INDEX

Explanations

user questions

This neuron detects tokens that are part of the model/assistant's outputs — especially long, content-rich assistant turns (code blocks or extended explanations).

New Auto-Interp

Configuration

Prompts (Dashboard)

238,145 prompts, 512 tokens each

Dataset (Dashboard)

lmsys + oasst1

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

0.61

’

0.45

 than

0.43

↵

0.42

to

0.41

//

0.39

্স

0.38

had

0.37

ैस

0.36

 mettre

0.35

POSITIVE LOGITS

 వివిధ

0.47

 различ

0.44

 مختلف

0.42

}^{+}=

0.40

एनए

0.40

 ليست

0.38

取决

0.38

图书

0.37

 లేదు

0.37

 interrelated

0.37

Activations Density 0.366%