LFM2.5-8B-A1B-DSpark-GGUF

LFM2.5-8B-A1B DSpark GGUF

GGUF quantizations of LiquidAI DSpark draft model for LFM2.5-8B-A1B.

Use with BeeLlama.cpp, a llama.cpp fork with advanced quantization features.

---

<div align="center"> <img src="https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/2b08LKpev0DNEk6DlnWkY.png" alt="Liquid AI" style="width: 100%; max-width: 100%; height: auto; display: inline-block; margin-bottom: 0.5em; margin-top: 0.5em;" /> <div style="display: flex; justify-content: center; gap: 0.5em; margin-bottom: 1em;"> <a href="https://playground.liquid.ai/"><strong>Try LFM</strong></a> • <a href="https://docs.liquid.ai/lfm/getting-started/welcome"><strong>Docs</strong></a> • <a href="https://leap.liquid.ai/"><strong>LEAP</strong></a> • <a href="https://discord.com/invite/liquid-ai"><strong>Discord</strong></a> </div> </div>

LFM2.5-8B-A1B-DSpark

LFM2.5-DSpark is a family of speculative-decoding draft models that adapt DSpark for the LFM2.5 architecture. They allow LFM2.5 models to run faster without degrading quality.

This is a drafter for LiquidAI/LFM2.5-8B-A1B. In SGLang, decoding runs about 2.6× faster. It also runs on-device on Apple silicon through the Metal backend.

Find more information about LFM2.5-DSpark in our blog post.

🗒️ Model Details

LFM2.5-8B-A1B-DSpark is a DSpark speculative-decoding draft model with the following features:

• Target model: LiquidAI/LFM2.5-8B-A1B
• Draft parameters: 327.7M (BF16)
• Backbone: 5 full attention layers, hidden_size=2048, intermediate_size=6144 with SiLU/SwiGLU, GQA with num_attention_heads=32 / num_key_value_heads=8, head_dim=64
• Extra heads: Markov head (rank 256) + confidence head
• Block size: 9
• Vocabulary: 128,000

Other models in the LFM2.5-DSpark family:

📊 Performance

Benchmarks

Speculative decoding is exact: the target verifies every proposed token, so the generated text is what the target would have produced on its own. See LiquidAI/LFM2.5-8B-A1B for performance benchmarks.

Acceptance

Mean accepted tokens per decoding step, by benchmark (1×H100, batch size 1, greedy decoding). Higher means more of the draft's proposed block is accepted per target forward pass, so decoding is faster (at block size 9, the ceiling is 10).

On-device and GPU Inference

🏃 How to run (SGLang)

Requires a build of SGLang with DSpark support for LFM2 / LFM2-MoE targets (PR #31041). Launch the target with the draft attached:

python -m sglang.launch_server \
  --model-path LiquidAI/LFM2.5-8B-A1B \
  --speculative-algorithm DSPARK \
  --speculative-draft-model-path LiquidAI/LFM2.5-8B-A1B-DSpark \
  --speculative-draft-attention-backend flashinfer \
  --disable-radix-cache --mem-fraction-static 0.75 --port 30000

Then query the OpenAI-compatible endpoint at http://localhost:30000/v1. The block size is read from the draft's config.json; the baseline is the same command without the three --speculative-* flags.

📬 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{liquidAI202626B,
  author  = {Liquid AI},
  title   = {LFM2.5-2.6B: Agents Everywhere},
  journal = {Liquid AI Blog},
  year    = {2026},
  note    = {www.liquid.ai/blog/lfm2-5-2-6b},
}

@article{liquidAI2026dspark,
  author = {Liquid AI},
  title = {LFM2.5-DSpark: Up to 3.2x Faster Inference from H100 to MacBook},
  journal = {Liquid AI Blog},
  year = {2026},
  note = {www.liquid.ai/blog/lfm2.5-dspark},
}