or permanent completion

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

Autoritní

-0.44

 informée

-0.38

MessageState

-0.38

MetaObject

-0.37

CERTIFICATE

-0.35

 cholera

-0.35

WireFormatLite

-0.35

 Choice

-0.35

 asisti

-0.35

hezza

-0.34

POSITIVE LOGITS

 danach

0.46

 afterwards

0.42

 PERMAN

0.42

Afterwards

0.41

permanent

0.41

__*/

0.41

 permanent

0.41

eqn

0.41

FirebaseUser

0.40

 purpoſe

0.40

Activations Density 0.049%

references to legal terminology and processes

afterwards or permanent completion

No Comments

No Known Activations