INDEX

Explanations

validation

np_max-act · gemini-2.0-flash

The neuron activates on terms referring to dataset splits (e.g., “validation,” “test”).

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

 Scaffold

-0.07

FAILURE

-0.07

 HuffPost

-0.07

_ratings

-0.07

(Field

-0.07

Ear

-0.07

(reader

-0.06

 roce

-0.06

줄

-0.06

 FileWriter

-0.06

POSITIVE LOGITS

 vlády

0.07

voy

0.06

؟↵

0.06

 sensed

0.06

іб

0.06

ับผ

0.06

 analy

0.05

lichen

0.05

 containment

0.05

statistics

0.05

Activations Density 0.002%

validation

The neuron activates on terms referring to dataset splits (e.g., “validation,” “test”).

No Comments

No Known Activations