INDEX

Explanations

phrases related to attention, coverage, and reception in various contexts

oai_token-act-pair · gpt-3.5-turbo

topics that have garnered significant attention or coverage

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

artifacts

-0.77

aco

-0.75

ILE

-0.65

uggest

-0.63

avid

-0.61

nel

-0.60

ZI

-0.60

reements

-0.60

 matters

-0.60

 Administ

-0.59

POSITIVE LOGITS

 traction

0.97

 scrutiny

0.90

 foothold

0.89

 resurgence

0.88

 flourish

0.86

 shine

0.84

 renaissance

0.80

 attention

0.79

 footing

0.78

 prominence

0.77

Activations Density 0.542%

phrases related to attention, coverage, and reception in various contexts

topics that have garnered significant attention or coverage

No Comments

No Known Activations