INDEX

Explanations

Tiny RNA

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

error or warning messages in code that indicate something is wrong or invalid.

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

This neuron identifies error-handling or error-reporting statements (e.g. throwing exceptions, printing or logging error/warning messages).

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

 which

-1.19

of

-1.16

to

-1.02

 including

-0.99

 around

-0.98

 Songtext

-0.98

-0.93

 estratégico

-0.92

 pacto

-0.88

 gerçekten

-0.87

POSITIVE LOGITS

but

1.18

もありました

1.09

却不

1.06

もありますが

1.02

够了

1.02

 arroz

1.02

پ

0.97

எ

0.96

澳大利亚

0.96

Activations Density 0.001%

Tiny RNA

error or warning messages in code that indicate something is wrong or invalid.

This neuron identifies error-handling or error-reporting statements (e.g. throwing exceptions, printing or logging error/warning messages).

No Comments

No Known Activations

Tiny RNA

error or warning messages in code that indicate something is wrong or invalid.

This neuron identifies error-handling or error-reporting statements (e.g. throwing exceptions, printing or logging error/warning messages).

No Comments

No Known Activations