INDEX

Explanations

memory allocation and management

The neuron strongly activates on technical descriptions involving memory blocks and cache coherence (e.g. references to “memory block,” “cache,” “coherence,” directory updates, L2/L3 victim entries, etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 Accepting

0.62

ભ્ય

0.57

simpleType

0.57

 parseInt

0.55

讃

0.55

람

0.53

 boolean

0.53

 accepting

0.52

oldown

0.52

 Bhajan

0.51

POSITIVE LOGITS

 address

1.67

address

1.64

地址

1.62

 memory

1.60

Address

1.55

 addresses

1.53

memory

1.51

内存

1.51

 adress

1.51

 Address

1.50

Activations Density 0.066%