INDEX

Explanations

feasible

np_max-act · gemini-2.0-flash

The neuron spotlights evaluative terms and modal constructions (like “can,” “feasible,” “effective,” “useful,” “integrated,” etc.) that assert a method’s or intervention’s feasibility or effectiveness.

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

 Strikes

-0.07

.Option

-0.07

 reviews

-0.06

 DOCUMENT

-0.06

Yaw

-0.06

ifie

-0.06

 confessed

-0.06

),"

-0.06

 результ

-0.06

Coeff

-0.06

POSITIVE LOGITS

(inst

0.06

-return

0.06

.addListener

0.06

 đổ

0.06

 Norse

0.06

-metal

0.06

 nghiên

0.06

_past

0.06

-girl

0.06

 programmed

0.06

Activations Density 0.051%

feasible

The neuron spotlights evaluative terms and modal constructions (like “can,” “feasible,” “effective,” “useful,” “integrated,” etc.) that assert a method’s or intervention’s feasibility or effectiveness.

No Comments

No Known Activations