INDEX

Explanations

phrases related to different ways of doing things or expressing ideas

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

UAL

-0.70

Klu

-0.70

isite

-0.70

hunter

-0.68

 Bots

-0.66

çīĪ

-0.63

sites

-0.63

 Ster

-0.62

 Boat

-0.62

 Extras

-0.60

POSITIVE LOGITS

 communicating

1.11

 dealing

1.05

 coping

1.00

 avoiding

0.97

 ensuring

0.97

 expressing

0.96

 combating

0.96

 achieving

0.95

 interpreting

0.95

 preserving

0.94

Activations Density 0.058%

phrases related to different ways of doing things or expressing ideas

No Comments

No Known Activations