""" GRPO / RLVR on GSM8K (DeepSeek-R1 style), from scratch -- no critic, group-relative advantages, verifiable reward. Per iteration: for each prompt sample a group of G completions, score each with the GSM8K verifier, compute group-relative advantages, and update with a token-level clipped surrogate + KL-to-reference penalty. An arithmetic warm-up curriculum runs first so the policy gets non-zero reward variance before facing full GSM8K. PYTHONPATH=. python scripts/train_grpo.py PYTHONPATH=. torchrun --standalone --nproc_per_node=2 scripts/train_grpo.py """ from __future__ import annotations import time import torch from config.post_training_config import GRPOConfig from data_loader.prompt_dataset import get_prompt_iterator from src.post_training.cli import parse_config_with_json from src.post_training.chat_template import decode, encode_prompt from src.post_training.distributed import ddp_setup, ddp_wrap, cleanup, reduce_scalar from src.post_training.evaluation import gsm8k_accuracy, load_gsm8k_eval from src.post_training.grpo import group_advantages, grpo_loss from src.post_training.logging_utils import MetricsLogger from src.post_training.optim import configure_optimizer from src.post_training.rewards import reward_gsm8k from src.post_training.rollout import compute_logprobs, rollout_prompts from src.post_training.utils import ( amp_autocast, load_backbone_from_ckpt, make_frozen_copy, save_stage_ckpt, set_seed, unwrap, ) def seq_lengths_from_mask(response_mask, prompt_lens): N, T = response_mask.shape pos = torch.arange(T, device=response_mask.device) last = torch.where(response_mask, pos[None, :], torch.full_like(response_mask, -1, dtype=torch.long)).max(dim=1).values return torch.where(last >= 0, last + 1, prompt_lens) def main(): cfg, _ = parse_config_with_json(GRPOConfig, "configs/grpo.json") ctx = ddp_setup(cfg.device) set_seed(cfg.seed + ctx.rank) policy = load_backbone_from_ckpt(cfg, cfg.sft_ckpt, ctx.device) ref = make_frozen_copy(policy, device=ctx.device) policy_ddp = ddp_wrap(policy, ctx) optimizer = configure_optimizer(unwrap(policy_ddp), cfg.lr, weight_decay=0.0) eval_set = None logger = MetricsLogger("grpo", cfg.log_dir, use_wandb=cfg.use_wandb, wandb_project=cfg.wandb_project) if ctx.is_main else None if ctx.is_main: print(f"GRPO from {cfg.sft_ckpt} | group_size={cfg.group_size} | world={ctx.world_size}") warm_it = get_prompt_iterator(cfg.curriculum_path, cfg.prompts_per_iter, rank=ctx.rank, world_size=ctx.world_size, seed=cfg.seed) main_it = get_prompt_iterator(cfg.prompt_path, cfg.prompts_per_iter, rank=ctx.rank, world_size=ctx.world_size, seed=cfg.seed) G = cfg.group_size for it in range(cfg.iterations): rows = next(warm_it if it < cfg.curriculum_iters else main_it) # Replicate each prompt G times, group-contiguously. base_prompts = [encode_prompt([{"role": "user", "content": r["prompt"]}]) for r in rows] prompts = [p for p in base_prompts for _ in range(G)] golds = [r.get("gold") for r in rows for _ in range(G)] policy.eval() with amp_autocast(cfg.amp_dtype, ctx.device): seqs, rmask, plens = rollout_prompts(policy, prompts, cfg.rollout_len, device=ctx.device, temperature=cfg.temperature, top_p=cfg.top_p if cfg.top_p < 1 else None) resp = rmask[:, 1:] seq_lens = seq_lengths_from_mask(rmask, plens) responses = [decode(seqs[i, plens[i]:seq_lens[i]].tolist()) for i in range(len(prompts))] rewards = torch.tensor([reward_gsm8k(responses[i], golds[i]) for i in range(len(prompts))], device=ctx.device, dtype=torch.float32) adv = group_advantages(rewards, G) with torch.no_grad(), amp_autocast(cfg.amp_dtype, ctx.device): old_logp, _ = compute_logprobs(policy, seqs, rmask, temperature=cfg.temperature, requires_grad=False) ref_logp, _ = compute_logprobs(ref, seqs, rmask, temperature=cfg.temperature, requires_grad=False) old_logp, ref_logp = old_logp.float(), ref_logp.float() policy.train() N = seqs.size(0) agg = {"loss": 0.0, "kl": 0.0, "clipfrac": 0.0, "n": 0} for _ in range(cfg.grpo_epochs): perm = torch.randperm(N, device=ctx.device) for s in range(0, N, max(1, G)): # minibatch ~ one group's worth mb = perm[s:s + max(1, G)] with amp_autocast(cfg.amp_dtype, ctx.device): new_logp, _ = compute_logprobs(policy_ddp, seqs[mb], rmask[mb], temperature=cfg.temperature, requires_grad=True) loss, st = grpo_loss(new_logp.float(), old_logp[mb], ref_logp[mb], adv[mb], resp[mb], clip=cfg.clip, kl_coef=cfg.kl_coef) optimizer.zero_grad(set_to_none=True) loss.backward() torch.nn.utils.clip_grad_norm_(policy_ddp.parameters(), cfg.grad_clip) optimizer.step() agg["loss"] += loss.item(); agg["kl"] += st["kl"]; agg["clipfrac"] += st["clipfrac"]; agg["n"] += 1 mean_reward = reduce_scalar(rewards.mean().item(), ctx) # Fraction of groups with non-zero reward spread (informative groups). grp_std = rewards.view(-1, G).std(dim=1) informative = reduce_scalar((grp_std > 1e-6).float().mean().item(), ctx) resp_len = reduce_scalar(resp.float().sum(1).mean().item(), ctx) n = max(1, agg["n"]) if ctx.is_main and it % 5 == 0: phase = "warmup" if it < cfg.curriculum_iters else "gsm8k" print(f"iter {it}[{phase}] | reward {mean_reward:.3f} | informative {informative:.2f} | " f"loss {agg['loss']/n:.4f} | KL {agg['kl']/n:.4f} | clipfrac {agg['clipfrac']/n:.3f} | resp_len {resp_len:.0f}") if logger: logger.log(it, {"reward": mean_reward, "informative_groups": informative, "loss": agg["loss"]/n, "kl": agg["kl"]/n, "resp_len": resp_len}) if ctx.is_main and it > 0 and it % cfg.eval_every == 0: if eval_set is None: eval_set = load_gsm8k_eval("test", limit=200) res = gsm8k_accuracy(unwrap(policy_ddp), eval_set, device=ctx.device, max_new_tokens=cfg.rollout_len) print(f" [eval] iter {it} | GSM8K test acc {res['accuracy']:.3f} ({res['correct']}/{res['n']})") if logger: logger.log(it, {"gsm8k_acc": res["accuracy"]}) if ctx.is_main and it > 0 and it % cfg.save_every == 0: save_stage_ckpt(cfg.out_ckpt, unwrap(policy_ddp), optimizer, stage="grpo", cfg=cfg, step=it, metrics={"reward": mean_reward}) if ctx.is_main: save_stage_ckpt(cfg.out_ckpt, unwrap(policy_ddp), optimizer, stage="grpo", cfg=cfg, step=cfg.iterations, metrics={}) print(f"Done GRPO -> {cfg.out_ckpt}") if logger: logger.close() cleanup(ctx) if __name__ == "__main__": main()