INDEX

Explanations

sequences of numbers or symbols, particularly those indicative of formatting or coding

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

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

protoimpl

-0.76

 المعيارى

-0.63

دانشنامهٔ

-0.59

StructEnd

-0.57

DeleteBehavior

-0.53

 springfox

-0.51

 BoxFit

-0.50

 ſche

-0.49

SequentialGroup

-0.49

ArrowToggle

-0.48

POSITIVE LOGITS

 Pellegrini

0.44

 Aussicht

0.44

httphttps

0.44

 Cárdenas

0.44

jenigen

0.43

 heutige

0.43

 proceeded

0.41

 Krankheiten

0.41

 föruts

0.40

kwür

0.40

Activations Density 0.138%

sequences of numbers or symbols, particularly those indicative of formatting or coding

No Comments

No Known Activations