microsoft/qlib · error · NotImplementedError

optimizer {} is not supported!

Error message

optimizer {} is not supported!

What it means

Raised in HIST.__init__ when `optimizer` is not "adam" or "gd" (case-insensitive). HIST (graph-structured stock prediction) wires up only these two optimizers at construction; anything else raises NotImplementedError immediately.

Source

Thrown at qlib/contrib/model/pytorch_hist.py:141

        if self.seed is not None:
            np.random.seed(self.seed)
            torch.manual_seed(self.seed)

        self.HIST_model = HISTModel(
            d_feat=self.d_feat,
            hidden_size=self.hidden_size,
            num_layers=self.num_layers,
            dropout=self.dropout,
            base_model=self.base_model,
        )
        self.logger.info("model:\n{:}".format(self.HIST_model))
        self.logger.info("model size: {:.4f} MB".format(count_parameters(self.HIST_model)))
        if optimizer.lower() == "adam":
            self.train_optimizer = optim.Adam(self.HIST_model.parameters(), lr=self.lr)
        elif optimizer.lower() == "gd":
            self.train_optimizer = optim.SGD(self.HIST_model.parameters(), lr=self.lr)
        else:
            raise NotImplementedError("optimizer {} is not supported!".format(optimizer))

        self.fitted = False
        self.HIST_model.to(self.device)

    @property
    def use_gpu(self):
        return self.device != torch.device("cpu")

    def mse(self, pred, label):
        loss = (pred - label) ** 2
        return torch.mean(loss)

    def loss_fn(self, pred, label):
        mask = ~torch.isnan(label)

        if self.loss == "mse":
            return self.mse(pred[mask], label[mask])

View on GitHub (pinned to 79633dd950)

Solutions

  1. Use optimizer="adam" or optimizer="gd".
  2. Subclass HIST to install a custom torch optimizer after super().__init__.

Example fix

# before
HIST(optimizer="sgd", ...)

# after
HIST(optimizer="gd", ...)
Defensive patterns

Strategy: validation

Validate before calling

assert params["optimizer"].lower() in {"adam", "gd"}, "HIST supports only 'adam' and 'gd'"

Type guard

def is_supported_optimizer(name: str) -> bool:
    return isinstance(name, str) and name.lower() in {"adam", "gd"}

Try / catch

try:
    model = HIST(**params)
except NotImplementedError as e:
    if "optimizer" in str(e):
        params["optimizer"] = "adam"
        model = HIST(**params)
    else:
        raise

Prevention

When it happens

Trigger: HIST(optimizer="sgd"/"adamw"/..., ...) — "sgd" again being the common trap since qlib's alias is "gd". Fires during model instantiation in the workflow.

Common situations: Copying optimizer names from other models/papers; hyper-parameter sweeps including unsupported values.

Related errors

AI-assisted analysis of microsoft/qlib@79633dd950 (2026-08-15). Data as JSON: /api/errors/6369844a2f526fce. Report an issue: GitHub.