INDEX

Explanations

references to processes and tools related to research and technical methodologies

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

NameInMap

-0.73

 AssemblyVersion

-0.69

featureID

-0.65

....)

-0.65

ReusableCell

-0.60

 autorytatywna

-0.60

Personensuche

-0.59

 wireType

-0.59

+#+

-0.58

 Taktlose

-0.58

POSITIVE LOGITS

 yalnızca

0.38

 jedoch

0.34

 lecz

0.33

 usan

0.33

empêcher

0.32

 Afterward

0.32

ほうが

0.32

ạm

0.32

 perciò

0.32

 zufolge

0.32

Activations Density 1.394%

references to processes and tools related to research and technical methodologies

No Comments

No Known Activations