{"slug":"graniet-grpo-rl-training","source_name":"graniet/grpo-rl-training","name":"Graniet/Grpo Rl Training","description":"Expert guidance for GRPO/RL fine-tuning with TRL for reasoning and task-specific model training","version":1,"lift":{"pass_rate_delta_pts":13.64,"pass_rate_pct":100,"total_cases":22,"passed_cases":22,"tokens_delta_pct":178.4,"turns_delta_pct":0,"verdict":"pass","benchmark_model":"gemini-3.6-flash","grading_method":"judged","completed_at":"2026-08-23T05:20:31.212200+00:00"},"skill_score":1,"benchmark_models":[{"model":"gemini-3.6-flash","headline":true,"delta_pts":13.64,"with_pass_pct":100,"without_pass_pct":86.4,"tokens_delta_pct":178.4,"turns_delta_pct":0,"total_cases":22,"cases_aggregated":21,"verdict":"pass","never_hurt":true,"completed_at":"2026-08-23T05:20:31.212200+00:00","run_id":"00d33988-a5f2-43fb-9868-e0874f900472","latest_attempt":{"id":"00d33988-a5f2-43fb-9868-e0874f900472","runner_model":"gemini-3.6-flash","overall_verdict":"pass","completed_at":"2026-08-23T05:20:31.212200+00:00","gate":null},"version_number":1,"is_latest_version":true,"gate":null}],"trust":{"skill_safety":"passed","safety_status":"clean","intent_verdict":"safe","content_status":"clean","indexable":true},"license":"MIT","install_count":0,"total_uses":0,"total_users":0,"manifest_hash":"481a4169cf82f6487d517905a7642b41c37a0d524ef8aeccfecef4859cedc048","install_name":"graniet-grpo-rl-training","version_id":"82b13e3f-66c1-429e-a428-5fba48131b66","body_sha256":"481a4169cf82f6487d517905a7642b41c37a0d524ef8aeccfecef4859cedc048","render_hash":"5e47200f9f726d3d798719249a25dc6279742673d1485b1aefd40cbbe5d94d42","raw_url":"https://app.decimal.ai/s/graniet-grpo-rl-training/SKILL.md","scorecard_url":"https://app.decimal.ai/skills/graniet-grpo-rl-training"}