INDEX

Explanations

dates mentioned in the format "MM/DD/YYYY" with a high activation value

oai_token-act-pair · gpt-3.5-turbo

date formats

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

 Levant

-0.72

 biography

-0.67

 paed

-0.66

 lamb

-0.65

 consolation

-0.65

 revolution

-0.64

 pressuring

-0.63

 pressured

-0.63

 pressure

-0.63

 McCarthy

-0.62

POSITIVE LOGITS

3.48

2.56

2.33

2.17

2.15

2.11

2.06

1.99

Activations Density 0.012%

dates mentioned in the format "MM/DD/YYYY" with a high activation value

date formats

No Comments

No Known Activations