INDEX

Explanations

mentions of specific persons and their roles or actions

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

ISO

-0.78

Downloadha

-0.76

daq

-0.75

ÙĲ

-0.75

lehem

-0.75

ibaba

-0.75

hari

-0.74

arios

-0.73

ãĥ´ãĤ¡

-0.73

needed

-0.72

POSITIVE LOGITS

own

1.46

 wife

1.46

 father

1.30

 nephew

1.30

 daughter

1.29

 grandson

1.28

 successor

1.26

 granddaughter

1.25

panic

1.25

 eldest

1.25

Activations Density 1.597%

mentions of specific persons and their roles or actions

No Comments

No Known Activations