Le Tuan Thanh issues

Results 4 issues of


                                            Le Tuan Thanh

How to construct the global graph of all relations from Wikidata?

Could you tell me about the detail of the way you build the global graph of all relations from Wikidata?

Can not change max_input_len of Encoder while building engine in Encoder-Decoder model (T5)

I built the engines for T5 model with the following scripts for the latest version of TensorRT-LLM: ``` export MODEL_DIR="path_to_t5_model" # or "flan-t5-small" export MODEL_NAME="t5model" export MODEL_TYPE="t5" export INFERENCE_PRECISION="float16" export...

Enc-Dec C++ Runtime Paged KV - Inflight Batching output junks while inference with multiple input texts

I try inference my T5 model with C++ runtime used Paged KV at the commit `b777bd64750abf30ca7eda48e8b6ba3c5174aafd`. Its result is normal when inference with single input text, but with multiple input...

triaged

Investigating

[Encoder-Decoder] LoRA - BART not working - LoraParams and input dims don't match, lora tokens 1 input tokens 0

* I am following the example of BART LoRA at [here](https://github.com/NVIDIA/TensorRT-LLM/tree/main/examples/enc_dec#run-bart-with-lora). I cannot convert the engines without flag `--context_fmha disable`. I change `test_remove_padding = False` then running `run.py` raises the...