INDEX

Explanations

phrases indicating a sequence or relationship between actions or events

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

 chrétien

-0.45

 curé

-0.44

 Grecs

-0.42

 cahier

-0.41

konomi

-0.41

Примечания

-0.40

 américa

-0.40

 amerik

-0.39

 conscience

-0.38

 église

-0.38

POSITIVE LOGITS

 gyhoeddwyd

0.63

:✨

0.60

 utafitiHapana

0.59

+#+#

0.59

 IBOutlet

0.57

0.56

 Audiodateien

0.56

 ditambah

0.55

脚注の使い方

0.54

ngdoc

0.54

Activations Density 0.052%

phrases indicating a sequence or relationship between actions or events

No Comments

No Known Activations