INDEX

Explanations

political terms and references related to Republican party or conservative viewpoints

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 Palest

-0.73

uras

-0.72

 Atlantis

-0.66

 Britann

-0.65

 Thai

-0.64

ĨĴ

-0.64

ijk

-0.63

ãĤ¨ãĥ«

-0.63

 Bulldogs

-0.63

iful

-0.62

POSITIVE LOGITS

ervative

1.09

ervatives

0.96

 elected

0.92

 majorities

0.87

boro

0.86

 lawmakers

0.83

 Party

0.81

 appoint

0.80

 voted

0.79

endez

0.78

Activations Density 0.029%

political terms and references related to Republican party or conservative viewpoints

No Comments

No Known Activations