{"slug": "crowdtensor-volunteer-training-beta-and-draft-qwen2-5-7b-gsm8k-campaign-rfc", "title": "CrowdTensor volunteer-training Beta and Draft Qwen2.5-7B GSM8K Campaign RFC", "summary": "CrowdTensor, an Apache-2.0 open-source protocol for checkpointed volunteer model-training campaigns, completed a feasibility run using Qwen/Qwen2.5-7B-Instruct and openai/gsm8k, achieving an accuracy improvement from 71.875% to 74.219%. The preregistered threshold was passed, but no statistical significance or broad-capability claim is made. A smaller Founding SmolLM2/WikiText-2 campaign now serves live aggregate status, though the project still lacks independent physical multi-host evidence and permissionless adversarial safety.", "body_md": "CrowdTensor is an Apache-2.0 open-source protocol for checkpointed volunteer model-training Campaigns. A Campaign pins immutable model/data revisions and gives admitted CPU/GPU/TPU Cells bounded work. Accepted LoRA deltas advance one auditable checkpoint; contributor disappearance pauses or reassigns work instead of discarding committed progress.\n\nThe completed feasibility run used:\n\n`Qwen/Qwen2.5-7B-Instruct@a09a35458c702b33eeacc393d103063234e8bc28`\n\n`openai/gsm8k@740312add88f781978c0658806c59bc2815b9866`\n\n`71.875% -> 74.219%`\n\nThe practical preregistered threshold passed; the paired bootstrap interval included zero, so no statistical-significance or broad-capability claim is made.\n\nA smaller Founding SmolLM2/WikiText-2 Campaign now serves live aggregate status. Two maintainer-operated private Kaggle GPU Cells seeded its first round through the public HTTPS contribution path. Enrollment remains controlled, and the project still lacks independent physical multi-host evidence and permissionless adversarial safety.\n\nI would value review of the Draft 7B RFC, especially the fresh-holdout design, the 256-to-1,024-step extension rule, quantized 16 GB GPU work units, delta validation, and maintainer/rollback policy.", "url": "https://wpnews.pro/news/crowdtensor-volunteer-training-beta-and-draft-qwen2-5-7b-gsm8k-campaign-rfc", "canonical_source": "https://discuss.huggingface.co/t/crowdtensor-volunteer-training-beta-and-draft-qwen2-5-7b-gsm8k-campaign-rfc/178211#post_1", "published_at": "2026-07-26 07:18:33+00:00", "updated_at": "2026-07-26 07:33:13.630923+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-research", "ai-infrastructure"], "entities": ["CrowdTensor", "Qwen/Qwen2.5-7B-Instruct", "openai/gsm8k", "SmolLM2", "WikiText-2", "Kaggle"], "alternates": {"html": "https://wpnews.pro/news/crowdtensor-volunteer-training-beta-and-draft-qwen2-5-7b-gsm8k-campaign-rfc", "markdown": "https://wpnews.pro/news/crowdtensor-volunteer-training-beta-and-draft-qwen2-5-7b-gsm8k-campaign-rfc.md", "text": "https://wpnews.pro/news/crowdtensor-volunteer-training-beta-and-draft-qwen2-5-7b-gsm8k-campaign-rfc.txt", "jsonld": "https://wpnews.pro/news/crowdtensor-volunteer-training-beta-and-draft-qwen2-5-7b-gsm8k-campaign-rfc.jsonld"}}