INDEX

Explanations

references to specific media or entertainment titles that start with "My"

oai_token-act-pair · gpt-3.5-turbo

references to specific titles or names in the "My" series

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

 bott

-0.78

ooz

-0.71

 yourselves

-0.69

itud

-0.67

 ramps

-0.64

 forcefully

-0.64

 sounded

-0.64

inelli

-0.62

igham

-0.62

 characterized

-0.59

POSITIVE LOGITS

stery

1.63

riad

1.59

anmar

1.39

stic

1.34

ths

1.29

croft

1.11

chal

1.10

ster

1.08

ocard

1.04

Space

1.00

Activations Density 0.055%

references to specific media or entertainment titles that start with "My"

references to specific titles or names in the "My" series

No Comments

No Known Activations