INDEX

Explanations

titles or headings in articles

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

chrom

-0.63

 crou

-0.61

aples

-0.60

 bases

-0.60

DERR

-0.60

EEK

-0.60

ometers

-0.60

bos

-0.59

 base

-0.58

 tripod

-0.58

POSITIVE LOGITS

"#

1.00

 Dreams

0.83

 Beware

0.81

"<

0.80

selves

0.79

 titled

0.76

 Dear

0.74

 Fahrenheit

0.73

 Nemesis

0.73

 Endless

0.72

Activations Density 0.041%

titles or headings in articles

No Comments

No Known Activations