Research Scientist at Meta FAIR | Coding & Computer use agents | CS PhD @UMich
Seattle, WA
Joined May 2016
- Tweets152
- Following451
- Followers967
- Likes793
Shengyi Qian retweeted
Metas Muse Glimmer tech report is pretty interesting.
Metas Muse Glimmer report describes training agents inside simulated digital lives, complete with 10–14 contacts, weeks of messages, calendars and documents. The agents practice tasks spanning multiple days, while training trajectories containing hallucinated results or unverified claims of success are discarded.
Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.
It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.
Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol
Shengyi Qian retweeted
Meta's Muse Glimmer report cites our 2023 weight averaging paper (COLM'24). Their final training recipe also uses uniform averaging of 600 steps apart checkpoints, (what we proposed).
Great to see our work being adopted. Thanks, @AIatMeta!
If you wanna understand the science of weight averaging keep reading 🧵 (1/n).
Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.
It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.
Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol
Shengyi Qian retweeted
Technical report of Muse Glimmer, a small and nimble model that is well-rounded. Enjoy!
Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.
It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.
Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol
Shengyi Qian retweeted
The Muse Glimmer technical report is out today! It is our 30B open-weight model (Apache 2.0) distilled from Muse Spark, built to run agentic workloads entirely on consumer hardware — one RTX 5090, 32GB.
Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.
It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.
Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol
Shengyi Qian retweeted
We just released details about our work on Muse Glimmer! Check it out!
Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.
It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.
Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol
Shengyi Qian retweeted
Very excited to share the tech report on Muse Glimmer.
It was great fun to work on this with folks at @AIatMeta.
Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.
It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.
Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol
Shengyi Qian retweeted
Excited to share the Muse Glimmer technical report! It was a great collaborative effort across multiple teams, and I'm glad to have been part of it!
Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.
It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.
Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol
🚀 Honored to be a core contributor to Muse Glimmer! The model is open — and so is the homework. Our tech report lays out how we built the open-weight model from pretraining to alignment.
Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.
It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.
Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol
Excited to see Muse Spark Open released! Three things I particularly appreciate: (1) efficiency before scaling, (2) opening up frontier architecture ideas, and (3) multimodal-native design from day one. A few personal thoughts 🧵
2/ I love seeing frontier architecture details shared openly, e.g. LatentMoE. @Kimi_Moonshot explores a similar idea with different routing and expert choices. Open weights are also about sharing the design tradeoffs that help move the whole field forward.
3/ Multimodal-native from day one matters to me. We considered multimodality from pretraining, not as a late add-on. Unlike @Alibaba_Qwen's text-only open-weight release, Muse Spark Open shares multimodal capabilities too. The foundation—and what we open—both matter.
Shengyi Qian retweeted
Muse Wonder, aka Watermelon is out today!! The consumer use cases of CUA are obvious from Muse, a particular use case I am super bullish about is verification.
CUA offers a feature-complete harness (GUI + code) for the agent to verify its creation, give itself feedback, and self-improve, aka CUA as verification.
This demo uses Muse Wonder to generate a globe web-app from a reference image and prompt. Two agents, same model: (a) a generation main agent, (b) a CUA verification sub-agent. They iterate until the CUA verifier is satisfied. This is an interesting test-time scaling with CUA.
As usual, work done with the incredible CUA and visual coding team at Meta MSL.
@taoyds, @yuning_pro, @EdwardSun0909
What a journey! It's amazing to see how far we've come in such a short time.
Building frontier AI is a massive team effort, and I'm grateful to work alongside so many talented people pushing the boundaries of what's possible.
Proud of the team, and excited for what's next! 🚀
Shengyi Qian retweeted
Today we are launching Muse Wonder and Muse Wonder Full, our best models yet, with state-of-the-art performance on personal and professional agentic tasks. They are built for long running, multi step work. Muse Wonder is available today in the Meta Model API and Muse Code, and Muse Wonder Full in Muse Code.
Shengyi Qian retweeted
muse wonder (full) 🍉 is our largest and most capable model yet!
we've made huge leaps with every release, and the pace of progress keeps accelerating. there is no wall and superintelligence is coming
Shengyi Qian retweeted
Muse Wonder is out! 🍉
Evals saturate quickly these days, so you have to evolve evals as models improve. We spent much time building Personal Superintelligence Index, including internal evals like FocusBench, to both measure PSI progress and give us clear guidance for training!
🍉 MUSE WONDER 🍉
our research process starts with evals. we developed the Personal Superintelligence Index (PSI) internally to help us measure the important dimensions of what actually makes a model useful for personal agents, vs just focusing on external benchmarks.
the end result is that our model is incredible for personal agents in many subtle and important ways, including Active Memory and Active Focus.
i'm especially proud of our research team for taking a particular focus on developing useful models. the whole point of building advanced frontier models is to make them useful to real people, and we have thought about that every step of the way in our research process.
muse wonder is quite advanced across the board, and certainly has big model smell (which we quantified!), but is truly state of the art for personal agents.
Excited to share Muse Wonder and Muse Wonder Full!
We launched 🍉 (Muse Wonder) and 🍉 Ultra (Muse Wonder Full) today!
What excites me most is how much closer we’re getting to agents you can actually delegate work to. Personal agents that understand you and your life, and professional agents that can take on complex, long-running projects. 🍉 is designed for personal and professional agent use
Pretty Wonder Full to see this come together just a year after starting the agent team at TBD Labs. Huge credit to the incredible team that made it happen!
