INDEX

Explanations

the word "wholly" with different levels of activation where 10 indicates the strongest match

oai_token-act-pair · gpt-3.5-turbo

expressions of totality or completeness

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

anwhile

-0.80

ebus

-0.77

 Frie

-0.71

 Commissioners

-0.70

imeters

-0.70

vals

-0.69

aminer

-0.69

 flyers

-0.69

 Cards

-0.68

 pods

-0.67

POSITIVE LOGITS

BuyableInstoreAndOnline

1.04

Buyable

0.91

 reliant

0.89

 liqu

0.75

 behold

0.74

entimes

0.74

alysed

0.73

 uncon

0.73

unt

0.73

 wholly

0.71

Activations Density 0.004%

the word "wholly" with different levels of activation where 10 indicates the strongest match

expressions of totality or completeness

No Comments

No Known Activations