INDEX

Explanations

sentences where someone is speaking or quoting someone else

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

Pg

-0.85

Newsletter

-0.83

Cf

-0.76

osc

-0.72

ItemThumbnailImage

-0.72

ogether

-0.70

_-

-0.68

ItemImage

-0.68

conservancy

-0.67

wk

-0.66

POSITIVE LOGITS

 Donald

0.86

 embattled

0.84

 homosexuality

0.76

 Antarctica

0.70

 rapper

0.69

 billionaire

0.69

 Adolf

0.68

 Tyrann

0.67

 constitu

0.67

 Tehran

0.66

Activations Density 0.249%

sentences where someone is speaking or quoting someone else

No Comments

No Known Activations