INDEX

Explanations

quantitative statements or descriptions mentioning a specific number

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

opia

-0.72

BuyableInstoreAndOnline

-0.72

cum

-0.69

acus

-0.67

anything

-0.66

ooter

-0.66

ãģĻ

-0.65

sein

-0.64

ageddon

-0.63

anium

-0.62

POSITIVE LOGITS

 commenters

1.28

 commentators

1.23

 observers

1.05

 scholars

0.98

 analysts

0.96

 reviewers

0.93

 studies

0.91

 countries

0.90

 researchers

0.89

 authors

0.88

Activations Density 0.215%

quantitative statements or descriptions mentioning a specific number

No Comments

No Known Activations