Yi Cui

onekq

AI & ML interests

Benchmark, Code Generation Model

Recent Activity

Organizations

MLX Community's profile picture ONEKQ AI's profile picture

onekq's activity

posted an update about 1 hour ago
view post
Post
69
I added OneSQL 3B to the model family, and its GGUF/AWQ/MLX quantizations. This model can fit into more places, and comfortably run on Apple M1 devices with twice the throughput (half the generation time) of its 7B sibling.

onekq-ai/onesql-v01-qwen-67d8e3eb1611c5532bb90c5f
  • 1 reply
·
reacted to clem's post with 🔥❤️ about 16 hours ago
view post
Post
3421
Before 2020, most of the AI field was open and collaborative. For me, that was the key factor that accelerated scientific progress and made the impossible possible—just look at the “T” in ChatGPT, which comes from the Transformer architecture openly shared by Google.

Then came the myth that AI was too dangerous to share, and companies started optimizing for short-term revenue. That led many major AI labs and researchers to stop sharing and collaborating.

With OAI and sama now saying they're willing to share open weights again, we have a real chance to return to a golden age of AI progress and democratization—powered by openness and collaboration, in the US and around the world.

This is incredibly exciting. Let’s go, open science and open-source AI!
·
reacted to John6666's post with 👍 1 day ago
posted an update 1 day ago
view post
Post
1294
Adding MLX version of OneSQL 7B for MacBook (Apple Silicon) users
onekq-ai/OneSQL-v0.1-Qwen-7B-MLX-4bit

This model has the best accuracy among all quantized versions (AWX, GGUF etc.), which I am very happy about.

I tested this model on my MacBook Air with M1 processor and 8GB of RAM, which is the lower bound of Apple Silicon, also the earliest and still the most popular. On average it took 16 seconds to generate a SQL query, and one minute in the worst case. If you own a newer MacBook with M2 or M3, the speed should be considerably faster.

I hope the MLX team will improve inference speed by software tricks (definitely doable) in the future. Meanwhile, if you find the current inference speed acceptable, you are more than welcome to enjoy this model. 🤗
posted an update 3 days ago
posted an update 4 days ago
view post
Post
1705
I really appreciate the comments and thoughts on my last post. It feels great reading them. Coming to think of it, HF is the best community to collect and define us. You do need a community to keep yourself going. 🤗

Now let me turn to the positive side, that is the model creators are incredibly empowered. Software folks must know the term "breaking changes". I spent a good portion of my lifetime building services. We had lots of great ideas. But at least half of the time they were rolled back after launch because they broke too-big-to-fail customers. Then the ideas died in vain.

Now with models, the value prop becomes "this model is smarter than the last one, but you will have to rebuild everything in order to use it" (shouldn't happen with OneSQL because the contract is simple to begin with). This would never fly in the software world.

Phenomena like this almost never happened. The closest precedence I could relate is "Intel Inside".
  • 1 reply
·
reacted to JLouisBiz's post with 👍 4 days ago
view post
Post
1487
https://www.youtube.com/watch?v=Jn7zMmPClIc

# Automating Tasks with Emacs: Speech-to-Command Magic Explained

Welcome to our latest tutorial where we explore the powerful capabilities of Emacs, a renowned text editor, as we dive into automating tasks using speech-to-command functionality. This guide will walk you through setting up and using Emacs Lisp functions to transcribe speech into actionable commands, seamlessly integrating with your database to execute tasks.

# What You'll Learn:

- How to set up speech recognition in Emacs.
- Transcribing speech into commands using Emacs Lisp.
- Executing commands from a database with real-time feedback.
- Enhancing productivity by automating repetitive tasks with your voice.

## Step-by-Step Guide:

Introduction to Emacs Lisp: Understand the basics of Emacs Lisp and how it can be used for automation.

Setting Up Speech Recognition: Learn how to configure Emacs to recognize and transcribe your speech.

Transcribing Speech to Commands: Discover how to convert spoken words into executable Emacs commands.

Database Integration: See how commands are matched with database entries to perform specific tasks.

Real-Time Feedback: Experience how Emacs provides real-time feedback by speaking the results of executed commands.

## Why Use Emacs for Automation?

Emacs is not just a text editor; it's a versatile tool that can be customized to fit your workflow. By leveraging its scripting capabilities with Emacs Lisp, you can create a personalized automation environment that responds to your voice, making your work more efficient and intuitive.

## Conclusion:

By the end of this video, you'll have a functional setup that allows you to control Emacs with your voice, opening up new possibilities for productivity. Whether you're a seasoned Emacs user or new to the platform, this tutorial will provide valuable insights into the power of automation.
posted an update 6 days ago
view post
Post
2216
Open source models are immutable, this is a big pain.

When you open source a piece of software, users leave their feedbacks via issues or PRs. You can merge their feedbacks in semi real time, this creates a positive cycle. Then you have a community.

LLMs don't have these nice micro steps. There are no hot fixes. Even a minor version bump is an endeavor. I'm quite confident my model is being used by teams somewhere. But until next launch, it's awfully quiet.

I don't know the solution. Just a regular lament before weekend. 🤗

·
reacted to JLouisBiz's post with 🔥 6 days ago
view post
Post
1614
In this exciting demonstration, we explore how you can enhance your productivity with cutting-edge features right at your fingertips. Experience seamless speech recognition and automatic text correction on GNU/Linux systems using just a couple of mouse clicks!

https://www.youtube.com/watch?v=51jEUtjrARo

What You'll Discover:

Speech Recognition: Activate by pressing *Mouse Button 9*. Say goodbye to typing fatigue as our system effortlessly converts spoken words into digital text.

Automatic LLM Text Correction: Press Mouse Button 8 for instant, intelligent corrections. Our advanced language model ensures your writing is polished and precise.

Why You Should Watch:

✅ Boost Your Efficiency
🔍 Simplify Complex Tasks
💡 Enhance Writing Quality

Whether you're a developer looking to streamline coding or someone who spends hours typing reports, this demonstration will show how these features can transform the way you work.

Don't miss out on discovering an innovative approach that integrates speech recognition and text correction into your daily routine with ease!

💬 Drop a comment below if you have questions or want to share how these features could benefit your workflow.
posted an update 8 days ago
posted an update 9 days ago
posted an update 10 days ago
posted an update 13 days ago
view post
Post
3741
Folks, let's get ready.🥳 We will be busy soon. 😅🤗https://github.com/huggingface/transformers/pull/36878
replied to their post 13 days ago
replied to their post 14 days ago
posted an update 14 days ago
view post
Post
1568
I like to benchmark 💵o1-pro💵 but it is way too expensive for me 🤦‍♂️
·
replied to their post 14 days ago
view reply

You can infer on CPU, but it will be very very slow 😕

posted an update 15 days ago
view post
Post
453
The majority of OneSQL downloads went to the lowest end (7B-GGUF). I didn't expect this at all. The accuracy of this variant is the lowest, as the tradeoff for its small size.

Like all LLMs, coding models hallucinate too. The wrong answers they give are only inches away from the right answers. In case of SQL, the code is not only presentable, but also executable, hence returning the wrong rows.

I'm clueless, and curious how users will deal with this.
  • 3 replies
·
posted an update 16 days ago
view post
Post
2288
Introducing 🎉 OneSQL-v0.1🥳, our first text-to-SQL model based on Qwen2.5-Coder. This model has achieved an EX score of 63.33 on the BIRD leaderboard (https://bird-bench.github.io/).

The model family includes 7B and 32B,
onekq-ai/onesql-v01-qwen-67d8e3eb1611c5532bb90c5f
and can be also found on Ollama (https://ollama.com/onekq/OneSQL-v0.1-Qwen)

My goal is to make OneSQL the most usable open-weights model for text-to-SQL. I'm currently working on best practices to help users use this model the right away and avoid pitfalls. After that, I plan to train the next version to push for a higher EX score.

Enjoy this model and feel free to share comments/questions 🤗
  • 1 reply
·