INDEX

Explanations

titles or headings containing numbered lists or bullet points

oai_token-act-pair · gpt-3.5-turbo

section headings or labels associated with chapters and lessons

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

anmar

-0.67

gnu

-0.66

ende

-0.64

sap

-0.64

 murd

-0.62

intent

-0.61

rue

-0.61

antically

-0.60

etimes

-0.60

 opin

-0.59

POSITIVE LOGITS

 Introduction

0.71

æĸ¹

0.70

 Dungeons

0.64

SPONSORED

0.64

 onwards

0.63

 Raid

0.61

ayn

0.60

 Anniversary

0.59

 Sched

0.59

 introduction

0.59

Activations Density 0.251%

titles or headings containing numbered lists or bullet points

section headings or labels associated with chapters and lessons

No Comments

No Known Activations