INDEX

Explanations

phrases related to ease or difficulty in processes or actions

oai_token-act-pair · gpt-3.5-turbo

phrases indicating ability or capacity to perform actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

minus

-0.85

cloth

-0.76

soDeliveryDate

-0.71

 teasp

-0.68

hot

-0.66

 Gone

-0.65

anova

-0.65

 Tears

-0.64

ãĤ¦ãĤ¹

-0.64

 Mayhem

-0.64

POSITIVE LOGITS

 communicate

1.47

 navigate

1.38

 cope

1.29

 interact

1.28

 diagnose

1.25

 integrate

1.25

 locate

1.22

 defend

1.21

 compete

1.20

 negotiate

1.19

Activations Density 0.194%

phrases related to ease or difficulty in processes or actions

phrases indicating ability or capacity to perform actions

No Comments

No Known Activations