INDEX

Explanations

sequences associated with programming constructs or structures

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

 дописавши

-0.93

fromnode

-0.93

 الرياضيه

-0.93

principalColumn

-0.92

oredCriteria

-0.90

 betweenstory

-0.88

ंदीखरीदारी

-0.86

 snippetHide

-0.86

 CreateTagHelper

-0.84

المناصب

-0.83

POSITIVE LOGITS

0.47

0.46

↵↵

0.45

We

0.44

↵

0.44

0.43

View

0.43

The

0.41

0.40

Activations Density 0.026%

sequences associated with programming constructs or structures

No Comments

No Known Activations