INDEX

Explanations

references to tables and their arrangement in a space

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

脚注の使い方

-0.43

 nahilalakip

-0.37

 sexually

-0.32

 ervan

-0.31

ujudkan

-0.31

ISION

-0.31

 wektor

-0.30

țion

-0.30

 invokingState

-0.30

 része

-0.29

POSITIVE LOGITS

 Dining

0.68

 tablecloth

0.66

AddTagHelper

0.66

LEncoder

0.65

Dining

0.64

Dine

0.63

 Meal

0.61

 dining

0.61

 table

0.60

 diners

0.60

Activations Density 0.004%

references to tables and their arrangement in a space

No Comments

No Known Activations