INDEX

Explanations

verbs followed by 'to' indicating an ability or possibility

oai_token-act-pair · gpt-3.5-turbo

phrases indicating capability or potential actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Coff

-0.74

 Respons

-0.70

ELD

-0.68

prem

-0.65

 Rising

-0.64

listed

-0.64

soDeliveryDate

-0.63

mails

-0.62

 Closing

-0.62

backer

-0.60

POSITIVE LOGITS

 withstand

1.29

 cope

1.25

 afford

1.24

 overcome

1.17

 convince

1.15

 muster

1.15

 accommodate

1.15

 achieve

1.13

 penetrate

1.11

 capitalize

1.11

Activations Density 0.105%

verbs followed by 'to' indicating an ability or possibility

phrases indicating capability or potential actions

No Comments

No Known Activations