CrowdGPT Launches Decentralized Training Network for Independent GPUs
Start here
CrowdGPT has launched a decentralized platform that trains a single language model across independent GPUs without large-scale infrastructure. Each connected device downloads the global weights from HuggingFace, trains on a 10MB shard of tokenized data, and submits a compressed weight delta; the server checks each delta for NaN/Inf values and for size and norm bounds before averaging it into the global model. The dataset is fully open and anyone can contribute to it, and model weights are automatically pushed to HuggingFace for public auditing and testing. CrowdGPT presents the setup as a way to coordinate many independent GPUs into one cohesive model without heavy infrastructure.
Generated from reports · updated 8 h ago
Report timeline
Follow the reports to see the event from different sides.
- Hacker News · AICrowdGPT: A Decentralized Network Training One Language Model Across Independent GPUs
CrowdGPT coordinates independent GPUs to train a single language model without large-scale infrastructure. Each connected device downloads global weights from HuggingFace, trains on a 10MB tokenized data shard, and submits a compressed weight delta that the server validates for NaN/Inf values, size and norm bounds before averaging into the global model.
Heat for this event
- Current, comparable range
- 8
- Peak, comparable range
- 10November 16 09:00
- Change in 24 hours
- –
The trend compares the same participants over hours observed continuously, so its range can be narrower than the current heat count. Hover or tap the chart for hourly heat; the arrow keys move through it.