# gemma-4-E2B-it-dspark by Yosef Worku Alemneh: Open Model Source: https://savrn.com/models/gemma-4-e2b-it-dspark Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve gemma-4-E2B-it-dspark (336M 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 | 0.7 GB | 0.8 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 0.3 GB | 0.4 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 0.2 GB | 0.2 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/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](https://savrn.com/ai-index/pricing/gpus), read Oct 12, 2026. [gemma-4-E2B-it-dspark on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/gemma-4-e2b-it-dspark/gpus) ## Model Card By Yosef Worku Alemneh, published under apache-2.0, revision dc7ef5788ba8. A DSpark draft model for speculative decoding with [google/gemma-4-E2B-it](https://savrn.com/models/gemma-4-e2b-it) as the verifier, trained with [speculators](https://github.com/vllm-project/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. This makes it a lossless speedup. Mean acceptance length is 2.89 tokens committed per verification step, up to 5.53 on math_reasoning. On an A100 at concurrency 1, that gives a throughput speedup over the verifier alone of 3.74× on math_reasoning and 2.42× on HumanEval. Training code: [rasyosef/train-dspark-draft-models](https://github.com/rasyosef/train-dspark-draft-models). Trained on 100,000 samples. ### Usage vLLM loads the verifier automatically from the config, so don't pass it separately. ``` vllm serve rasyosef/gemma-4-E2B-it-dspark --port 8000 --gpu-memory-utilization 0.8 ``` Then query the OpenAI-compatible endpoint at http://localhost:8000/v1. ### Details [Read the full model card (880 words)](https://savrn.com/models/gemma-4-e2b-it-dspark/card) ## Configuration Architecture DSparkDraftModel ## Identity and Version Repository rasyosef/gemma-4-E2B-it-dspark Publisher Yosef Worku Alemneh Task Not stated by the source Modality Other Library transformers Parameters 336M parameters Languages Not stated by the source Revision dc7ef5788ba82322afb410ce9601d0be6db34748 First published 2026-09-23 Last updated 2026-09-26 ## Files and Weights 8 files, 671.2 MB in total. The weights are 1 file totalling 671.2 MB in safetensors. Weights1 file · 671.2 MB Configuration4 files · 5.2 KB Documentation1 file · 7.2 KB Other1 file · 789 B Repository1 file · 1.5 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model.safetensors | Weights | 671.2 MB | 527de14bb127 | | config.json | Configuration | 2.0 KB | — | | config.py | Configuration | 2.3 KB | — | | training_state.json | Configuration | 51 B | — | | val_metrics.json | Configuration | 812 B | — | | README.md | Documentation | 7.2 KB | — | | train_command.txt | Other | 789 B | — | | .gitattributes | Repository | 1.5 KB | — | ## License and Download License apache-2.0 Access Open weights, no gate Download size 671.2 MB [Download from Yosef Worku Alemneh](https://huggingface.co/rasyosef/gemma-4-E2B-it-dspark) Released by Yosef Worku Alemneh through its official repository on Hugging Face. [Read the license](https://www.apache.org/licenses/LICENSE-2.0). ## Built From - Derived from [google/gemma-4-E2B-it](https://savrn.com/models/gemma-4-e2b-it) ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 671.2 MB | | 16-bit | 0.7 GB | | 8-bit | 0.3 GB | | 4-bit | 0.2 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About gemma-4-E2B-it-dspark ### How much GPU memory does gemma-4-E2B-it-dspark need? About 0.8 GB at 16-bit and 0.2 GB at 4-bit: the weights (336M parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run gemma-4-E2B-it-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. ### Can I use gemma-4-E2B-it-dspark commercially? Yes. gemma-4-E2B-it-dspark is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. ## Yosef Worku Alemneh [All models and datasets](https://savrn.com/model-publishers/rasyosef) ## Versions - [dc7ef5788ba8](https://savrn.com/models/gemma-4-e2b-it-dspark/versions/dc7ef5788ba8) · current 2026-09-26 ## Explore More - [All models under apache-2.0](https://savrn.com/models/licenses/apache-2-0) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-09-26. - [Hugging Face record](https://huggingface.co/rasyosef/gemma-4-E2B-it-dspark) - [How the hub is built](https://savrn.com/model-hub/methodology)