INDEX

Explanations

will if

np_max-act · gemini-2.0-flash

The neuron activates on future-tense or modal auxiliaries (e.g. “will,” “can,” “should,” “may”) that signal commitments, permissions, or obligations in delivery and service instructions.

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

Negative Logits

ky

-0.08

 организ

-0.07

emb

-0.06

 freed

-0.06

ture

-0.06

 requiring

-0.06

aly

-0.06

TTL

-0.06

StartTime

-0.06

 vind

-0.06

POSITIVE LOGITS

 searchable

0.07

 typeof

0.06

ehir

0.06

_cases

0.06

getic

0.06

ndern

0.06

-click

0.06

 stej

0.06

.."

0.06

NewItem

0.06

Activations Density 0.023%

will if

The neuron activates on future-tense or modal auxiliaries (e.g. “will,” “can,” “should,” “may”) that signal commitments, permissions, or obligations in delivery and service instructions.

No Known Activations

will if

The neuron activates on future-tense or modal auxiliaries (e.g. “will,” “can,” “should,” “may”) that signal commitments, permissions, or obligations in delivery and service instructions.

No Known Activations