INDEX

Explanations

words related to legal or political actions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 tremend

-0.72

 Sabha

-0.69

CLASSIFIED

-0.67

 Skydragon

-0.67

GGGG

-0.63

 Expend

-0.62

ãĥīãĥ©ãĤ´ãĥ³

-0.61

 tert

-0.60

"_

-0.58

ItemThumbnailImage

-0.58

POSITIVE LOGITS

ongyang

0.86

oshenko

0.81

 Nieto

0.78

mint

0.77

seed

0.76

erville

0.73

olate

0.73

inces

0.72

eworthy

0.72

ocene

0.72

Activations Density 0.040%

words related to legal or political actions

No Comments

No Known Activations