INDEX

Explanations

articles and guides that are likely to be helpful or useful to a broad audience

oai_token-act-pair · gpt-3.5-turbo

phrases indicating usefulness or assistive guidance

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

Â·Â·

-0.70

Lock

-0.67

 Shutdown

-0.66

 Skydragon

-0.66

 absentee

-0.65

 Wait

-0.65

":""},{"

-0.64

 blackout

-0.64

 barric

-0.63

Shut

-0.62

POSITIVE LOGITS

 enlight

1.18

 helpful

1.15

 insightful

1.14

 informative

1.13

 insights

1.11

 insight

1.07

 useful

1.05

 usefulness

1.03

 learners

1.02

 constructive

1.00

Activations Density 0.644%

articles and guides that are likely to be helpful or useful to a broad audience

phrases indicating usefulness or assistive guidance

No Comments

No Known Activations