refactor: separate old policy and ref model in GRPO strategy

- Split single ref_model into old_model (importance sampling ratio) and ref_model (frozen KL regularizer)
- Move ref_model/old_model creation from strategy __init__ to TrainContextBuilder, pass as explicit parameters
- Remove periodic sync_ref_model + sync_interval; add sync_old_model for external rollout loop to call
- DPOStrategy also receives ref_model from builder
- Fix std to use unbiased=False (population std per GRPO paper)
- Remove redundant tests (test_grpo_kl_zero_at_init, test_grpo_no_sync_interval_param)
- Remove --grpo_sync_interval CLI arg
This commit is contained in:
2026-07-14 20:03:45 +08:00
parent 3e0007fc91
commit 2c7a71a9c0
4 changed files with 80 additions and 58 deletions
+35 -25
View File
@@ -56,19 +56,32 @@ def _make_batch(
}
def _make_frozen_copy(model, device):
"""Create a frozen copy of ``model`` with independent weights loaded."""
config = _make_config()
copy = AutoRegressiveLM(config).to(device=device)
copy.load_state_dict(model.state_dict())
copy.requires_grad_(False)
copy.eval()
return copy
@pytest.fixture
def grpo_strategy():
"""Build a GRPOStrategy with a small real model and fake executor."""
device = "cuda" if torch.cuda.is_available() else "cpu"
model, config = _make_model(device)
old_model = _make_frozen_copy(model, device)
ref_model = _make_frozen_copy(model, device)
strategy = GRPOStrategy(
model=model,
device=device,
old_model=old_model,
ref_model=ref_model,
clip_eps=0.2,
kl_coef=0.01,
group_size=4,
sync_interval=200,
model_fn=lambda c=config: AutoRegressiveLM(c).to(device=device),
executor=_FakeExecutor(),
)
@@ -108,6 +121,16 @@ def test_grpo_ref_model_not_updated(grpo_strategy):
assert p.grad is None
def test_grpo_old_model_not_updated(grpo_strategy):
"""Backward should not populate gradients on old_model."""
strategy, device = grpo_strategy
batch = _make_batch(device=device)
loss = strategy.compute_loss(batch)
loss.backward()
for p in strategy.old_model.parameters():
assert p.grad is None
def test_grpo_prompt_tokens_masked(grpo_strategy):
"""When only prompt-equivalent tokens are unmasked (response mask all 0),
the policy loss should be zero (no valid tokens contribute)."""
@@ -127,45 +150,32 @@ def test_grpo_identical_rewards_zero_advantage(grpo_strategy):
batch = _make_batch(device=device)
batch["rewards"] = torch.ones(batch["rewards"].shape, device=device)
loss = strategy.compute_loss(batch)
# At init policy == ref, so ratio == 1, KL == 0; advantage == 0.
# At init policy == old == ref, so ratio == 1, KL == 0; advantage == 0.
assert loss.item() == pytest.approx(0.0, abs=1e-5)
def test_grpo_kl_zero_at_init(grpo_strategy):
"""At initialization policy == ref_model, so KL penalty must be 0."""
def test_grpo_sync_old_model(grpo_strategy):
"""sync_old_model copies current policy weights into old_model."""
strategy, device = grpo_strategy
batch = _make_batch(device=device)
# Make rewards distinct so advantages are non-zero (isolates KL term).
loss = strategy.compute_loss(batch)
# KL term is 0 at init; loss is purely policy surrogate.
# With ratio==1, surr1==surr2==advantage, so policy_loss = -mean(|adv|).
# Just assert KL portion is negligible by checking loss is finite and
# re-running after a model update increases loss magnitude.
assert torch.isfinite(loss).item()
def test_grpo_sync_ref_model(grpo_strategy):
"""sync_ref_model copies current policy weights into ref_model."""
strategy, device = grpo_strategy
# Perturb policy model so it differs from ref.
# Perturb policy model so it differs from old.
with torch.no_grad():
for p in strategy.model.parameters():
p.add_(0.05)
# ref_model should still hold original weights (differ from policy).
# old_model should still hold original weights (differ from policy).
policy_sd = strategy.model.state_dict()
ref_sd = strategy.ref_model.state_dict()
old_sd = strategy.old_model.state_dict()
differs_before = any(
not torch.allclose(policy_sd[k], ref_sd[k]) for k in policy_sd if k in ref_sd
not torch.allclose(policy_sd[k], old_sd[k]) for k in policy_sd if k in old_sd
)
assert differs_before
strategy.sync_ref_model()
strategy.sync_old_model()
ref_sd_after = strategy.ref_model.state_dict()
old_sd_after = strategy.old_model.state_dict()
matches = all(
torch.allclose(policy_sd[k], ref_sd_after[k])
torch.allclose(policy_sd[k], old_sd_after[k])
for k in policy_sd
if k in ref_sd_after
if k in old_sd_after
)
assert matches