去中心化推理Decentralised inference
Gensyn:可审计的去中心化训练Gensyn: auditable decentralised training
交集:把训练过程做成可复现、可审计的执行环境,让外部能验证模型是怎么训出来的,而不是只看最终权重。
适用范围:官方发布 open-1b/REE/open-8b 等可审计训练与可复现执行环境;它是研究/生产平台,不是通用 GPU 租用市场。
未解决:可审计性覆盖到哪一层(数据、代码、随机性、硬件)、大规模任务成本、以及验证者与训练者串谋风险,官方材料未给出完整答案。Intersection: make training reproducible and auditable so outsiders can verify how a model was trained, not just inspect final weights.
Scope: the official site announces open-1b/REE/open-8b and a reproducible execution environment; it is a research/production platform, not a generic GPU rental marketplace.
Open: how deep auditability covers data, code, randomness and hardware; the cost of large runs; and collusion between trainers and verifiers are not fully answered in the public material.
关键参数Key specs
- 协议状态 / status
- 已部署
- 可审计对象 / audit
- 训练过程 + 执行环境
- 公开模型 / models
- open-1b / open-8b
- 执行环境 / REE
- Reproducible Execution Environment
官方原文怎么说What the source says
open-1b: the first model you don’t have to trust
Introducing REE: Reproducible Execution Environment
open-8b: an auditable model, in training
资料与许可Source and licence
- 资料时间Source checked
- 2026-11-19
- 许可 / 类型Licence / kind
- official-specification
- 许可说明Licence evidence
- https://www.gensyn.ai/
- 类目Category
- 去中心化推理Decentralised inference