INDEX

Explanations

instances in text that involve performing an action or deed

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

case

-0.64

hog

-0.63

)=(

-0.62

 Printed

-0.59

fields

-0.59

 Entered

-0.58

Hok

-0.57

 Madagascar

-0.57

lights

-0.57

 Methods

-0.57

POSITIVE LOGITS

omsday

1.11

pez

1.08

ppel

1.06

oms

0.95

herty

0.94

lez

0.92

vet

0.91

ggy

0.90

gging

0.88

xx

0.86

Activations Density 2.397%

instances in text that involve performing an action or deed

No Comments

No Known Activations