INDEX

Explanations

benzo, resection, Pandas, DSS

np_acts-logits-general · gemini-2.5-flash-lite

drug and pharmaceutical names or terms related to software uninstallation.

oai_token-act-pair · claude-3-7-sonnet-20250219 Triggered by @neilrathi

The neuron fires on isolated, content‐carrying “long” tokens – i.e. relatively rare, domain‐specific words (like software commands “uninstall,” chemical prefixes “benzo,” proper‐name fragments “Pand,” etc.) rather than on common function words.

oai_token-act-pair · o4-mini Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_10/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

You

-3.52

</i>

-2.91

’

-2.64

.”

-2.55

Of

-2.50

 Here

-2.44

In

-2.39

 Because

-2.38

↵↵

-2.36

POSITIVE LOGITS

2.86

the

2.56

for

2.53

暌

2.52

2.50

猻

2.47

 faune

2.44

carnation

2.38

 télécommande

2.33

Activations Density 0.009%

benzo, resection, Pandas, DSS

drug and pharmaceutical names or terms related to software uninstallation.

The neuron fires on isolated, content‐carrying “long” tokens – i.e. relatively rare, domain‐specific words (like software commands “uninstall,” chemical prefixes “benzo,” proper‐name fragments “Pand,” etc.) rather than on common function words.

No Comments

No Known Activations

benzo, resection, Pandas, DSS

drug and pharmaceutical names or terms related to software uninstallation.

The neuron fires on isolated, content‐carrying “long” tokens – i.e. relatively rare, domain‐specific words (like software commands “uninstall,” chemical prefixes “benzo,” proper‐name fragments “Pand,” etc.) rather than on common function words.

No Comments

No Known Activations