INDEX

Explanations

proper names, particularly those of people

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

 Bulgar

-0.68

å£«

-0.67

 Archdemon

-0.66

imeters

-0.66

ashtra

-0.65

 Bulgarian

-0.64

 Vengeance

-0.63

Letter

-0.62

ç¥ŀ

-0.60

acea

-0.60

POSITIVE LOGITS

rentice

0.87

zik

0.81

enhagen

0.75

rees

0.70

rick

0.68

isch

0.68

isson

0.66

aylor

0.64

annon

0.64

elson

0.63

Activations Density 0.021%

proper names, particularly those of people

No Comments

No Known Activations