INDEX

Explanations

Code

np_max-act · gemini-2.0-flash

The neuron consistently activates on terms and symbols involved in error handling or exception raising in code (e.g. “Exception,” “raise,” “fail,” “error,” “throw,” “exit”).

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Boots

-0.07

.lua

-0.07

(pixel

-0.07

 premises

-0.06

χή

-0.06

 possibilit

-0.06

ábado

-0.06

 Playstation

-0.06

Supplier

-0.06

POSITIVE LOGITS

 allergy

0.06

 footh

0.06

.XPATH

0.06

 начала

0.06

/Auth

0.06

 publi

0.06

-news

0.06

 tropical

0.06

 citiz

0.06

개를

0.06

Activations Density 0.032%

Code

The neuron consistently activates on terms and symbols involved in error handling or exception raising in code (e.g. “Exception,” “raise,” “fail,” “error,” “throw,” “exit”).

No Comments

No Known Activations

Code

The neuron consistently activates on terms and symbols involved in error handling or exception raising in code (e.g. “Exception,” “raise,” “fail,” “error,” “throw,” “exit”).

No Comments

No Known Activations