INDEX

Explanations

neuronal plasticity and connectivity

New Auto-Interp

Top Features by Cosine Similarity

Configuration

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Negative Logits

BOOLE

-0.11

 preload

-0.11

 Trev

-0.10

 Boeh

-0.10

$MESS

-0.10

 human

-0.09

 stap

-0.09

aeper

-0.09

eoq

-0.09

 flats

-0.09

POSITIVE LOGITS

 plastic

0.28

 Plastic

0.24

å¡ĳ

0.15

 plast

0.15

 plastics

0.13

 connections

0.12

 dend

0.12

 pruning

0.12

 spine

0.12

ne

0.11

Activations Density 0.028%

No Known Activations