INDEX

Explanations

phrases related to quality or superiority, often involving comparisons

oai_token-act-pair · gpt-3.5-turbo

references to top rankings or ratings

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

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

 Lauder

-0.59

ATIONS

-0.57

ments

-0.57

Mub

-0.55

ajor

-0.55

 Daylight

-0.54

cci

-0.54

lished

-0.54

Cla

-0.54

 daylight

-0.53

POSITIVE LOGITS

ographical

1.31

ography

1.19

 notch

1.15

ographic

1.14

ographically

1.10

eka

1.03

ology

1.01

ographies

1.00

deck

1.00

most

0.97

Activations Density 0.045%

phrases related to quality or superiority, often involving comparisons

references to top rankings or ratings

No Comments

No Known Activations