howtospark
models / laguna-s-2-1

118B params~8.1B activebfloat16MoE 256×10GQA 48:81.0M ctxopenmdw-1.1
Sparks
~250 GB
~191 GB
~132 GB
~73 GB
~48 GB
~46 GB
~45 GB
110 usable
REAP dialno pruning · 256 experts
Spec decode
Context33K / 1.0MKV
On the Sparks — 1× · 2-bit experts + NVFP4 dense · 33K ctx · FP16 KV~365K ctx fits usable
spark
65 GB free
Expert planes28.5* GB2-bit
110 usable
45* / 110 GB
41% of usable
Expert planesDense weightsKV cacheActivations + graphsOS reserve

* segment sizes marked with an asterisk are estimates pending a measured run

Estimated per-node footprint, weights + KV sharded across the Spark (tensor parallel). Overhead ~8 GB/node; KV at FP16.

layer map
48 layersfull attention×12sliding window×36
in
out

one MoE layer · 256 experts, 10 fire per token

each expert ≈ 9M params

resident: 48 MoE layers × 256 experts × 9M 114.0B

active/token: 48 × 10 × 9M 4.5B experts

+ 3.5B always-on dense (attention, embeddings, shared stack) = 8.1B active

Eval scores (compare all)

Reported

70.2Terminal-Bench 2.1
78.5SWE-bench Multilingual
59.4SWE-bench Pro
40.4DeepSWE
46.2SWE Atlas (Codebase QnA)
49.7Toolathlon Verified

Recipes (all recipes)

Worked deployments that run this model on DGX Spark — each shows how it fills each node's unified memory.

Optimize
Verdict

118B params, ~235 GB at bfloat16 — doesn't fit natively, but ~59 GB at 4-bit does.

Make it fit

Get the weights (and headroom for KV) inside the 128 GB pool — the hard gate.

Quantize to 4-bit

At 235 GB native it doesn't fit the 110 GB budget; 4-bit weight-only quantization gets it to ~59 GB.

235 GB → ~59 GB

Make it fast

Once it fits: fewer bytes per token and fewer decode steps against 273 GB/s.

Add a speculative decoder

Decode is bandwidth-bound at ~67.8 tok/s ceiling; a trained EAGLE-style draft multiplies tokens per weight-read.

Prune experts for speed

Fewer experts ⇒ smaller working set ⇒ better cache/page behavior, even when it already fits.

Make it yours

Change what the model does — adapt, edit, or steer it — independent of size.

Fine-tune it

LoRA/QLoRA adapts behavior on one Spark without touching the base weights.