INDEX

Explanations

locations or addresses

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

METHOD

-0.74

tools

-0.71

film

-0.69

sticks

-0.67

henko

-0.65

hops

-0.65

cloth

-0.64

aceutical

-0.64

accompan

-0.63

biased

-0.63

POSITIVE LOGITS

rium

1.25

 dusk

0.90

las

0.90

 least

0.89

 Mile

0.87

 Sunrise

0.82

yp

0.81

 sunset

0.80

 weekends

0.80

 Fort

0.79

Activations Density 0.099%

locations or addresses

No Comments

No Known Activations