INDEX

Explanations

political terms related to government and 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_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

 ILCS

-0.77

 cubic

-0.71

 harvest

-0.71

brance

-0.69

 upkeep

-0.68

sew

-0.68

 destiny

-0.66

 extinct

-0.66

 tide

-0.66

geries

-0.65

POSITIVE LOGITS

 sarcast

1.45

 remarks

1.42

 stating

1.36

 rhet

1.29

 remark

1.28

 praising

1.26

 criticizing

1.24

 saying

1.22

 comments

1.21

 citing

1.15

Activations Density 8.685%

political terms related to government and organizations

No Comments

No Known Activations