INDEX

Explanations

phrases related to actions and reactions, including shouting, yelling, running, and expressing commands or statements

oai_token-act-pair · gpt-3.5-turbo

audience reactions and dynamics within narrative contexts

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

DonaldTrump

-0.64

 Canaveral

-0.61

Specifically

-0.59

luster

-0.59

 divest

-0.57

BuyableInstoreAndOnline

-0.56

.''.

-0.55

 domestically

-0.55

quartered

-0.55

uments

-0.54

POSITIVE LOGITS

emonium

0.69

girl

0.67

escape

0.62

 Zombie

0.62

agar

0.58

xual

0.56

erker

0.56

boy

0.56

 Technique

0.55

olor

0.55

Activations Density 1.744%

phrases related to actions and reactions, including shouting, yelling, running, and expressing commands or statements

audience reactions and dynamics within narrative contexts

No Comments

No Known Activations