INDEX

Explanations

references to violent actions or weapons

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

weapons and attacks

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

verwijspagina

-0.61

msgTypes

-0.53

 dezelve

-0.49

 propOrder

-0.49

 للمعارف

-0.48

__(/*!

-0.47

 esigenze

-0.46

transQ

-0.46

endpush

-0.46

hamshire

-0.46

POSITIVE LOGITS

 weapon

0.51

 weapons

0.40

 attack

0.40

weapon

0.40

 deadly

0.37

 powerful

0.37

 attacks

0.36

attack

0.35

KeySpec

0.35

Weapon

0.34

Activations Density 0.144%

references to violent actions or weapons

weapons and attacks

No Comments

No Known Activations