INDEX

Explanations

code snippets and programming-related terms

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_33/width_131k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.33.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

 Majefty

-0.52

 kuiten

-0.51

 pleaſure

-0.51

 ſtre

-0.49

neſs

-0.48

 ſta

-0.47

 quæ

-0.45

wiſe

-0.44

 ſch

-0.44

 Monfieur

-0.43

POSITIVE LOGITS

 cucumber

0.67

 Cucumber

0.63

Cucumber

0.58

cucumber

0.57

 cucumbers

0.53

CUC

0.51

🥒

0.50

 огур

0.50

 tread

0.50

contentLoaded

0.48

Activations Density 0.003%

code snippets and programming-related terms

No Comments

No Known Activations