LFM2.5-24B-A2B-DSpark needs about 1.2 GB of memory at F16 (a 0.7 GB file) with an 8K context window, so it fits on a 4 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 0.3 GB.
LFM2.5-24B-A2B-DSpark: a draft sidecar for speculative decoding
LFM2.5-24B-A2B-DSpark is a draft sidecar that accelerates generation by proposing tokens for the target model to verify. It contains 327.7M draft parameters and shares embeddings and LM head with the target. The model is built on a 5‑layer dflash backbone.
Good at
- Speculative decoding
- Token proposal acceleration
- Draft verification
Limits
- Requires paired target model
- No vision support
- Large VRAM needed for target
Best for
- Speed‑critical inference
- Edge deployment with draft sidecar
- Optimizing throughput on GPUs
Who it’s for. Developers building inference pipelines that need faster generation. The model is suitable for users who can pair it with a compatible target and have the necessary GPU resources.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| F16 | 0.7 GB | 1.2 GB | Fits | Get file → |
The only available GGUF quant is F16, 0.66 GB in size. It requires about 1.2 GB of VRAM at 8 k context, which is approximate. For lower memory budgets, a different target quantization should be chosen, but the draft itself remains the same size.
What GPU do you need for LFM2.5-24B-A2B-DSpark?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is F16 (0.7 GB file, about 1.2 GB in use).
Which LFM2.5-24B-A2B-DSpark quant should you choose?
F16 (0.7 GB) is the middle option in this repository. The unquantized model. It gives original quality and needs the most memory, so it is mainly for conversion or reference.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about LFM2.5-24B-A2B-DSpark
How much VRAM does LFM2.5-24B-A2B-DSpark need?
About 1.2 GB at F16 with an 8K context window. That is the 0.7 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run LFM2.5-24B-A2B-DSpark on an 8 GB GPU?
Yes, with F16: the 0.7 GB file needs about 1.2 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of LFM2.5-24B-A2B-DSpark GGUF?
It depends on the quant: F16 is 0.7 GB.
What context length does LFM2.5-24B-A2B-DSpark support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.1 GB at 8K and 0.3 GB at 32K.
Where can I download LFM2.5-24B-A2B-DSpark GGUF?
The files are on Hugging Face in the LiquidAI/LFM2.5-24B-A2B-DSpark-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does LFM2.5-24B-A2B-DSpark use?
Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run LFM2.5-24B-A2B-DSpark on 8 GB VRAM?
The draft alone needs about 1.2 GB at 8 k context, but the target model must also fit in VRAM. 8 GB may be insufficient for the full pipeline.
Does the model support vision tasks?
No, the model has no vision capabilities.
What is the context length of the draft?
The draft was trained on 131072 tokens and supports that context length.