INDEX

Explanations

phrases and names with varying structures and formats, potentially related to a specific category or context

oai_token-act-pair · gpt-3.5-turbo

various types of media titles and references, particularly those related to films or shows

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 occas

-0.80

 princ

-0.79

 elbows

-0.70

 prosec

-0.70

 enthusi

-0.70

 trusting

-0.67

udeau

-0.66

 linger

-0.66

aturdays

-0.66

 blogging

-0.65

POSITIVE LOGITS

Contents

1.48

Synopsis

1.19

Overview

1.13

Appearances

1.05

Description

0.97

Background

0.94

edit

0.90

Introduction

0.90

rawdownloadcloneembedreportprint

0.90

Bi

0.88

Activations Density 0.082%

phrases and names with varying structures and formats, potentially related to a specific category or context

various types of media titles and references, particularly those related to films or shows

No Comments

No Known Activations