INDEX

Explanations

references to projects

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 17-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_17/width_65k/average_l0_125

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.17.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

paravant

-0.85

 uſed

-0.79

 whoſe

-0.78

 Theſe

-0.76

 estekak

-0.75

 vectorstock

-0.75

 myſelf

-0.74

 Мексичка

-0.74

 écl

-0.73

httphttps

-0.73

POSITIVE LOGITS

 projects

1.27

 Projects

1.05

 project

1.04

 Project

1.00

Project

0.97

projects

0.97

Projects

0.96

PROJECT

0.91

 PROJECT

0.91

project

0.88

Activations Density 0.023%

references to projects

No Comments

No Known Activations