LFM2-350M
Liquid AI · 2025-07-10 · 354M parameters
LFM2-350M is the smallest of the four post-trained checkpoints in Liquid AI's July 10, 2025 LFM2 launch, a hybrid language model for on-device and edge deployment. Its 16 layers combine 10 double-gated short-range convolution blocks with six grouped-query attention blocks at hidden size 1,024, and the card gives a 32,768-token context, a 10-trillion-token training budget and eight languages. The BF16 checkpoint contains 354,483,968 parameters, and the tokenizer assigns 64,400 distinct IDs within 65,536 embedding rows.