google-research/timesfm · error · ValueError

Continuous quantile head is not supported for horizons > {se

Error message

Continuous quantile head is not supported for horizons > {self.model.os}.

What it means

When forecast_config.use_continuous_quantile_head is enabled, quantiles can only be produced for horizons up to self.model.os (derived from output_quantile_len=1024 / output_patch_len=128, i.e. 8 output patches). A larger max_horizon with the continuous quantile head raises ValueError during compile().

Source

Thrown at src/timesfm/timesfm_2p5/timesfm_2p5_torch.py:416

        new_context := math.ceil(fc.max_context / self.model.p) * self.model.p,
      )
      fc = dataclasses.replace(fc, max_context=new_context)
    if fc.max_horizon % self.model.o != 0:
      logging.info(
        "When compiling, max horizon needs to be multiple of the output patch"
        " size %d. Using max horizon = %d instead.",
        self.model.o,
        new_horizon := math.ceil(fc.max_horizon / self.model.o) * self.model.o,
      )
      fc = dataclasses.replace(fc, max_horizon=new_horizon)
    if fc.max_context + fc.max_horizon > self.model.config.context_limit:
      raise ValueError(
        "Context + horizon must be less than the context limit."
        f" {fc.max_context} + {fc.max_horizon} >"
        f" {self.model.config.context_limit}."
      )
    if fc.use_continuous_quantile_head and (fc.max_horizon > self.model.os):
      raise ValueError(
        f"Continuous quantile head is not supported for horizons > {self.model.os}."
      )
    self.forecast_config = fc

    def _compiled_decode(horizon, inputs, masks):
      if horizon > fc.max_horizon:
        raise ValueError(
          f"Horizon must be less than the max horizon. {horizon} > {fc.max_horizon}."
        )

      inputs = (
        torch.from_numpy(np.array(inputs)).to(self.model.device).to(torch.float32)
      )
      masks = torch.from_numpy(np.array(masks)).to(self.model.device).to(torch.bool)
      batch_size = inputs.shape[0]

      if fc.infer_is_positive:
        is_positive = torch.all(inputs >= 0, dim=-1, keepdim=True)

View on GitHub (pinned to 331c6d33cb)

Solutions

1. Lower max_horizon to <= self.model.os (e.g. 1024) when use_continuous_quantile_head=True.
2. Set use_continuous_quantile_head=False if horizons beyond os are required.
3. Forecast iteratively: compile within the supported range and roll predictions forward.
4. Check TimesFM_2p5_200M_Definition.output_quantile_len and output_patch_len to compute the valid horizon cap.

Example fix

// before
fc = ForecastConfig(max_horizon=2048, use_continuous_quantile_head=True)  # ValueError
// after
fc = ForecastConfig(max_horizon=1024, use_continuous_quantile_head=True)
model.compile(fc)

os_cap = 1024 // 128  # output_quantile_len / output_patch_len = 8
if fc.use_continuous_quantile_head and fc.max_horizon > os_cap * 128:
    fc = dataclasses.replace(fc, max_horizon=os_cap * 128, use_continuous_quantile_head=False)

try:
    model.compile(fc)
except ValueError as e:
    if "Continuous quantile head" in str(e):
        fc = dataclasses.replace(fc, use_continuous_quantile_head=False)
        model.compile(fc)
    else:
        raise

When it happens

Trigger: model.compile(ForecastConfig(use_continuous_quantile_head=True, max_horizon=N)) where N exceeds self.model.os — beyond roughly 1024 forecast steps for the 2.5 200M model.

Common situations: Enabling continuous quantiles with a very long horizon (e.g. 2048) for long-range forecasting; mixing the quantile-head flag with a horizon tuned for the default quantile path.

Related errors

• Context + horizon must be less than the context limit. {fc.m
• Horizon must be less than the max horizon. {horizon} > {fc.m
• Activation: {config.activation} not supported.
• Output dims must be a multiple of 4: {config.output_dims} %
• Memory dimension ({self.in_features}) must be divisible by '

AI-assisted analysis of google-research/timesfm@331c6d33cb (2026-08-29). Data as JSON: /api/errors/d052208b79f48181. Report an issue: GitHub.