LiquidAI/LFM3-1.7B
AI model · https://huggingface.co/LiquidAI/LFM3-1.7B
First seen by Oriel on .
AI summary
LFM3-1.7B is a 1.73B-parameter tri-hybrid text model designed for local agents, tool calling, structured outputs, and efficient domain specialization on edge hardware.
- Its 24-layer architecture combines 13 gated short convolutions, five sliding-window attention layers, and six global attention layers with a 256K-token context window.
- At 4-bit precision, it requires about 1 GB of memory and supports local inference through GGUF, MLX, and ONNX variants.
- At 32K context, it prefills at least 2.5x faster and decodes at least 1.8x faster than every compared model across three tested devices.
- Training used 22T pre-training tokens, followed by SFT, mixed reinforcement learning, multi-domain on-policy distillation, and agentic RL in real harnesses.
- The model is unsuitable for complex multi-step reasoning or agentic coding because of its size, and retrieval or search is recommended for knowledge-heavy tasks.
#language models #edge ai #llm agents #tool calling
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Likes
Total likes reported by the source.
- Hugging Face Models: 1,897, latest observation ; +1,786 over observed period since
Downloads (30d)
Rolling 30-day downloads. A change in this value is not new downloads during the observed period.
- Hugging Face Models: 106,961, latest observation ; +100,650 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hugging Face Models | LiquidAI/LFM3-1.7B | https://huggingface.co/LiquidAI/LFM3-1.7B Read discussion |