INDEX

Explanations

words related to specific locations

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

UAL

-0.73

arians

-0.70

ulia

-0.68

 Norn

-0.66

CTR

-0.65

ÑĮ

-0.63

 embodiments

-0.63

URA

-0.61

 antiquity

-0.61

aya

-0.60

POSITIVE LOGITS

pper

1.36

vers

1.35

zzi

1.30

ilers

1.28

ven

1.27

ffee

1.22

pping

1.21

pped

1.21

zzle

1.20

vert

1.19

Activations Density 3.426%

words related to specific locations

No Comments

No Known Activations