TensorRT-LLMs/cpp/tensorrt_llm
Robin Kobus b5bc0a9fcd
chore: Add output of first token to additional generation outputs (#3205)
- Updated the first dimension of additional output tensors to match mMaxNewTokens.
- Copy output of last context token to generation outputs.
- Adjusted the expected output size calculations in unit tests to reflect the correct maximum output length.

Signed-off-by: Robin Kobus <19427718+Funatiq@users.noreply.github.com>
2025-04-02 20:14:16 +08:00
..
batch_manager chore: Add output of first token to additional generation outputs (#3205) 2025-04-02 20:14:16 +08:00
common fix: segfault in cudaDriverWrapper (#3017) 2025-04-02 08:55:19 +02:00
cutlass_extensions/include/cutlass_extensions feat: Update cutlass (#2981) 2025-03-26 22:36:27 +08:00
executor feat: Variable-Beam-Width-Search (VBWS) Part2 (#3133) 2025-04-02 12:31:28 +08:00
executor_worker Update TensorRT-LLM (#2792) 2025-02-18 21:27:39 +08:00
kernels fix: segfault in cudaDriverWrapper (#3017) 2025-04-02 08:55:19 +02:00
layers feat: Variable-Beam-Width-Search (VBWS) Part2 (#3133) 2025-04-02 12:31:28 +08:00
plugins perf: Add optimizations for deepseek in min latency mode (#3093) 2025-04-02 09:05:24 +08:00
pybind Revert "refactor: Replace DecoderFinishedEvent with CudaEvent in decoder clas…" (#3183) 2025-04-01 12:49:27 +08:00
runtime refactor: Simplify disableLookahead and improve numDecodingEngineTokens handling (#3103) 2025-04-01 18:47:31 +08:00
thop [feat] open source fp8_blockscale_gemm (#3071) 2025-04-02 12:12:52 +08:00
CMakeLists.txt [feat] open source fp8_blockscale_gemm (#3071) 2025-04-02 12:12:52 +08:00