INDEX

Explanations

references to historical events or political and economic topics

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

 ILCS

-0.59

ank

-0.56

 Kara

-0.55

 charism

-0.55

 hemor

-0.53

 Alley

-0.52

essen

-0.52

ATT

-0.52

authent

-0.51

OL

-0.51

POSITIVE LOGITS

 outweigh

1.27

 coincides

1.23

 varies

1.21

 coincided

1.19

 depends

1.18

 outwe

1.16

 exceeds

1.13

 reflects

1.12

 implies

1.12

 constitutes

1.11

Activations Density 2.534%

references to historical events or political and economic topics

No Comments

No Known Activations