INDEX

Explanations

headlines or titles of news articles

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 routine

-0.61

Ì¶

-0.59

 otherwise

-0.56

esp

-0.55

YS

-0.54

SUP

-0.52

 Californ

-0.52

ctor

-0.52

 Haram

-0.51

 Americ

-0.50

POSITIVE LOGITS

ing

1.68

ed

1.34

quartered

1.18

ificant

1.08

ificantly

1.03

edly

1.01

edIn

0.95

ership

0.95

quarters

0.94

Activations Density 0.251%

headlines or titles of news articles

No Comments

No Known Activations