INDEX

Explanations

regex substitution and splitting

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

the `sed -i` command or methods/functions that perform in-place modifications (indicated by `!` or similar mutation operators).

oai_token-act-pair · claude-4-5-sonnet Triggered by @bfgcn7frf5

The neuron fires on code tokens that indicate in-place text substitution or regex replacement commands (e.g. sed -i, re.sub(...), map!, character‐class escapes).

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

</em>

-1.67

</h2>

-1.63

-1.61

-1.60

But

-1.54

Despite

-1.49

).

-1.41

Ｇ

-1.40

来年

-1.40

Like

-1.40

POSITIVE LOGITS

емости

1.55

 июня

1.55

 这个

1.54

jenigen

1.52

 декабря

1.52

år

1.52

":

1.48

푦

1.47

 such

1.46

 για

1.46

Activations Density 0.005%

regex substitution and splitting

the `sed -i` command or methods/functions that perform in-place modifications (indicated by `!` or similar mutation operators).

The neuron fires on code tokens that indicate in-place text substitution or regex replacement commands (e.g. sed -i, re.sub(...), map!, character‐class escapes).

No Comments

No Known Activations

regex substitution and splitting

the `sed -i` command or methods/functions that perform in-place modifications (indicated by `!` or similar mutation operators).

The neuron fires on code tokens that indicate in-place text substitution or regex replacement commands (e.g. sed -i, re.sub(...), map!, character‐class escapes).

No Comments

No Known Activations