INDEX

Explanations

keywords related to legal and political news 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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

estones

-0.81

contact

-0.73

anticipated

-0.71

foreseen

-0.71

aceae

-0.71

 Finder

-0.70

 linem

-0.69

 gauge

-0.66

 peaks

-0.66

joining

-0.65

POSITIVE LOGITS

 hypocrisy

1.59

 arrogance

1.40

 hypocritical

1.39

 irresponsible

1.34

 hypoc

1.30

 stupidity

1.30

 ignorance

1.28

 hypocr

1.27

 incompetence

1.27

 disingen

1.26

Activations Density 4.124%

keywords related to legal and political news events

No Comments

No Known Activations