INDEX

Explanations

the word "upside" or variations of it

oai_token-act-pair · gpt-3.5-turbo

phrases related to positive and negative outcomes, particularly emphasizing "upside" and "downside."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

eph

-0.75

charge

-0.74

elle

-0.71

charges

-0.68

 Conduct

-0.68

Est

-0.67

 Chapters

-0.67

Count

-0.65

 Count

-0.65

 Lists

-0.65

POSITIVE LOGITS

 upside

4.26

 downside

1.86

 sideways

1.40

 ceiling

1.25

 ceilings

1.19

 inverted

1.18

 flip

1.02

 backwards

1.01

 flipped

1.01

 unpredict

1.00

Activations Density 0.014%

the word "upside" or variations of it

phrases related to positive and negative outcomes, particularly emphasizing "upside" and "downside."

No Comments

No Known Activations