INDEX

Explanations

mentions of being accused of various actions or behaviors

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

soDeliveryDate

-0.75

DragonMagazine

-0.72

 Mehran

-0.69

MRI

-0.69

soon

-0.64

meric

-0.62

lords

-0.62

arta

-0.62

Sport

-0.61

late

-0.61

POSITIVE LOGITS

 creating

1.57

 putting

1.52

 delivering

1.50

 maintaining

1.49

 completing

1.48

 having

1.46

 keeping

1.45

 preserving

1.45

 providing

1.44

 bringing

1.44

Activations Density 5.545%

mentions of being accused of various actions or behaviors

No Comments

No Known Activations