sgl-project/sglang · error · ValueError

DSPARK requires explicit layer_ids for aux hidden capture.

Error message

DSPARK requires explicit layer_ids for aux hidden capture.

What it means

set_dspark_layers_to_capture (kimi_linear.py:797) requires an explicit list of layer indices to capture; passing None (the sentinel meaning 'capture defaults') raises ValueError. Unlike some models, Kimi-Linear has no default capture set, so callers must name the layers.

Source

Thrown at python/sglang/srt/models/kimi_linear.py:797

                quant_config=quant_config,
                prefix=maybe_prefix(prefix, "lm_head"),
            )
        else:
            self.lm_head = PPMissingLayer()
        logit_scale = getattr(self.config, "logit_scale", 1.0)
        self.logits_processor = LogitsProcessor(config=config, logit_scale=logit_scale)
        self.capture_aux_hidden_states = False

    def get_input_embeddings(self):
        return self.model.embed_tokens

    def set_dspark_layers_to_capture(self, layer_ids: list[int]) -> None:
        if self.pp_group.world_size > 1:
            raise NotImplementedError("DSPARK aux hidden capture requires PP=1.")
        if not self.pp_group.is_last_rank:
            return
        if layer_ids is None:
            raise ValueError(
                "DSPARK requires explicit layer_ids for aux hidden capture."
            )
        self.capture_aux_hidden_states = True
        self.model.dspark_layers_to_capture = list(layer_ids)

    @torch.no_grad()
    def forward(
        self,
        input_ids: torch.Tensor,
        positions: torch.Tensor,
        forward_batch: ForwardBatch,
        inputs_embeds: Optional[torch.Tensor] = None,
        pp_proxy_tensors: Optional[PPProxyTensors] = None,
    ) -> torch.Tensor:
        hidden_states = self.model(
            input_ids,
            positions,
            forward_batch,

View on GitHub (pinned to 0132848349)

Solutions

1. Pass an explicit list, e.g. set_dspark_layers_to_capture([24, 48])
2. Default unset CLI options to a concrete layer list (or skip calling the API entirely)
3. Validate layer ids against model.config.num_hidden_layers before calling

Example fix

# before
model.set_dspark_layers_to_capture(args.capture_layers)  # None when flag omitted
# after
if args.capture_layers:
    model.set_dspark_layers_to_capture(list(args.capture_layers))

assert layer_ids is not None and len(layer_ids) > 0

def valid_capture_layers(ids) -> bool:
    return isinstance(ids, (list, tuple)) and len(ids) > 0 and all(isinstance(i, int) for i in ids)

When it happens

Trigger: Calling model.set_dspark_layers_to_capture(None) or passing an uninitialized list[int] variable that evaluates to None.

Common situations: Adapting generic training-hook code that assumes a default layer set exists; optional CLI arg for capture layers left unset and forwarded verbatim.

Understand the failure class

Background: "missing required argument" and "the following required arguments were not provided": what required-argument errors mean and how to fix them — this error's family across 20 libraries.

Related errors

• DSpark requires --speculative-dspark-block-size to be positi
• DSpark speculative_num_draft_tokens must be >= 2 (= gamma +
• --enable-linear-replayssm-spec with DSPARK/DFLASH requires a
• DSPARK requires explicit layer_ids for aux hidden capture.
• DSPARK aux hidden capture requires PP=1.

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/af2c9b2c729ef337. Report an issue: GitHub.