INDEX

Explanations

phrases that indicate relationships or structures within a hierarchy

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

 Wikiseite

-0.46

XmlAccessorType

-0.45

ACHUSET

-0.44

 нет

-0.43

 terasa

-0.42

-------

-0.42

crí

-0.42

 приме

-0.41

riel

-0.40

tsi

-0.40

POSITIVE LOGITS

 Мексичка

0.62

 Савезне

0.52

 AssemblyCompany

0.50

 PyTuple

0.50

WithMany

0.49

setVerticalGroup

0.49

 travaillons

0.45

parsedMessage

0.44



0.44

PhysRevLett

0.43

Activations Density 0.031%

phrases that indicate relationships or structures within a hierarchy

No Comments

No Known Activations