Open-weight model
Llama-3.2-3B-Instruct-DSpark
by Yosef Worku Alemneh rasyosef/Llama-3.2-3B-Instruct-DSpark
Llama-3.2-3B-Instruct-DSpark is an open-weight model from Yosef Worku Alemneh, released under llama3.2. It has 750M parameters. At 16-bit it needs about 1.8 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
A DSpark draft model for speculative decoding with unsloth/Llama-3.2-3B-Instruct as the verifier, trained with speculators.
Runs On
What it takes to serve Llama-3.2-3B-Instruct-DSpark (750M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 1.5 GB | 1.8 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.7 GB | 0.9 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.4 GB | 0.4 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 20, 2026.
Llama-3.2-3B-Instruct-DSpark on every accelerator the SAVRN Index prices, at every precision
Model Card
A DSpark draft model for speculative decoding with unsloth/Llama-3.2-3B-Instruct as the verifier, trained with speculators. The drafter proposes 8 tokens at a time and the verifier checks them in one forward pass, so output is identical to running the verifier alone — a lossless speedup. Mean acceptance length is 3.27 tokens committed per verification step, up to 4.92 on mathreasoning. vLLM loads the verifier automatically from the config — don't pass it separately. Then query the OpenAI-compatible endpoint at http://localhost:8000/v1. 5 Qwen3 layers (hidden size 3072, intermediate size 8192, 24 attention heads over 8 KV heads, sliding-window attention with a 2048-token window), ~0.9B…
Excerpt from the card by Yosef Worku Alemneh, licensed llama3.2.
Configuration
- Architecture
- DSparkDraftModel
Identity and Version
- Repository
- rasyosef/Llama-3.2-3B-Instruct-DSpark
- Publisher
- Yosef Worku Alemneh
- Task
- Not stated by the source
- Modality
- Other
- Library
- transformers
- Parameters
- 750M parameters
- Languages
- Not stated by the source
- Revision
- 19b6ed73f2900e306c24b39605b1b9ab9783e8e5
- First published
- 2026-09-20
- Last updated
- 2026-09-20
Files and Weights
10 files, 2.8 GB in total. The weights are 3 files totalling 2.8 GB in pt, safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 1.5 GB | b19dc44309e3 |
| optimizer_state_dict.pt | Weights | 1.3 GB | 891ff840a70a |
| scheduler_state_dict.pt | Weights | 1.6 KB | fc7dddc6f27f |
| config.json | Configuration | 2.1 KB | — |
| config.py | Configuration | 2.3 KB | — |
| training_state.json | Configuration | 51 B | — |
| val_metrics.json | Configuration | 809 B | — |
| README.md | Documentation | 4.7 KB | — |
| train_command.txt | Other | 779 B | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- llama3.2
- Access
- Open weights, no gate
- Download size
- 2.8 GB
Released by Yosef Worku Alemneh through its official repository on Hugging Face.
Built From
- Derived from unsloth/Llama-3.2-3B-Instruct
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 2.8 GB |
| 16-bit | 1.5 GB |
| 8-bit | 0.7 GB |
| 4-bit | 0.4 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About Llama-3.2-3B-Instruct-DSpark
How much GPU memory does Llama-3.2-3B-Instruct-DSpark need?
About 1.8 GB at 16-bit and 0.4 GB at 4-bit: the weights (750M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Llama-3.2-3B-Instruct-DSpark on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What license is Llama-3.2-3B-Instruct-DSpark released under?
llama3.2, as its publisher declares it. Read the license text before commercial use.