INDEX

Explanations

quantitative information such as numbers and figures

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

rador

-0.80

ovie

-0.71

aimon

-0.68

axter

-0.66

rait

-0.66

inelli

-0.66

INTON

-0.65

IAL

-0.65

outer

-0.64

 Despair

-0.64

POSITIVE LOGITS

plates

0.99

enance

0.90

metry

0.87

 crunch

0.83

plate

0.81

esses

0.74

 plates

0.72

limits

0.70

pad

0.70

number

0.70

Activations Density 0.447%

quantitative information such as numbers and figures

No Comments

No Known Activations