INDEX

Explanations

numerical values related to margins and errors

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

uni

-0.75

emis

-0.72

arus

-0.70

tics

-0.67

brance

-0.66

ergy

-0.66

oxide

-0.66

JECT

-0.65

endants

-0.63

uras

-0.63

POSITIVE LOGITS

 margin

1.11

 margins

1.03

 separating

0.80

 SHARES

0.75

 favoring

0.72

 landslide

0.69

margin

0.68

ageddon

0.67

irth

0.65

 Advantage

0.64

Activations Density 0.025%

numerical values related to margins and errors

No Comments

No Known Activations