@maximelabonne

Massive unlock: pre-training for $0 Alexander Amini (@xanamini) Three years ago we started working on a stealth project that we weren’t sure we’d ever talk about publicly… until today. Breakthrough: Introducing LFM-Zero: the first foundation model trained on 0 tokens. No pretraining. No finetuning. No data. Instead, we initialize from an implicit probabilistic prior over the underlying data-generating process, allowing the model to converge without ever observing data. LFM-Zero matches or surpasses models trained on 10T+ tokens across reasoning, coding, and multimodal tasks. Turns out that pretraining was just regularization that was holding us back. > Read our Tech Report here: tinyurl.com/lfm-zero — https://nitter.net/xanamini/status/2039403320247480469#m

→ View original post on X — @maximelabonne, 2026-04-01 22:11 UTC

Very interesting results about the fine-tunability of different models. 👀 LFM2 is more flexible than alternatives. It also confirms some common knowledge about RL degrading fine-tuneability. Jacek Golebiowski (@j_golebiowski) We benchmarked 15 small language models across 9 tasks to find out which one you should actually fine-tune. The most surprising result: Liquid AI's LFM2-350M ranked #1 for tunability. 350M parameters, absorbing training signal more effectively than models 20x its size. The entire LFM2 family swept the top 3 spots. No other architecture came close. LFM2-350M: avg rank 2.11 (±0.89) LFM2-1.2B: avg rank 3.44 LFM2.5-1.2B-Instruct: avg rank 4.89 That tight CI means it's consistent across every task type, not just a few lucky benchmarks. — https://nitter.net/j_golebiowski/status/2033611679645266280#m

→ View original post on X — @maximelabonne, 2026-03-16 19:06 UTC