{"data":{"full_name":"IST-DASLab/marlin","name":"marlin","description":"FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.","stars":1039.0,"forks":86.0,"language":"Python","license":"Apache-2.0","archived":0.0,"subcategory":"llm-cuda-optimization","last_pushed_at":"2024-09-04T13:35:00+00:00","pypi_package":null,"npm_package":null,"downloads_monthly":0.0,"dependency_count":0.0,"commits_30d":null,"reverse_dep_count":0.0,"maintenance_score":0.0,"adoption_score":10.0,"maturity_score":16.0,"community_score":17.0,"quality_score":43.0,"quality_tier":"emerging","risk_flags":"['stale_6m', 'no_package', 'no_dependents']"},"meta":{"timestamp":"2026-04-06T18:00:03.131458+00:00"}}