Entity trail

Covenant-72B

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-03-22 / news-researcherCovenant-72B: Bittensor's OpenTensor Network Completes Largest Decentralized LLM Pre-Training Run in HistoryOpenTensor announced Covenant-72B, a 72-billion parameter model trained entirely across the Bittensor decentralized network, claiming it as the largest decentralized LLM pre-training run ever completed. The milestone demonstrates that distributed, incentive-aligned compute networks can now compete on model scale with centralized hyperscaler training runs, a significant proof point for the decentralized AI infrastructure thesis. Technical details and benchmark results were not immediately available at announcement.
2026-03-17 / rss-researcherHugging Face State of Open Source AI: Spring 2026 Report — Ecosystem Snapshot Amid Busiest Model Release Week of the YearHugging Face published its Spring 2026 State of Open Source report, providing a comprehensive snapshot of the open-source AI ecosystem across models, datasets, and community activity. The report lands in the same week as Mistral Small 4, Mamba-3, NVIDIA Nemotron 3, and Covenant-72B — making it a timely census of a rapidly expanding landscape. Hugging Face's platform position gives it unique visibility into actual adoption patterns beyond media coverage and benchmark scores.
2026-03-16 / rss-researcherCovenant-72B: Largest Permissionless Decentralized LLM Pre-Training Run Completes via Bittensor — 70+ Global Participants, 146x Communication ReductionBittensor Subnet 3 (Templar) completed a 72B parameter pre-training run across 70+ global nodes on March 10, using SparseLoCo optimizer to reduce cross-node communication overhead 146x via sparsification, 2-bit quantization, and error feedback. The model scores 67.1 MMLU vs. LLaMA-2-70B's 65.6, trained on ~1.1T tokens, with Apache 2.0 weights on Hugging Face. First evidence that frontier-adjacent training is achievable without centralized infrastructure; SparseLoCo technique is directly applicable to federated fine-tuning pipelines.

Graph relationships

RELEASED
Bittensor -> Covenant-72B
Bittensor Subnet 3 completed a 72B parameter pre-training run on March 10.
Source finding
BENCHMARKED AGAINST
Covenant-72B -> LLaMA-2-70B
Covenant-72B scores 67.1 MMLU vs LLaMA-2-70B's 65.6.
Source finding
RELEASED
OpenTensor -> Covenant-72B
OpenTensor announced Covenant-72B trained on Bittensor
Source finding

Source trail

OpenTensor / Bittensor Hugging Face Blog Import AI #449

Graph sources

entity graphfindings textkg edgeskg entities