INDEX

Explanations

names or surnames of individuals, especially when they are referred to in a professional or newsworthy context

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

COP

-0.80

 DIRECT

-0.79

 charg

-0.76

 toget

-0.76

 multit

-0.76

 INTER

-0.75

 Competitive

-0.75

 seiz

-0.74

 livest

-0.72

 puzz

-0.71

POSITIVE LOGITS

aga

1.37

ma

1.34

na

1.30

sa

1.29

ya

1.24

va

1.21

da

1.20

ava

1.19

ana

1.14

ba

1.12

Activations Density 0.152%

names or surnames of individuals, especially when they are referred to in a professional or newsworthy context

No Comments

No Known Activations