INDEX

Explanations

specific geographical locations and significant historical events

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

isson

-0.58

ãĤ³

-0.56

milo

-0.54

ãģ¾

-0.53

 preferred

-0.52

uay

-0.52

skirts

-0.52

instead

-0.51

ById

-0.51

çīĪ

-0.49

POSITIVE LOGITS

 history

1.41

history

1.30

 ever

1.27

 EVER

1.19

ever

1.13

 since

0.99

 History

0.98

era

0.93

 imaginable

0.91

since

0.89

Activations Density 0.228%

specific geographical locations and significant historical events

No Comments

No Known Activations