INDEX

Explanations

terms and attributes related to data formatting and precision

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

 nobis

-0.36

 voluto

-0.33

 Técnico

-0.32

strerror

-0.31

iż

-0.30

 moule

-0.30

læg

-0.30

 pouces

-0.30

 sportivo

-0.30

Kanpo

-0.29

POSITIVE LOGITS

featureID

0.77

 estekak

0.77

 يتيمه

0.71

 للمعارف

0.70

CppMethod

0.69

MigrationBuilder

0.69

 незавершена

0.68

tagHelperRunner

0.68

 GenerationType

0.66

tonode

0.64

Activations Density 0.001%

terms and attributes related to data formatting and precision

No Comments

No Known Activations