NVIDIA/TensorRT-Model-Connect

Parent epic: #1430 This issue groups the N–Z family migration issues under the epic's native sub-issue hierarchy. GitHub limits a parent to 100 direct sub-issues, while this migration contains 112 family issues. Each nested issue is still scoped to one family. The epic's child-issue checklist records the current Not started / In progress status and links existing migration PRs. This grouping issue has no implementation PR or separate acceptance criteria; close it when its family sub-issues are complete.

Parent epic: #1430 This issue groups the A–M family migration issues under the epic's native sub-issue hierarchy. GitHub limits a parent to 100 direct sub-issues, while this migration contains 112 family issues. Each nested issue is still scoped to one family. The epic's child-issue checklist records the current Not started / In progress status and links existing migration PRs. This grouping issue has no implementation PR or separate acceptance criteria; close it when its family sub-issues are complete.

Summary Create one polished demonstration that communicates the core TensorRT Model Connect experience: Build a supported model. Run the generated bundle. Produce an observable result. Tasks Select and pin a representative model and checkpoint with reliable setup, build, and runtime behavior. Define the supported hardware, operating system, TensorRT Model Connect revision, and dependency versions. Keep the primary path to the minimum required commands. Show the expected bundle artifact and observable inference output. Demonstrate that the deployed native runtime does not require Python or PyTorch. Include approximate setup, build, and inference times while identifying them as hardware-dependent. Provide copyable commands and troubleshooting guidance for common failures. Prepare both a live-demo path and prerecorded fallback assets. Validate the complete workflow from a clean environment. Add an automated smoke check for the documented command and output contract where prac

This was generated by AI during triage. Request type Future model support consideration Current status Wan 3.0 has been released as a hosted product/API, but its model weights have not been open-sourced. TensorRT-Model-Connect should not begin implementation based on unofficial or unverifiable artifacts. Reconsider adding Wan 3.0 only after Alibaba/Wan-AI publishes official downloadable weights and the material needed to reproduce the reference behavior. Problem and use case Alibaba released Wan 3.0 as its latest video-generation model after a public preview that began on August 6, 2026. TensorRT-Model-Connect currently lists Wan 2.1 T2V and Wan 2.2 TI2V support, but it has no Wan 3.0 workload or model family coverage. Wan 3.0 is currently available through Alibaba Cloud Model Studio under the wan3.0-video API. At the time this issue was filed, the official Wan GitHub organization and Wan-AI model hubs did not provide open-source Wan 3.0 model weights, a model card, or a reference imp

Parent Inspired by #963, which identified the same dependency-reimplementation pattern in CLI configuration parsing. This is a separate scope and does not replace or modify #963. What to build Use the repository's existing nlohmann/json dependency for dataset JSONL input and for CLI and benchmark JSONL output. Remove the independent field extractors, unescape routines, escape routines, and stream-built JSON objects while preserving the existing record schemas and per-line processing behavior. Acceptance criteria Dataset records are decoded with nlohmann/json and retain the current required/optional field behavior. CLI and benchmark result records are constructed as JSON values and serialized by the library. The local JSON escape, unescape, integer-field, and string-field implementations are removed. Tests cover quotes, backslashes, control characters, Unicode, missing fields, wrong field types, and malformed JSONL records. Existing output field names and value types remain compat