LFM2.5 8B A1B DSpark Draft v1 needs about 0.8 GB of memory at Q4_K_M (a 0.2 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 3 quants, from 0.2 GB (Q4_K_M) to 0.7 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 0.3 GB.
LFM2.5 8B A1B DSpark Draft v1 – a lightweight draft sidecar for LFM2.5
LFM2.5-8B-A1B-DSpark is a speculative‑decoding draft model with 327.7 M draft parameters. It shares token embeddings and the LM head with the target LFM2.5‑8B‑A1B model. The draft contains 5 attention layers, a rank‑256 Markov head and a confidence head.
Good at
- Fast speculative decoding
- Exact draft verification
- Low VRAM footprint
Limits
- Requires paired target LFM2.5‑8B‑A1B‑GGUF
- No vision support
- License is other
Best for
- On‑device inference on Apple silicon
- Low‑latency text generation
- Edge deployment with limited VRAM
Who it’s for. Developers building low‑memory LLM applications that need faster inference. It is suitable for users who already have the target LFM2.5‑8B‑A1B model and want to add speculative decoding.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 0.2 GB | 0.8 GB | Fits | Get file → | |
| Q8_0 | 0.4 GB | 0.9 GB | Fits | Get file → | |
| F16 | 0.7 GB | 1.2 GB | Fits | Get file → |
For 8 GB VRAM, use Q4_K_M (0.2 GB file, ~0.8 GB memory at 8 k context). If 12 GB VRAM is available, Q8_0 (0.36 GB file, ~0.9 GB memory) offers higher precision. F16 (0.66 GB file, ~1.2 GB memory) is suitable only for 16 GB+ VRAM setups.
What GPU do you need for LFM2.5 8B A1B DSpark Draft v1?
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 8B A1B DSpark Draft v1 quant should you choose?
Q4_K_M is the usual starting point for LFM2.5 8B A1B DSpark Draft v1: it balances file size and quality well. Here it is a 0.2 GB file.
The Q8_0 file is 0.4 GB, about 1.8 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about LFM2.5 8B A1B DSpark Draft v1
How much VRAM does LFM2.5 8B A1B DSpark Draft v1 need?
About 0.8 GB at Q4_K_M with an 8K context window. That is the 0.2 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 8B A1B DSpark Draft v1 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 8B A1B DSpark Draft v1 GGUF?
It depends on the quant: Q4_K_M is 0.2 GB, Q8_0 is 0.4 GB, F16 is 0.7 GB.
What context length does LFM2.5 8B A1B DSpark Draft v1 support?
The GGUF header lists a trained context length of 128,000 tokens (125K). 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 8B A1B DSpark Draft v1 GGUF?
The files are on Hugging Face in the LiquidAI/LFM2.5-8B-A1B-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 8B A1B DSpark Draft v1 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 LFM2.5‑8B‑A1B‑DSpark on 8 GB VRAM?
Yes, the Q4_K_M quantization uses about 0.8 GB of VRAM at 8 k context, which fits within an 8 GB GPU.
Does this model support vision?
No, LFM2.5‑8B‑A1B‑DSpark is a text‑only draft sidecar.
Do I need the target model to use this draft?
Yes, the draft shares embeddings and the LM head with the target LFM2.5‑8B‑A1B model, so both must be loaded together.