2.a. Crawling task as a transaction 2.b. Crawled data as an asset
2.c. Fault-tolerant distributed high performance crawling
2.d. RAFT consensus algorithm
2.e. Decentralized data storage
2.f. DFSM decoding path
2.g. Transfer learning for updating the model parameters
3. gPredict: Decentralized gradient learning framework
3.a. Training task as a transaction
3.b. Trained model as an asset
3.c. Floating Point mapping
3.d. Hyper-Parameters routing
3.e. Results reducing
4. gCompute: Internet-scale AI dApps solution
4.a. AI model as an asset
4.b. Scalable
4.c. Controllable
4.d. Difficult
4.e. Maintainable
4.f. Responsive
4.g. Concurrent