INDEX

Explanations

arguments and discussions around a specific topic or issue with some focus on criticism and disagreement

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

Pwr

-0.55

fortunately

-0.53

 irony

-0.48

 spokesman

-0.48

 consolation

-0.47

 partName

-0.47

 fortunately

-0.46

 explains

-0.46

 nonetheless

-0.46

cffffcc

-0.45

POSITIVE LOGITS

thood

0.58

 virginity

0.51

 unpopular

0.51

versive

0.51

equality

0.49

 unfavorable

0.49

 unfairly

0.48

 supposedly

0.48

 purportedly

0.48

 homosexuality

0.47

Activations Density 11.048%

arguments and discussions around a specific topic or issue with some focus on criticism and disagreement

No Comments

No Known Activations