INDEX

Explanations

phrases related to actions, particularly those involving usage and creation

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

 QtCore

-0.38

 Newport

-0.38

 pinulongan

-0.38

phors

-0.37

Cea

-0.37

Newport

-0.37

Fac

-0.36

 tbody

-0.35

 Serv

-0.35

 مج

-0.35

POSITIVE LOGITS

modelBuilder

0.48

 bricolaje

0.45

 AssemblyVersion

0.44

WebVitals

0.44

 sagesse

0.44

 bahía

0.43

 ejército

0.43

 keselamatan

0.42

第一时间

0.42

 okam

0.42

Activations Density 0.063%

phrases related to actions, particularly those involving usage and creation

No Comments

No Known Activations