INDEX

Explanations

references to controversies or dramatic events related to specific individuals or organizations

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pheus

-0.78

eryl

-0.77

illary

-0.77

xual

-0.76

cules

-0.74

 Seym

-0.74

vironment

-0.74

ignty

-0.73

 Marketable

-0.71

ettings

-0.71

POSITIVE LOGITS

idious

1.54

 paced

1.39

eners

1.18

ened

1.16

ener

0.97

ening

0.94

cgi

0.94

nesses

0.90

 asleep

0.88

lane

0.87

Activations Density 0.052%

references to controversies or dramatic events related to specific individuals or organizations

No Comments

No Known Activations