INDEX

Explanations

phrases indicating various purposes and uses of information or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

PerformLayout

-0.46

Being

-0.43

vesen

-0.43

 suivant

-0.42

io

-0.42

 Realität

-0.41

 lest

-0.41

想到

-0.41

жено

-0.40

ble

-0.40

POSITIVE LOGITS

 purpose

2.54

 purposes

2.49

purposes

2.13

purpose

2.06

 Purpose

2.04

 Purposes

2.01

Purpose

1.94

 PURPOSE

1.85

PURPOSE

1.71

 purpoſe

1.71

Activations Density 0.292%

phrases indicating various purposes and uses of information or actions

No Comments

No Known Activations