ljupco/LFM2.5-8B-A1B-GGUF overview
LFM2.5 8B A1B — GGUF mixed 4 bit quant A GGUF conversion of LiquidAI's LFM2.5 8B A1B MoE, ~1B active for llama.cpp / vllm.cpp, with a mixed quantization: the b…
Runs locally from ~4.45 GB disk (8 GB VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| LFM2.5-8B-A1B-Q4_0h.gguf | GGUF | Q4_0H | 4.45 GB | Download |
Model Details
Model README
---
license: other
base_model: LiquidAI/LFM2.5-8B-A1B
quantized_by: ljupco
tags:
- llama.cpp
- gguf
- lfm2.5
---
LFM2.5-8B-A1B — GGUF (mixed 4-bit quant)
A GGUF conversion of LiquidAI's LFM2.5-8B-A1B (MoE, ~1B active) for llama.cpp /
vllm.cpp, with a mixed quantization: the bulk of the weights are 4-bit (Q4_0) while
the most sensitive tensors keep a higher precision. This is one of the four models
benchmarked in the three-engine report:
Usage
llama-cli -m LFM2.5-8B-A1B-Q4_0h.gguf -p "The capital of France is" -n 64Credits and Acknowledgements
This is a quantization of LFM2.5-8B-A1B by Liquid AI. We are deeply grateful
to Liquid AI for the LFM2.5 family, its gated-delta / shortconv architecture, and for
publishing the weights openly. This work builds directly on theirs, and we thank them
profusely. See the report above for the full acknowledgement.
Run ljupco/LFM2.5-8B-A1B-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models