INDEX

Explanations

mentions of the word "multiple."

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

Dialogue

-0.71

instead

-0.69

iquette

-0.68

ESE

-0.66

NER

-0.65

NEY

-0.64

brance

-0.64

nothing

-0.63

atown

-0.63

 Notting

-0.63

POSITIVE LOGITS

 sclerosis

1.79

xes

1.60

 iterations

1.24

 layers

1.16

 occasions

1.13

 versions

1.11

 generations

1.11

 viewpoints

1.11

 instances

1.10

 perspectives

1.10

Activations Density 0.038%

mentions of the word "multiple."

No Comments

No Known Activations