Page 636 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

3 Use Case Requirements

• REQ-01: For text generation, we would require the following resources:
GPU: Nvidia A100 (80 GB vRAM PCIe) x 2
CPU: 64vCore, Memory: 64 GiB
• REQ-02: For the toxic check model, we would require the following resources:
GPU: A100 80GB PCIe MIG 1g-10GB
CPU: 2vCore, Memory: 24 GiB
• REQ-03: For the tone change model, we would require the following resources:
GPU: A100 80GB PCIe MIG 1g-10GB
CPU: 2vCore, Memory: 24 GiB
• REQ-04: For the relevance check model, we would require the following resources:
GPU: A100 80GB PCIe MIG 3g-40GB
CPU: 2vCore, Memory: 24 GiB
• REQ-05: For the vector store, we would require the following resources:
Milvus vector store + Google Cloud Storage
vCU: Auto-scaling vCU and Storage

4 Sequence Diagram

5 References

• [1] https:// elice .io/ en/ newsroom/ eliceschool -aidt
• [2] https:// elice -library .elice .io - Pilot Deployment

600