INDEX

Explanations

words related to signing up or registering for services or events

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

ĸļ

-0.78

»Ĵ

-0.75

 Islands

-0.72

gypt

-0.63

ecause

-0.62

agy

-0.61

 Remastered

-0.61

 Afric

-0.60

 Warfare

-0.58

amily

-0.56

POSITIVE LOGITS

atories

1.15

ificantly

1.08

posted

1.00

alled

1.00

ging

1.00

atory

0.98

atures

0.97

posts

0.97

ature

0.96

ATURES

0.95

Activations Density 4.867%

words related to signing up or registering for services or events

No Comments

No Known Activations