INDEX

Explanations

phrases or words related to physical actions in a confrontational or intense context

oai_token-act-pair · gpt-3.5-turbo

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

soDeliveryDate

-0.73

actionDate

-0.69

 Recent

-0.66

category

-0.64

 Bloomberg

-0.64

 endorsements

-0.63

isSpecialOrderable

-0.60

ishable

-0.60

 Kardash

-0.59

 recent

-0.59

POSITIVE LOGITS

 hesitated

1.29

 proceeded

1.22

 panicked

1.20

 replied

1.18

 responded

1.18

 screamed

1.16

 countered

1.15

 grabbed

1.14

 laughed

1.12

 awoke

1.10

Activations Density 0.410%

phrases or words related to physical actions in a confrontational or intense context

No Comments

No Known Activations