INDEX

Explanations

phrases related to worthiness and value

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

\{\\

-0.47

 jabatan

-0.46

ArrowToggle

-0.46

Nix

-0.45

enumi

-0.45

はじめに

-0.44

ʌ

-0.44

ImageContext

-0.43

Parcelize

-0.43

 gangen

-0.43

POSITIVE LOGITS

 Worthy

1.26

 worthy

1.20

worthy

1.07

 достой

0.85

 deserving

0.81

 digna

0.80

 digne

0.80

 unworthy

0.78

 digno

0.76

Worth

0.69

Activations Density 0.005%

phrases related to worthiness and value

No Comments

No Known Activations