INDEX

Explanations

phrases related to legal actions or criminal activities

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

Birth

-0.64

Ink

-0.62

grave

-0.60

alam

-0.60

repre

-0.60

olia

-0.58

 Emer

-0.58

bourg

-0.58

 Wond

-0.57

ortium

-0.57

POSITIVE LOGITS

swick

1.05

aways

0.99

gs

0.91

dy

0.88

escape

0.88

ners

0.85

ways

0.84

Disney

0.83

af

0.81

nin

0.80

Activations Density 4.026%

phrases related to legal actions or criminal activities

No Comments

No Known Activations