INDEX

Explanations

quotations marked with double quotation marks

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

 accomp

-0.98

 carrier

-0.81

 agric

-0.77

 viral

-0.77

 adjud

-0.75

 dubbed

-0.75

 liner

-0.75

 cleanup

-0.74

 brim

-0.74

 flared

-0.74

POSITIVE LOGITS

We

1.59

Nobody

1.59

There

1.58

Everybody

1.58

Whoever

1.57

Absolutely

1.57

If

1.55

It

1.53

Everyone

1.51

Activations Density 0.112%

quotations marked with double quotation marks

No Comments

No Known Activations