LFM2-350M-Math
Based on LFM2-350M, LFM2-350M-Math is a tiny reasoning model designed for tackling tricky math problems.
You can find more information about other task-specific models in this blog post.
📄 Model details
Generation parameters: We strongly recommend using greedy decoding with a temperature=0.6, top_p=0.95, min_p=0.1, repetition_penalty=1.05.
System prompt: We recommend not using any system prompt.
Supported languages: English only.
Chat template: LFM2 uses a ChatML-like chat template as follows:
<|startoftext|><|im_start|>user
Find the sum of all integer bases $b>9$ for which $17_{b}$ is a divisor of $97_{b}$.<|im_end|>
<|im_start|>assistant
<|cot_start|>First, we need to convert $17_{b}$ and $97_{b}$ into base 10. [...]<|im_end|>
You can automatically apply it using the dedicated .apply_chat_template() function from Hugging Face transformers.
[!WARNING] ⚠️ The model is intended for single-turn conversations.
📈 Performance
Reasoning enables models to better structure their thought process, explore multiple solution strategies, and self-verify their final responses. Augmenting tiny models with extensive test-time compute in this way allows them to even solve challenging competition-level math problems. Our benchmark evaluations demonstrate that LFM2-350M-Math is highly capable for its size.
As we are excited about edge deployment, our goal is to limit memory consumption and latency. Our post-training recipe leverages reinforcement learning to explicitly bring down response verbosity where it is not desirable. To this end, we combine explicit reasoning budgets with difficulty-aware advantage re-weighting. Please refer to our separate blog post for a detailed post-training recipe.
🏃 How to run
- Hugging Face: LFM2-350M
- llama.cpp: LFM2-350M-Math-GGUF
- LEAP: LEAP model library
You can use the following Colab notebooks for easy inference and fine-tuning:
| Notebook | Description | Link |
|---|---|---|
| Inference | Run the model with Hugging Face's transformers library. | |
| SFT (TRL) | Supervised Fine-Tuning (SFT) notebook with a LoRA adapter using TRL. | |
| DPO (TRL) | Preference alignment with Direct Preference Optimization (DPO) using TRL. | |
| SFT (Axolotl) | Supervised Fine-Tuning (SFT) notebook with a LoRA adapter using Axolotl. | |
| SFT (Unsloth) | Supervised Fine-Tuning (SFT) notebook with a LoRA adapter using Unsloth. |
📬 Contact
- Got questions or want to connect? Join our Discord community
- If you are interested in custom solutions with edge deployment, please contact our sales team.
Citation
@article{liquidai2025lfm2,
title={LFM2 Technical Report},
author={Liquid AI},
journal={arXiv preprint arXiv:2511.23404},
year={2025}
}