INDEX

Explanations

references to memorabilia or the act of memorization

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

IRO

-0.68

OHN

-0.66

 Prometheus

-0.61

roxy

-0.60

 methane

-0.60

Conclusion

-0.60

 Canary

-0.60

 Wenger

-0.60

 Simulator

-0.59

ACS

-0.59

POSITIVE LOGITS

abilia

1.64

ably

1.06

ographed

1.05

ific

1.02

 memor

1.01

istically

1.00

icol

0.95

ized

0.95

ruction

0.93

brance

0.93

Activations Density 0.008%

references to memorabilia or the act of memorization

No Comments

No Known Activations