INDEX

Explanations

words related to political and economic policies and conditions, specifically related to tax, government benefits, social programs, and public figures

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

canon

-1.13

PCR

-0.81

 prank

-0.79

editor

-0.78

ocular

-0.77

£ı

-0.76

 Cannes

-0.74

SCP

-0.72

Sensor

-0.71

 Canon

-0.71

POSITIVE LOGITS

 incomes

1.52

 subsidies

1.43

 Unemployment

1.39

 subsidized

1.36

 pensions

1.32

 unemployment

1.32

 uninsured

1.32

elfare

1.32

 earners

1.31

 income

1.31

Activations Density 7.050%

words related to political and economic policies and conditions, specifically related to tax, government benefits, social programs, and public figures

No Comments

No Known Activations