INDEX

Explanations

words related to political news and international relations

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

 Darrell

-0.82

 Bronx

-0.79

 Alto

-0.79

alach

-0.78

emetery

-0.78

 Bucc

-0.76

brow

-0.73

 Died

-0.73

ospel

-0.72

 folders

-0.72

POSITIVE LOGITS

 diplomatic

1.46

 treaties

1.42

 treaty

1.41

 diplomats

1.41

 NATO

1.38

 diplomacy

1.33

 Nato

1.31

 diplomat

1.29

ilateral

1.29

 Treaty

1.28

Activations Density 6.424%

words related to political news and international relations

No Comments

No Known Activations