INDEX

Explanations

names of places or people with specific attributes

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

rouse

-0.78

 partName

-0.71

bourg

-0.70

ishop

-0.69

grave

-0.69

unicip

-0.69

uliffe

-0.69

Interested

-0.69

evin

-0.67

gomery

-0.67

POSITIVE LOGITS

 Atlantis

0.95

 Hearts

0.88

 Cups

0.80

 Thieves

0.79

 Mankind

0.76

 Persia

0.76

¥µ

0.76

 Omaha

0.74

 mankind

0.73

Wak

0.72

Activations Density 0.075%

names of places or people with specific attributes

No Comments

No Known Activations