INDEX

Explanations

Disadvantages and problems

np_max-act · gemini-2.0-flash

The neuron selectively lights up on words that express negation or qualification (e.g. “no,” “relatively,” “little”) in technical-style descriptions.

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

multiply

-0.06

 نرم

-0.06

AutoresizingMaskIntoConstraints

-0.06

yclerview

-0.06

(Component

-0.06

 учит

-0.06

(city

-0.06

 Dickinson

-0.06

 flashlight

-0.06

.street

-0.06

POSITIVE LOGITS

 indica

0.07

 Targets

0.06

 monsters

0.06

 underwent

0.06

 Surgery

0.06

_that

0.06

 pave

0.06

-num

0.06

 oppose

0.06

Activations Density 0.187%

Disadvantages and problems

The neuron selectively lights up on words that express negation or qualification (e.g. “no,” “relatively,” “little”) in technical-style descriptions.

No Comments

No Known Activations