INDEX

Explanations

numerical values

oai_token-act-pair · gpt-3.5-turbo

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

 belong

-0.80

 belongs

-0.76

 separat

-0.68

 Goodbye

-0.66

 Badge

-0.65

izable

-0.65

 Stability

-0.65

å£«

-0.65

 Carry

-0.64

 Tags

-0.64

POSITIVE LOGITS

 able

1.34

 unable

1.26

 surprised

1.20

 hesitant

1.19

 reluctant

1.18

 alerted

1.15

 unaware

1.14

 aware

1.13

 astonished

1.11

 pleased

1.11

Activations Density 0.271%

numerical values

No Comments

No Known Activations