INDEX

Explanations

"c", parentheses, and numbers

np_max-act · gemini-2.0-flash

The neuron selectively activates on SMILES‐format tokens (element symbols, bond symbols like “=”, ring‐closure digits, and bracketed atom specifications), i.e. it detects SMILES chemical‐structure notation.

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

鋼

-0.07

sburg

-0.07

ότητα

-0.07

.nlm

-0.06

規

-0.06

儀

-0.06

 Salvador

-0.06

냐

-0.06

 목록

-0.06

 irregular

-0.06

POSITIVE LOGITS

úp

0.07

нів

0.07

走

0.06

 elektr

0.06

 demi

0.06

TRUE

0.06

ball

0.06

mente

0.06

(Person

0.06

 fall

0.06

Activations Density 0.003%

"c", parentheses, and numbers

The neuron selectively activates on SMILES‐format tokens (element symbols, bond symbols like “=”, ring‐closure digits, and bracketed atom specifications), i.e. it detects SMILES chemical‐structure notation.

No Comments

No Known Activations

"c", parentheses, and numbers

The neuron selectively activates on SMILES‐format tokens (element symbols, bond symbols like “=”, ring‐closure digits, and bracketed atom specifications), i.e. it detects SMILES chemical‐structure notation.

No Comments

No Known Activations