INDEX

Explanations

expressions related to coordinated actions or initiatives

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

 TextInputType

-0.64

XmlAccessorType

-0.56

 Maldonado

-0.51

unbind

-0.49

 windows

-0.47

 Codable

-0.46

Royaume

-0.46

box

-0.44

 Marín

-0.44

windows

-0.44

POSITIVE LOGITS

 Effort

1.34

 Efforts

1.30

effort

1.29

 efforts

1.27

efforts

1.25

Effort

1.25

 effort

1.24

 esfuerzo

0.91

 esforço

0.90

 esfuerzos

0.83

Activations Density 0.012%

expressions related to coordinated actions or initiatives

No Comments

No Known Activations