INDEX

Explanations

proper nouns related to current events or people

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

 Yamaha

-0.68

ĪĴ

-0.66

uyomi

-0.66

 GOODMAN

-0.65

 Tinder

-0.64

 Chero

-0.64

 track

-0.63

ãĥĹ

-0.60

ItemTracker

-0.60

 timeout

-0.59

POSITIVE LOGITS

overed

1.27

rossover

1.27

ursor

1.25

oding

1.22

rouch

1.20

URRENT

1.19

ologne

1.18

auldron

1.18

ategor

1.18

ursive

1.17

Activations Density 0.867%

proper nouns related to current events or people

No Comments

No Known Activations