sgl-project/sglang · critical · ValueError

Kimi expert-pack index is truncated

Error message

Kimi expert-pack index is truncated

What it means

While iterating the index entries (index_count of them from the header), a read of GGML_PACK_ENTRY.size bytes came back short — the index region is truncated relative to the declared entry count. The file ends before the full index.

Source

Thrown at python/sglang/srt/layers/moe/expert_pack.py:817

            if len(raw_header) != GGML_PACK_HEADER.size:
                raise ValueError("Kimi expert-pack header is truncated")
            index_digest.update(raw_header)
            magic, version, header_size, index_count, data_start = (
                GGML_PACK_HEADER.unpack(raw_header)
            )
            if (
                magic != GGML_PACK_MAGIC
                or version != 1
                or header_size != GGML_PACK_HEADER.size
                or index_count != expected_entry_count
                or data_start != int(pack_manifest["data_start"])
            ):
                raise ValueError("Kimi expert-pack header does not match its manifest")

            for index in range(index_count):
                raw_entry = stream.read(GGML_PACK_ENTRY.size)
                if len(raw_entry) != GGML_PACK_ENTRY.size:
                    raise ValueError("Kimi expert-pack index is truncated")
                index_digest.update(raw_entry)
                name_raw, expert, reserved, offset, nbytes = GGML_PACK_ENTRY.unpack(
                    raw_entry
                )
                object_index, physical_role_id = divmod(index, len(KIMI_PHYSICAL_ROLES))
                layer_index, expected_expert = divmod(object_index, expected_experts)
                layer = active_layers[layer_index]
                physical_role = KIMI_PHYSICAL_ROLES[physical_role_id]
                name = _fixed_string(name_raw)
                match = KIMI_EXPERT_RE.fullmatch(name)
                if (
                    match is None
                    or int(match.group("layer")) != layer
                    or match.group("role") != physical_role
                    or expert != expected_expert
                    or reserved != 0
                ):
                    raise ValueError(

View on GitHub (pinned to 0132848349)

Solutions

1. Re-download/rebuild the pack; verify full file size and SHA-256 against the manifest before loading
2. If self-produced, re-run packing ensuring all 92 layers x experts x roles entries were written

min_size = data_start + index_count * ENTRY_SIZE
assert pack_path.stat().st_size >= min_size, 'index region truncated'

When it happens

Trigger: File truncated mid-index (partial download), or header index_count inflated relative to actual entries; also follows from a data_start/size mismatch upstream.

Common situations: See trigger scenarios.

Related errors

• Kimi active routed MoE layers must be exactly 1..92
• Kimi manifest expert-pack path does not match pack_path
• Kimi expert-pack size does not match its manifest
• Kimi expert-pack physical role order is unsupported
• Kimi expert-pack role sizes do not match object bytes

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/795f16e6bc5ddb94. Report an issue: GitHub.