INDEX

Explanations

neuron findings and results

findings and conclusions

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

SearchParams

0.37

кість

0.36

列

0.36

 envol

0.35

朼

0.35

搜索引擎

0.34

搜索

0.33

 perfekt

0.32

 goog

0.32

 शेफाली

0.32

POSITIVE LOGITS

 findings

3.39

findings

3.23

Findings

3.13

 Findings

3.05

 conclusions

2.27

results

1.97

 results

1.94

 conclusiones

1.90

 результаты

1.84

 discoveries

1.82

Activations Density 0.085%