INDEX

Explanations

keywords in technical or formal contexts like advertisements or corrections

oai_token-act-pair · gpt-3.5-turbo

instances of numbers and formatting typical of structured content like comments or annotations

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

oun

-1.02

 metic

-0.88

 eleph

-0.88

 analys

-0.86

 successive

-0.86

 disciplines

-0.84

 mism

-0.83

 developmental

-0.81

 exha

-0.81

 acute

-0.81

POSITIVE LOGITS

Advertisements

2.19

UPDATE

1.85

Images

1.80

Comments

1.79

Update

1.76

Featured

1.75

Related

1.75

Sources

1.74

Tags

1.74

SOURCE

1.74

Activations Density 0.402%

keywords in technical or formal contexts like advertisements or corrections

instances of numbers and formatting typical of structured content like comments or annotations

No Comments

No Known Activations