Stable Diffusion 3 License Revamped Amid Blowback, Promising Better ModelStability AI released a more permissive license for Stable Diffusion 3 amid recent controversy, but does it go far enough?

Stable Diffusion 3 License Revamped Amid Blowback, Promising Better Model

Stability AI released a more permissive license for Stable Diffusion 3 amid recent controversy, but does it go far enough?

Stable Diffusion 3 License Revamped Amid Blowback, Promising Better ModelStability AI released a more permissive license for Stable Diffusion 3 amid recent controversy, but does it go far enough?

In brief

• Stable Diffusion users pushed back last month after Stability AI launched restrictive new licensing terms.
• The firm announced late Friday that it has relaxed its conditions, to mixed reactions from users.
• Users still cannot create a new foundational model by training it on SD3-generated work.

Stability AI unveiled a revamped Community License for its Stable Diffusion 3 (SD3) model, aiming to quell the firestorm of controversy that erupted following the initial release. The company's move comes on the heels of a ban by CivitAI, a major community hub, which had barred all SD3-related content due to licensing concerns.

“We acknowledge that our latest release, SD3 Medium, didn’t meet our community’s high expectations,” Stability said in an announcement late Friday. “We heard you and have made improvements to address your concerns and to continue to support the open-source community."

Under the new terms, Stability AI grants free use of SD3 for research, non-commercial, and limited commercial purposes. The license also allows individuals and businesses with annual revenues under $1 million to use the model without charge. Those exceeding this threshold must obtain a paid enterprise license.

In an interview with Stability, the company confirmed that it is OK to create custom SD3 models and improve on top of the base SD3. However, it’s forbidden to develop a new foundational model using images generated with SD3 as part of its training dataset—that is, training a Stable Diffusion competitor using material from the original model.

“Derivative products include any output derived from Stability AI's Foundational models, such as fine-tuned models or other creative outputs,” a spokesperson from Stability AI told Decrypt. “Examples of derivative works include SD3 fine-tunes, LoRA fine-tunes, adapters etc. and these can also be trained with SD3 output images.”

The license also says that “you are the owner of derivative works you create, subject to Stability AI’s ownership of the Stability AI materials and any derivative works made by or for Stability AI.” In other words, as long as those boundaries are respected, fine-tuning and profiting from it should not be against the terms and conditions.

“To safeguard our IP, it is not permitted to train new foundational AI models using SD3 outputs as training data, and all activity must adhere to our acceptable use policies,” the company spokesperson told Decrypt.

New Fiber Optics Tech Smashes Data Rate Record Expanded bandwidth yields a transmission rate of 402 terabits per second

New Fiber Optics Tech Smashes Data Rate Record Expanded bandwidth yields a transmission rate of 402 terabits per secondNew Fiber Optics Tech Smashes Data Rate Record 

Expanded bandwidth yields a transmission rate of 402 terabits per second 

An international team of researchers have smashed the world record for fiber optic communications through commercial-grade fiber. By broadening fiber’s communication bandwidth, the team has produced data rates four times as fast as existing commercial systems—and 33 percent better than the previous world record.

The researchers’ success derives in part from their innovative use of optical amplifiers to boost signals across communications bands that conventional fiber optics technology today less-frequently uses. “It’s just more spectrum, more or less,” says Ben Puttnam, chief senior researcher at the National Institute of Information and Communications Technology (NICT) in Koganei, Japan.

Puttnam says the researchers have built their communications hardware stack from optical amplifiers and other equipment developed, in part, by Nokia Bell Labs and the Hong Kong-based company Amonics. The assembled tech comprises six separate optical amplifiers that can squeeze optical signals through C-band wavelengths—the standard, workhorse communications band today—plus the less-popular U-, L-, S-, E-, and O-bands. (E- and O- bands are in the near-infrared; while S-band, C-band, L-, and O-bands are in what’s called short-wavelength infrared.)

All together, the combination of O, E, S, C, L, and U bands enables the new technology to push a staggering 402 terabits per second (Tbps) through the kinds of fiber optic cables that are already in the ground and underneath the oceans. Which is impressive when compared to the competition. 

“The world’s best commercial systems are 100 terabits per second,” Puttnam says. “So we’re already doing about four times better.” Then, earlier this year, a team of researchers at Aston University in the Birmingham, England boasted what at the time was a record-setting 301 Tbps using much the same tech as the joint Japanese-British work—plus sharing a number of researchers between the two groups.

Puttnam adds that if one wanted to push everything to its utmost limits, more bandwidth still could be squeezed out of existing cables.

“If you really push everything, if you filled in all the gaps, and you had every channel the highest quality you can arrange, then probably 600 [Tbps] is the absolute limit,” Puttnam says.

Getting to 402 Tbps—or 600

The “C” in C-band stands for “conventional”—and C-band is the conventional communications band in fiber optics in part because signals in this region of spectrum experience low signal loss from the fiber. “Fiber loss is higher as you move away from C-band in both directions,” Puttnam says.

For instance, in much of the E-band and O-band, the same phenomenon that causes the sky to be blue and sunsets to be pink and red—Rayleigh scattering—makes the fiber less transparent for these regions of the infrared spectrum. And just as a foggy night sometimes requires fog lights, strong amplification of signals can be all the more significant when the fiber is less transparent than it is for the comparatively high-transparency C-band.

“The world’s best commercial systems are 100 terabits per second. We’re already doing about four times better.”—BEN PUTTNAM, NICT

Previous efforts to increase fiber optic bandwidths have often relied on what are called doped-fiber amplifiers (DFA)—in which an optical signal enters a modified stretch of fiber that’s been doped with a rare-earth ion like erbium. When a pump laser is shined into the fiber, the dopant elements in the fiber are pushed into higher energy states. That allows photons from the optical signal passing through the fiber to trigger a stimulated emission from the dopant elements. The result is a stronger (i.e. amplified) signal exiting the DFA fiber stretch than the one that entered it.

Bismuth is the dopant of choice for the E band. But even bismuth DFAs are still just the least-bad option for boosting E-band signals.They can sometimes be inefficient, with higher noise rates, and more limited bandwidths.

So Puttnam says the team developed a DFA that is co-doped with both bismuth and germanium. Then they added to the mix a kind of filter developed by Nokia that optimizes the amplifier performance and improves the signal quality.

“So you can control the spectrum to compensate for the variations of the amplifier,” Puttnam says.

Ultimately, he says, the amplifier can still do its job without overwhelming the original signal.New Fiber Optics Tech Smashes Data Rate Record Expanded bandwidth yields a transmission rate of 402 terabits per second

Pongo - Mongo but on Postgres and with strong consistency benefits

Pongo - Mongo but on Postgres and with strong consistency benefits 

Flexibility or Consistency? Why not have both? Wouldn’t it be great to have MongoDB flexible schema and PostgreSQL consistency?

MongoDB is a decent database, but it gives headaches with its eventual consistency handling. I wrote about it a few times in past:

• Dealing with Eventual Consistency and Idempotency in MongoDB projections
• Long-polling, how to make our async API synchronous
• How to test event-driven projections

Don’t get me wrong, eventual consistency is fine. We need to learn to live with that, still… Undeniably, having strong consistency guarantees, transactions, read your own writing is great.

On Friday, I decided to spend my working day on the small proof of concept that I called Pongo.

What’s Pongo?

It’s a MongoDB-compliant wrapper on top of Postgres.

You can setup it like that:

import { pongoClient } from "@event-driven-io/pongo";

const connectionString =
  "postgresql://dbuser:secretpassword@database.server.com:5432/yourdb";

const pongoClient = pongoClient(postgresConnectionString);
const pongoDb = pongoClient.db();

const users = pongoDb.collection  

It will start internally with a PostgreSQL connection pool connected to your selected database.
Having that, you can then perform operations like:
const anita = { name: "Anita", age: 25 };

// Inserting
await pongoCollection.insertOne(roger);
await pongoCollection.insertOne(cruella);

const { insertedId } = await pongoCollection.insertOne(alice);
const anitaId = insertedId;

// Finding by Id
const anitaFromDb = await pongoCollection.findOne({ _id: anitaId });

// Updating
await users.updateOne({ _id: anitaId }, { $set: { age: 31 } });

// Deleting
await pongoCollection.deleteOne({ _id: cruella._id });

// Finding by Id
const anitaFromDb = await pongoCollection.findOne({ _id: anitaId });

// Finding more
const users = await pongoCollection.find({ age: { $lt: 40 } });
Internally, it’ll set up the collection as the PostgreSQL table with the key-value structure:

CREATE TABLE IF NOT EXISTS "YourCollectionName" (
    _id UUID PRIMARY KEY, 
    data JSONB
)
Essentially, it treats PostgreSQL as a key/value database. Sounds familiar? Yet, it’s a similar concept to Marten or, more correctly, to AWS DocumentDB (see here or there, they seem to be using Mongo syntactic sugar on top of AuroraDB with Postgres).
I explained in general strategy for migrating relational data to document-based that contrary to common belief, document data is structured but less rigidly, as in the relational approach. JSON has structure, but it is not enforced for each document. We can easily extend the schema for our documents, even for specific ones, by adding new fields. We should also not fail if the field we expect to exist, but doesn’t.
Handling semi-structured data in a relational database can be tricky, but PostgreSQL’s JSONB data type offers a practical solution. Unlike the plain text storage of the traditional JSON type, JSONB stores JSON data in a binary format. This simple change brings significant advantages in terms of performance and storage efficiency.
The binary format of JSONB means that data is pre-parsed, allowing faster read and write operations than text-based JSON. You don’t have to re-parse the data every time you query it, which saves processing time and improves overall performance. Additionally, JSONB supports advanced indexing options like GIN and GiST indexes, making searches within JSONB documents much quicker and more efficient.
Moreover, JSONB retains the flexibility of storing semi-structured data while allowing you to use PostgreSQL’s robust querying capabilities. You can perform complex queries, joins, and transactions with JSONB data, just as you can with regular relational data.