INDEX

Explanations

directional terms and positioning in a layout context

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

RegistryLite

-0.63

QMetaType

-0.51

GraphicsUnit

-0.47

 quedado

-0.45

parsedMessage

-0.44

IntoConstraints

-0.43

CodeAttribute

-0.43

setLayoutManager

-0.42

writeFieldEnd

-0.42

 disambiguazione

-0.42

POSITIVE LOGITS

InjectAttribute

0.40

 Studier

0.39

 appétit

0.38

 Bioaccumulative

0.37

 Censo

0.37

ner

0.36

InsertCommand

0.35

irkus

0.35

 Enzyme

0.35

 casinos

0.35

Activations Density 0.039%

directional terms and positioning in a layout context

No Comments

No Known Activations