INDEX

Explanations

instances of the token "<bos>", suggesting it identifies the start of new segments or sections in the text

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

 nakalista

-0.64

 مشين

-0.62

carteles

-0.54

 CreateTagHelper

-0.54

 henvisninger

-0.54

Obrázky

-0.54

 Bioaccumulative

-0.54

 protoimpl

-0.53

 bandou

-0.50

protoimpl

-0.50

POSITIVE LOGITS

<h1>

0.48

<h2>

0.42

↵↵

0.41

ContentAlignment

0.40

<blockquote>

0.39

Pra

0.39

MeasureSpec

0.39

0.38

中

0.37

 susi

0.37

Activations Density 0.054%

instances of the token "<bos>", suggesting it identifies the start of new segments or sections in the text

No Comments

No Known Activations