INDEX

Explanations

short identifier codes

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

identifiers ending in "id" (like "pid", "cid", "aid", "tid") in code and technical contexts.

oai_token-act-pair · claude-4-5-sonnet

Triggered by @bfgcn7frf5

The neuron is detecting identifier tokens for numeric IDs (e.g. “id,” “cid,” “pid,” “aid,” “nid,” etc.).

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

Negative Logits

 your

-1.82

 about

-1.77

to

-1.77

尛

-1.71

Naphthalene

-1.65

Toluene

-1.60

out

-1.59

</b>

-1.57

 incorrectly

-1.50

up

-1.46

POSITIVE LOGITS

ématiques

1.54

’，

1.52

Комментарии

1.48

ſten

1.48

จึง

1.43

见了

1.41

ньому

1.40

Ĉ

1.40

?"

1.36

 gourmet

1.35

Activations Density 0.011%

short identifier codes

identifiers ending in "id" (like "pid", "cid", "aid", "tid") in code and technical contexts.

The neuron is detecting identifier tokens for numeric IDs (e.g. “id,” “cid,” “pid,” “aid,” “nid,” etc.).

No Known Activations

short identifier codes

identifiers ending in "id" (like "pid", "cid", "aid", "tid") in code and technical contexts.

The neuron is detecting identifier tokens for numeric IDs (e.g. “id,” “cid,” “pid,” “aid,” “nid,” etc.).

No Known Activations