> For the complete documentation index, see [llms.txt](https://navyai.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://navyai.gitbook.io/docs/gpu-contribution-program/contribution-flows.md).

# Contribution Flows

Fact, graphics cards of sufficient quality for training LLM models, such as the RTX 3080 or higher, are exceedingly rare, while weaker graphics cards are readily available in abundance. So, why don't we merge all the low qualify graphic card to create a new one that qualify enough for training LLM ?

<figure><img src="/files/HNyn1CknrQq7Ec03hABQ" alt=""><figcaption></figcaption></figure>

Step 1: User contributes GPU configuration \
Step 2: NavyAI's productivity tracking system will evaluate and mine the performance of the configuration.\
Step 3: After completing the mining process, the performance of the configurations that the user contributes will be merged through the merging system and performance synthesis. \
Step 4: Create a virtual server with input performance taken from the merging system \
Step 5: Provide virtual servers to decentralized GPU platforms as a worker.