RL adapters and training artifacts for 3-stage PPO, GRPO-C, only A+C, only B+C, and GiGPO-C.
Independent publisher
Yitia
yinita
Models in Library2
Datasets in Library0
Models on Hugging Face100
Followers—
Compare the published evidence
This profile connects the 2 model and dataset records shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.
Published tasks: Text generation.
Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.
Models
Rank-16 LoRA adapters on Qwen/Qwen3.5-4B (revision 851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a), three methods: grpoabc, grpoc, grpocurriculumabc. - /checkpoint-XXXXXX/: adapter, receipt, trainingstate (every 25 GRPO groups + final) - /periodicdev/step-XXXXXX/: periodic dev evaluation (1056 cells, PENGUIN+CASTLE x 11 topologies, JEV-1.13.0 judge); summary.json + results.tar.gz Periodic dev is monitoring only; the fixed final checkpoint is the comparison endpoint. W&B: yinita/ps4mas-0922-training.