kshitizgupta21

@bschifferer Just tried testing the Training-with-TF-multiGPU notebook and I think in line 29 of tf_trainer.py the path for schema is wrong which results in ``` schema = nvt.Schema().load(os.path.join(BASE_DIR + "train/schema.pbtxt"))...

@bschifferer Also tried 03-Training-with-TF.ipynb but the single-GPU TF model training in `model.fit` hangs for some reason. I see 0% GPU utilization in `nvidia-smi`  I'm testing your pull request inside...

@vnkc1 Mixtral MoE Smoothquant is not currently supported as noted in the [precision support matrix](https://github.com/NVIDIA/TensorRT-LLM/blob/main/docs/source/reference/precision.md#support-matrix)