INDEX

Explanations

mathematical or programming constructs and terminology

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

-0.42

ValueStyle

-0.35

mayın

-0.34

’

-0.34

-0.33

‘

-0.32

layoutControl

-0.31

an

-0.30

 hierfür

-0.30

Phương

-0.28

POSITIVE LOGITS

UnusedPrivate

0.73

Tikang

0.68

 houſe

0.66

 snippetHide

0.65

 Houſe

0.65

ſelves

0.65

 purpoſe

0.61

contentLoaded

0.60

RTEE

0.60



0.60

Activations Density 0.045%

mathematical or programming constructs and terminology

No Comments

No Known Activations