INDEX

Explanations

logical implications or conclusions in mathematical proofs

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

mathematical ⇒ symbols

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.24.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

<em>

-0.46

<i>

-0.46

 Bewohner

-0.46

 unterkunft

-0.44

 Arbeitnehmer

-0.43

 Eigentümer

-0.43

 enfance

-0.40

 Pitman

-0.39

home

-0.38

 home

-0.38

POSITIVE LOGITS

Rightarrow

1.09

⇒

0.82

⇒

0.80

SerializeField

0.76

 InputDecoration

0.68

 noDo

0.68

featureID

0.67

AndroidJUnit

0.67

 AssemblyVersion

0.66

ſcher

0.66

Activations Density 0.005%

logical implications or conclusions in mathematical proofs

mathematical ⇒ symbols

No Comments

No Known Activations