INDEX

Explanations

as

np_max-act · gemini-2.0-flash

The neuron detects discourse-level connectives and transition words (e.g. “however,” “as,” “well,” “so,” “but”) that signal shifts or links in the text.

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

emodel

-0.07

-time

-0.06

 obligated

-0.06

，却

-0.06

bfd

-0.06

 şimdi

-0.06

Enemies

-0.06

-Time

-0.06

 companyId

-0.06

 Copyright

-0.06

POSITIVE LOGITS

as

0.09

 Autumn

0.07

AL

0.07

iore

0.07

As

0.07

_A

0.07

 WHICH

0.06

 Ар

0.06

-as

0.06

_ONCE

0.06

Activations Density 0.074%

as

The neuron detects discourse-level connectives and transition words (e.g. “however,” “as,” “well,” “so,” “but”) that signal shifts or links in the text.

No Comments

No Known Activations