microsoft/qlib · error · ValueError

model is not fitted yet!

Error message

model is not fitted yet!

What it means

Raised by GRUModel.predict when called before fit() completed. The `fitted` flag is set True only at the end of a successful fit; predict checks it first because the GRU weights are otherwise randomly initialized and predictions would be meaningless.

Source

Thrown at qlib/contrib/model/pytorch_gru.py:294

                        self.logger.info("early stop")
                        break

        self.logger.info("best score: %.6lf @ %d" % (best_score, best_epoch))
        self.gru_model.load_state_dict(best_param)
        torch.save(best_param, save_path)

        # Logging
        rec = R.get_recorder()
        for k, v_l in evals_result.items():
            for i, v in enumerate(v_l):
                rec.log_metrics(step=i, **{k: v})

        if self.use_gpu:
            torch.cuda.empty_cache()

    def predict(self, dataset: DatasetH, segment: Union[Text, slice] = "test"):
        if not self.fitted:
            raise ValueError("model is not fitted yet!")

        x_test = dataset.prepare(segment, col_set="feature", data_key=DataHandlerLP.DK_I)
        index = x_test.index
        self.gru_model.eval()
        x_values = x_test.values
        sample_num = x_values.shape[0]
        preds = []

        for begin in range(sample_num)[:: self.batch_size]:
            if sample_num - begin < self.batch_size:
                end = sample_num
            else:
                end = begin + self.batch_size

            x_batch = torch.from_numpy(x_values[begin:end]).float().to(self.device)

            with torch.no_grad():
                pred = self.gru_model(x_batch).detach().cpu().numpy()

View on GitHub (pinned to 79633dd950)

Solutions

  1. Call fit() to completion before predict().
  2. Persist the trained state (torch save_path) and reload the weights, then set model.fitted = True before predict.
  3. Wrap fit in code that aborts the pipeline on failure so predict is never reached unfitted.

Example fix

# before
model = GRUModel(**params)
model.predict(dataset)  # ValueError

# after
model = GRUModel(**params)
model.fit(dataset)
model.predict(dataset)
Defensive patterns

Strategy: validation

Validate before calling

assert model.fitted, "fit the model or load a checkpoint before predict"

Type guard

def is_fitted(model) -> bool:
    return bool(getattr(model, "fitted", False))

Try / catch

try:
    model.predict(dataset)
except ValueError as e:
    if "not fitted" in str(e):
        model.fit(dataset)
        model.predict(dataset)
    else:
        raise

Prevention

When it happens

Trigger: model.predict(dataset) on a fresh GRUModel; predict after a fit() that raised earlier (so fitted stayed False); using a new model instance in an inference-only script.

Common situations: Two-stage workflows (train script, then predict script) that re-instantiate the model instead of restoring it; exception swallowing in a training loop letting the code proceed to backtest.

Related errors

AI-assisted analysis of microsoft/qlib@79633dd950 (2026-08-15). Data as JSON: /api/errors/62adf267835c94e4. Report an issue: GitHub.