INDEX

Explanations

specific words or phrases that are emphasized or stand out in the text

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

orge

-0.66

Rated

-0.64

bro

-0.59

 Miko

-0.58

AAA

-0.57

gettable

-0.57

etheless

-0.55

essor

-0.54

stood

-0.54

Stage

-0.53

POSITIVE LOGITS

 lest

1.34

 hoping

1.08

 fearing

1.07

avoid

0.95

 precaution

0.94

 hopes

0.89

 because

0.87

 attempt

0.86

 ensuring

0.83

 appease

0.83

Activations Density 0.883%

specific words or phrases that are emphasized or stand out in the text

No Comments

No Known Activations