{"adoption": {"forks": 174, "observed_at": "2026-09-03T02:15:06.827285+00:00", "stars": 1201}, "canonical_url": "https://ross.abutalabs.com/products/kvpress", "card": {"archived": false, "artifact_type": "library", "description": "LLM KV cache compression made easy", "domain": ["large-language-models", "deep-learning", "machine-learning"], "enriched": true, "function": ["llm-inference", "machine-learning", "benchmarking"], "health_score": 99, "homepage": null, "language": "Python", "license": "Apache-2.0", "license_family": "permissive", "maturity": "active", "member_repos": ["NVIDIA/kvpress"], "name": "NVIDIA/kvpress", "platform": ["python", "cross-platform"], "pushed_at": "2026-09-02T07:41:41+00:00", "repo": "NVIDIA/kvpress", "stars": 1201, "tags": ["kv-cache", "kv-cache-compression", "long-context", "transformers", "pytorch", "huggingface", "memory-optimization", "gpu"], "topics": ["llm", "inference", "kv-cache", "kv-cache-compression", "long-context", "python", "pytorch", "transformers", "large-language-models"], "urls": [], "use_cases": ["compress kv cache for long-context llm inference", "reduce gpu memory when serving llms with long prompts", "benchmark kv cache compression methods", "implement a custom kv cache compression method", "run 1M token contexts on limited gpu memory", "integrate kv cache compression into a transformers pipeline"], "what_it_is": "kvpress is a Python library from NVIDIA that implements multiple KV cache compression methods and benchmarks for long-context LLM inference, built on Hugging Face transformers. It provides 'presses' that compress the key-value cache during prefilling to reduce memory usage, and simplifies developing new compression methods.", "when_to_avoid": ["you need KV cache compression for non-transformer or non-Hugging Face model runtimes", "you want a production inference server rather than a compression library", "your contexts are short enough that KV cache memory is not a bottleneck"], "when_to_choose": ["you deploy or fine-tune transformer LLMs with long contexts and hit KV cache memory limits", "you research KV cache compression and want a benchmarking framework with existing methods", "you already use Hugging Face transformers and PyTorch"]}, "data_as_of": "2026-08-30T08:39:29.467469+00:00", "members": [{"path": "/products/kvpress", "repo": "NVIDIA/kvpress", "role": "main", "score": 86}], "provenance": {"archived": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "artifact_type": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "description": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "domain": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "enriched": {"inputs": [], "kind": "computed", "method": "enrichment_status"}, "function": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "health_score": {"inputs": ["days_since_push", "days_since_release", "archived"], "kind": "computed", "method": "health_v1"}, "homepage": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "language": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "license": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "license_family": {"inputs": ["license"], "kind": "computed", "method": "license_family"}, "maturity": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "member_repos": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "name": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "platform": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "pushed_at": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "repo": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "stars": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "tags": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "topics": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "urls": {"kind": "observed", "observed_at": "2026-09-03T02:15:06.827285+00:00", "source": "github"}, "use_cases": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "what_it_is": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "when_to_avoid": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}, "when_to_choose": {"confidence": null, "enriched_at": "2026-08-30T06:22:12.262196+00:00", "kind": "inferred", "prompt_version": 1, "sources": [{"content_hash": "aeef658a5dcbbb659c997fbaf35859540fd7cfd059c8e0f9965149b18a58aa4b", "fetched_at": "2026-09-03T02:15:06.827285+00:00", "kind": "readme", "missing": false, "url": "https://github.com/NVIDIA/kvpress"}], "taxonomy_version": 1}}, "score": {"components": {"activity": 100, "longevity": 47, "rhythm": 91}, "computed_at": "2026-09-03T02:20:16.233290+00:00", "flags": [], "formula": "round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)", "inputs": {"age_days": 665, "days_push": 0, "days_rel": 62, "gap_med": 13.5, "n_releases_24m": 27}, "score": 86, "version": 2}, "staleness": {"enrichment_outdated": false, "low_confidence": false, "scrape_days": 3, "stale_scrape": false}}