INDEX

Explanations

references to specific dates and locations, possibly related to news events

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

lite

-0.23

talk

-0.22

âķĲ

-0.21

pload

-0.20

âķĲâķĲ

-0.20

osate

-0.20

ãĥĺ

-0.20

sch

-0.19

Ã¨

-0.18

govtrack

-0.18

POSITIVE LOGITS

ggles

0.20

 detrim

0.20

 outwe

0.17

 Mayo

0.16

girlfriend

0.16

 exerc

0.16

idth

0.16

assetsadobe

0.16

 Corona

0.16

cdn

0.16

Activations Density 16.890%

references to specific dates and locations, possibly related to news events

No Comments

No Known Activations