INDEX

Explanations

mathematical symbols and notation associated with protective measures or constraints in formal contexts

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

protect against opponent

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

 ProtoMessage

-0.51

 ſch

-0.49

 anhänger

-0.44

 dieß

-0.41

 exercí

-0.39

 nocy

-0.39

 stanga

-0.39

 miniaturka

-0.38

 rigue

-0.38

 leſs

-0.35

POSITIVE LOGITS

 للمعارف

0.73

MLLoader

0.71

yarnpkg

0.67

 <<<<<<<<<<<<<<

0.66

'{@

0.60

出版年

0.59

AddTagHelper

0.59

 שוליים

0.58

 تانيه

0.58

!*\

0.57

Activations Density 0.080%

mathematical symbols and notation associated with protective measures or constraints in formal contexts

protect against opponent

No Comments

No Known Activations