INDEX

Explanations

Predicting primary nucleophile

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

programming code with initialization blocks, particularly focusing on components or interfaces.

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

The neuron never fires—it does not detect any pattern in these inputs.

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

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_22/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

utiliz

-1.39

 וכן

-1.37

趼

-1.36

Fourni

-1.31

 perda

-1.29

 misschien

-1.27

 even

-1.26

出了一

-1.26

its

-1.23

 inaugura

-1.23

POSITIVE LOGITS

1.31

 więc

1.23

1.21

ｎ

1.19

1.18

ou

1.18

お

1.17

プロ

1.16

De

1.16

of

1.16

Activations Density 0.000%

Predicting primary nucleophile

programming code with initialization blocks, particularly focusing on components or interfaces.

The neuron never fires—it does not detect any pattern in these inputs.

No Comments

No Known Activations

Predicting primary nucleophile

programming code with initialization blocks, particularly focusing on components or interfaces.

The neuron never fires—it does not detect any pattern in these inputs.

No Comments

No Known Activations