INDEX

Explanations

your actual

np_max-act · gemini-2.0-flash

The neuron activates on instructional phrases telling the reader to replace placeholders (e.g. “your actual directories”), so it detects placeholder-replacement 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

VIP

-0.07

effect

-0.07

#@

-0.07

Convention

-0.06

優

-0.06

EX

-0.06

問題

-0.06

 unserialize

-0.06

 Weekly

-0.06

 CircularProgress

-0.06

POSITIVE LOGITS

 Bain

0.06

Nin

0.06

 Napoli

0.06

opc

0.06

inance

0.06

 scalp

0.06

hur

0.06

 dokument

0.06

issenschaft

0.06

.online

0.06

Activations Density 0.015%

your actual

The neuron activates on instructional phrases telling the reader to replace placeholders (e.g. “your actual directories”), so it detects placeholder-replacement instructions.

No Comments

No Known Activations