Search results

[Feature]: AutoDeploy: restructure unit tests

AutoDeploy<NV> AutoDeploy Backend<NV> AutoDeploy Backend

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

Task

#11166 In NVIDIA/TensorRT-LLM;

· lucaslie opened on Jan 31, 2026

[Bug]: AutoDeploy: triton attention produces gibberish output for non-greedy sampling

AutoDeploy<NV> AutoDeploy Backend<NV> AutoDeploy Backend

bugSomething isn't workingSomething isn't working

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Triton backend<NV>Related to NVIDIA Triton Inference Server backend<NV>Related to NVIDIA Triton Inference Server backend

Status: Open.

Bug

#11154 In NVIDIA/TensorRT-LLM;

· lucaslie opened on Jan 30, 2026

[Feature]: Set up perf benchmark for ops

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

#11151 In NVIDIA/TensorRT-LLM;

· nvchenghaoz opened on Jan 30, 2026

[Feature]: Better structure the custom op folder / tests

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

#11148 In NVIDIA/TensorRT-LLM;

· nvchenghaoz opened on Jan 30, 2026

[Feature]: fused_add_rms_norm Torch / Triton implementation

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Pytorch<NV>Pytorch backend related issues<NV>Pytorch backend related issues

Triton backend<NV>Related to NVIDIA Triton Inference Server backend<NV>Related to NVIDIA Triton Inference Server backend

Status: Open.

#11147 In NVIDIA/TensorRT-LLM;

· nvchenghaoz opened on Jan 30, 2026

[Feature]: Attention Support Coverage for Triton / Torch backend

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Pytorch<NV>Pytorch backend related issues<NV>Pytorch backend related issues

Triton backend<NV>Related to NVIDIA Triton Inference Server backend<NV>Related to NVIDIA Triton Inference Server backend

Status: Open.

#11146 In NVIDIA/TensorRT-LLM;

· nvchenghaoz opened on Jan 30, 2026

[Feature]: Triton backend for MOE

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Triton backend<NV>Related to NVIDIA Triton Inference Server backend<NV>Related to NVIDIA Triton Inference Server backend

Status: Open.

#11145 In NVIDIA/TensorRT-LLM;

· nvchenghaoz opened on Jan 30, 2026

[Feature]: Triton Linear Layers / Quant Op support

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Triton backend<NV>Related to NVIDIA Triton Inference Server backend<NV>Related to NVIDIA Triton Inference Server backend

Status: Open.

#11144 In NVIDIA/TensorRT-LLM;

· nvchenghaoz opened on Jan 30, 2026

[Feature]: Support chunked attention for Llama3 models

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

#11123 In NVIDIA/TensorRT-LLM;

· langzhao-netizen opened on Jan 29, 2026

[Feature]: Support for Qwen3-TTS (Multi-Codebook Audio Generation)

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

#11118 In NVIDIA/TensorRT-LLM;

· Saeedmatt3r opened on Jan 29, 2026

[Feature]: AutoDeploy: Deepseek-R1 accuracy test + verification

AutoDeploy<NV> AutoDeploy Backend<NV> AutoDeploy Backend

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

Task

#11117 In NVIDIA/TensorRT-LLM;

· lucaslie opened on Jan 29, 2026

[Feature]: AutoDeploy: MLA sharding revisited

AutoDeploy<NV> AutoDeploy Backend<NV> AutoDeploy Backend

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

Task

#11116 In NVIDIA/TensorRT-LLM;

· lucaslie opened on Jan 29, 2026