INDEX

Explanations

key terms or phrases signaling the end of a segment or section

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

Magikarp

-0.65

IGHTS

-0.64

 spots

-0.64

 suscept

-0.63

 rooft

-0.63

 thinner

-0.61

KER

-0.61

BILITY

-0.60

 marble

-0.59

 balcon

-0.59

POSITIVE LOGITS

angered

1.28

owment

1.21

urance

1.10

angering

1.10

orph

1.03

ogenous

1.00

ocrin

0.98

ocrine

0.97

owed

0.91

orse

0.90

Activations Density 0.041%

key terms or phrases signaling the end of a segment or section

No Comments

No Known Activations