INDEX

Explanations

proper nouns and historical 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_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

worldly

-0.84

dule

-0.70

ocial

-0.69

 footing

-0.69

natureconservancy

-0.68

ommel

-0.66

selves

-0.66

inka

-0.66

arge

-0.65

lust

-0.65

POSITIVE LOGITS

etc

0.97

0.92

 Cameroon

0.77

 Univ

0.76

 Lanc

0.75

 Born

0.74

 Calif

0.74

 ISBN

0.74

 Cosponsors

0.72

Phi

0.70

Activations Density 0.193%

proper nouns and historical figures

No Comments

No Known Activations