Quanquan Gu @quanquangu - Bluesky Profile

Self-Play Fine-Tuning of Diffusion Models for Text-to-Image Generation Fine-tuning Diffusion Models remains an underexplored frontier in generative artificial intelligence (GenAI), especially when compared with the remarkable progress made in fine-tuning Large Language M...

Papers #2-3: arxiv.org/abs/2402.10210 and arxiv.org/abs/2405.00675 from the incredible
@quanquangu.bsky.social. I really like how they explore new techniques for RLHF

20.12.2024 16:53 — 👍 3 🔁 3 💬 1 📌 0

Pretraining will only end once we find the optimal scaling law.

14.12.2024 08:07 — 👍 6 🔁 0 💬 1 📌 0

To better interpret the plot, draw a horizontal line representing a specific target validation loss. Find the points where this line intersects the curves for AdamW and MARS, which will allow you to determine how much speedup, in terms of training tokens, MARS achieves compared to AdamW.

05.12.2024 02:54 — 👍 0 🔁 0 💬 0 📌 0

Just added you.

03.12.2024 23:49 — 👍 1 🔁 0 💬 0 📌 0

GitHub - AGI-Arena/MARS: The official implementation of MARS: Unleashing the Power of Variance Reduction for Training Large Models The official implementation of MARS: Unleashing the Power of Variance Reduction for Training Large Models - AGI-Arena/MARS

With the delivery of MARS complete, the focus now shifts to delivering new architectures.

03.12.2024 23:48 — 👍 3 🔁 2 💬 0 📌 0

Just added you! Welcome!

03.12.2024 01:17 — 👍 1 🔁 0 💬 0 📌 0

Just added you.

02.12.2024 21:53 — 👍 0 🔁 0 💬 0 📌 0

Just added you.

01.12.2024 00:35 — 👍 1 🔁 0 💬 0 📌 0

Just added you!

30.11.2024 04:38 — 👍 1 🔁 0 💬 0 📌 0

Just added you!

29.11.2024 22:47 — 👍 1 🔁 0 💬 1 📌 0

Just added you.

29.11.2024 22:32 — 👍 1 🔁 0 💬 1 📌 0

This Thanksgiving, I want to express my heartfelt gratitude to all the students, colleagues, and collaborators who have contributed to the success of SPIN, SPPO, DPLM, GPM, MARS, and many other projects. Your hard work and dedication continue to be truly inspiring.

29.11.2024 03:22 — 👍 14 🔁 0 💬 0 📌 1

Just added you!

28.11.2024 19:29 — 👍 1 🔁 0 💬 0 📌 0

Just added you!

28.11.2024 19:16 — 👍 1 🔁 0 💬 0 📌 0

Just added you.

28.11.2024 19:14 — 👍 1 🔁 0 💬 0 📌 0

Anyone using their real name and interested is welcome!

28.11.2024 02:44 — 👍 0 🔁 0 💬 0 📌 0

Just added you. Welcome!

28.11.2024 01:48 — 👍 1 🔁 0 💬 0 📌 0

MARS is a unified framework that can be integrated with various precondition techniques. So it can be applied to PSGD. I believe @hessianfree.bsky.social has implemented MARS-PSGD.

28.11.2024 01:48 — 👍 3 🔁 0 💬 2 📌 0

Just added you!

28.11.2024 01:44 — 👍 1 🔁 0 💬 0 📌 0

Just added you.

28.11.2024 01:43 — 👍 1 🔁 0 💬 1 📌 0

Done!

28.11.2024 01:42 — 👍 1 🔁 0 💬 0 📌 0

Just added you.

28.11.2024 01:42 — 👍 0 🔁 0 💬 0 📌 0

Just added you!

28.11.2024 01:42 — 👍 0 🔁 0 💬 0 📌 0

Just added you!

28.11.2024 01:42 — 👍 1 🔁 0 💬 0 📌 0

Please reply to this message or DM me if you’d like to be added!

27.11.2024 20:48 — 👍 3 🔁 0 💬 3 📌 0

Just added you!

27.11.2024 20:46 — 👍 1 🔁 0 💬 0 📌 0

Have added both of you. Feel free to recommend other people.

27.11.2024 09:40 — 👍 1 🔁 0 💬 0 📌 0

Tulu 3 SFT mix trending on HuggingFace :D , next step make preferences and RL datasets more accessible.

26.11.2024 16:57 — 👍 15 🔁 2 💬 0 📌 0

OLMo 2 is out 🥳 7B and 13B trained on 5T tokens, and meticulousy instruction tuned using Tulu 3 recipe.

Simply the best fully open models yet.

Really proud of the work & the amazing team at
@ai2.bsky.social

26.11.2024 21:12 — 👍 263 🔁 44 💬 9 📌 2

Just added you there.

26.11.2024 21:08 — 👍 1 🔁 0 💬 0 📌 0

Quanquan Gu

Latest posts by quanquangu.bsky.social on Bluesky

@quanquangu is following 19 prominent accounts