INDEX

Explanations

positive adjectives and descriptors indicating high quality or excellence

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

berdayakan

-0.33

 umum

-0.31

 perbuatan

-0.30

 Ganzen

-0.30

 itself

-0.28

 kantoor

-0.28

 voorkomen

-0.27

may

-0.27

 oorsp

-0.27

time

-0.27

POSITIVE LOGITS

IndentedString

0.90

featureID

0.88

 المعيارى

0.87

Vidite

0.86

 disponibilités

0.79



0.75

<unused41>

0.74

<unused47>

0.74

<unused23>

0.74

<unused8>

0.74

Activations Density 0.014%

positive adjectives and descriptors indicating high quality or excellence

No Comments

No Known Activations