Hysen Labs
Open-source project
BaseIntelligence/base avatar
BaseIntelligence

base

[🧠] Base is a Bittensor subnet enabling decentralized collaborative AI research through multiple challenges, each focused on a specific objective where miners compete and contribute innovative code.

160 stars16 forksRustApache-2.0
DEEP OPEN-SOURCE ANALYSIS

Ingot links plant data to traceable process experiments

liuweichaox/Ingot is an open source system for manufacturing process tracing and optimization that runs on site. It connects production runs, quality results, candidate causes, and verification experiments without sending data off the factory floor.

What problem it addresses

Traditional process development relies on personal memory and scattered tables, and even when machines produce data, the conditions, process curves, and quality results often cannot be tied to the same real run. Ingot builds a trusted data loop first, then helps engineers see what a run actually used, link settings and inspection results to that run, compare comparable runs, and separate candidate causes from confounders and weak evidence.

How the system is put together

The product is composed of Edge, which connects controls, instruments, gateways, and business data sources, plus Process Executions, Manufacturing context, Inspections, Research, an Optimizer, and an Agent. Edge and Platform stay separate processes even when on one machine, with their own storage and recovery. The reference Compose topology is a single API, a single worker, and a single PostgreSQL instance.

The analysis methods

Ingot does not fix one advanced algorithm as the answer to every problem. It uses data quality statistics, matched comparisons and other statistical methods, mixed-effects models for context, controlled experiments for causes, and Gaussian process or constrained Bayesian optimization via BoTorch for expensive small-sample search. A language model helps understand the problem and interpret evidence through read-only tools, but it does not emit numeric process settings directly.

Editorial conclusion

The project is released under the MIT license and starts with a Docker Compose stack that brings up the engineering workbench on port 3000.

DEEP OPEN-SOURCE ANALYSIS

Official sources

Community notes

Community notes