INDEX

Explanations

terms related to introductory sections or written pieces

oai_token-act-pair · gpt-3.5-turbo

instances of the word "introduction."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

esthetic

-0.72

gem

-0.70

robe

-0.70

mic

-0.70

yss

-0.70

rone

-0.68

roph

-0.68

rage

-0.67

hap

-0.67

ramid

-0.64

POSITIVE LOGITS

 Takeru

0.78

ablishment

0.77

ounter

0.76

 introduction

0.75

urated

0.74

introdu

0.73

prise

0.73

APR

0.72

lishes

0.72

 introducing

0.70

Activations Density 0.012%

terms related to introductory sections or written pieces

instances of the word "introduction."

No Comments

No Known Activations