INDEX

Explanations

text related to research findings and studies

oai_token-act-pair · gpt-3.5-turbo

references to ongoing research and its implications

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Whilst

-1.25

 whilst

-1.24

[/

-0.99

Âł

-0.99

 ðŁĻĤ

-0.95

âĢ¦..

-0.95

"âĢ¦

-0.92

↵Âł

-0.91

 ÂłÂł

-0.89

 ðŁĺ

-0.89

POSITIVE LOGITS

 cellphone

0.78

 nonprofits

0.75

 pricey

0.74

dozen

0.72

earchers

0.72

___

0.71

 nonprofit

0.70

 automakers

0.69

 sprawling

0.69

 aides

0.64

Activations Density 1.739%

text related to research findings and studies

references to ongoing research and its implications

No Comments

No Known Activations