INDEX

Explanations

Question/Answer exchanges

np_max-act · gemini-2.0-flash

The neuron is sensitive to modal verbs and negation words (e.g. should, must, not, don’t) that signal instructions or prohibitions.

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

查

-0.08

	items

-0.07

WATCH

-0.07

flash

-0.06

_allow

-0.06

("."

-0.06

	LEFT

-0.06

。これ

-0.06

_PRODUCT

-0.06

 docker

-0.06

POSITIVE LOGITS

AGING

0.07

_|

0.07

{|

0.06

 трансп

0.06

 浙江

0.06

dab

0.06

ushi

0.06

uitive

0.06

.reducer

0.06

OMET

0.06

Activations Density 0.109%

Question/Answer exchanges

The neuron is sensitive to modal verbs and negation words (e.g. should, must, not, don’t) that signal instructions or prohibitions.

No Comments

No Known Activations