{"adoption": {"forks": 101, "observed_at": "2026-08-28T04:03:22.718030+00:00", "stars": 1049}, "canonical_url": "https://ross.abutalabs.com/products/ring-flash-attention", "card": {"archived": false, "artifact_type": "library", "description": "Ring attention implementation with flash attention", "domain": ["deep-learning", "large-language-models", "gpu-computing"], "enriched": true, "function": ["machine-learning", "llm-training", "gpu-computing"], "health_score": 60, "homepage": null, "language": "Python", "license": "MIT", "license_family": "permissive", "maturity": "active", "member_repos": ["zhuzilin/ring-flash-attention"], "name": "zhuzilin/ring-flash-attention", "platform": ["python"], "pushed_at": "2025-09-10T11:55:45+00:00", "repo": "zhuzilin/ring-flash-attention", "stars": 1049, "tags": ["ring-attention", "flash-attention", "context-parallelism", "distributed-training", "transformers", "long-context", "varlen", "huggingface", "gpu", "linux"], "topics": [], "urls": [], "use_cases": ["train LLMs with sequences longer than fit on one GPU", "enable context parallelism in HuggingFace training runs", "pack variable-length samples with ring attention", "use llama3-style context parallelism for varlen training", "reduce memory overhead of attention across multiple GPUs", "substitute flash attention in transformers models with ring attention"], "what_it_is": "A Python library implementing RingAttention on top of FlashAttention for distributed long-context transformer training. It provides varlen and batch APIs, zigzag/stripe variants, Llama3-style context parallelism, and a HuggingFace model adapter.", "when_to_avoid": ["single-GPU training where flash attention alone suffices", "inference-only workloads without distributed setup", "non-CUDA or non-NCCL environments"], "when_to_choose": ["you need to train or fine-tune transformers on very long sequences across multiple GPUs", "you want drop-in ring/context-parallel attention compatible with flash_attention_2 and HuggingFace models", "you need varlen (packed samples) support with low memory overhead"]}, "data_as_of": "2026-08-30T08:39:29.467469+00:00", "members": [{"path": "/products/ring-flash-attention", "repo": "zhuzilin/ring-flash-attention", "role": "main", "score": 44}], "provenance": {"archived": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "artifact_type": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "description": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "domain": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "enriched": {"inputs": [], "kind": "computed", "method": "enrichment_status"}, "function": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "health_score": {"inputs": ["days_since_push", "days_since_release", "archived"], "kind": "computed", "method": "health_v1"}, "homepage": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "language": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "license": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "license_family": {"inputs": ["license"], "kind": "computed", "method": "license_family"}, "maturity": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "member_repos": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "name": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "platform": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "pushed_at": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "repo": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "stars": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "tags": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "topics": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "urls": {"kind": "observed", "observed_at": "2026-08-28T04:03:22.718030+00:00", "source": "github"}, "use_cases": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "what_it_is": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "when_to_avoid": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}, "when_to_choose": {"confidence": null, "enriched_at": "2026-08-30T07:00:34.243085+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "70d29a2e5095710c1417982bbaf154d4a8e412da9ed77a39ecbe2587783b27bc", "fetched_at": "2026-08-28T04:03:22.718030+00:00", "kind": "readme", "missing": false, "url": "https://github.com/zhuzilin/ring-flash-attention"}], "taxonomy_version": 1}}, "score": {"components": {"activity": 41, "longevity": 66, "rhythm": 35}, "computed_at": "2026-09-02T17:46:02.011165+00:00", "flags": [], "formula": "round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)", "inputs": {"age_days": 924, "days_push": 357, "days_rel": 357, "gap_med": 42, "n_releases_24m": 8}, "score": 44, "version": 2}, "staleness": {"enrichment_outdated": false, "low_confidence": false, "scrape_days": 9, "stale_scrape": false}}