INDEX

Explanations

the word "Rat" followed by a number

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "Rat."

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

cu

-0.80

 seaf

-0.79

CVE

-0.77

 eclipse

-0.72

ssl

-0.66

ã

-0.65

 faked

-0.63

fecture

-0.63

 flaw

-0.63

 autop

-0.62

POSITIVE LOGITS

Rat

4.11

Rat

2.60

 Rats

2.13

rat

1.76

rat

1.74

 Ratt

1.52

 Rath

1.34

 rats

1.28

rats

1.24

Cod

1.15

Activations Density 0.023%

the word "Rat" followed by a number

mentions of the word "Rat."

No Comments

No Known Activations