ampere_model_library

ampere_model_library copied to clipboard

Add Unet KITS19 in PyTorch

2

MarcelWilnicki

jan-grzybek-ampere

kkontny

Adding fp16 mode whisper

1

Adds support for naitive FP16 in whisper OpenAI model.

kkontny

Save subcategory results to json

1

dkupnicki

Llama gptq int 8 models 7&13 b

kkontny

add gpt2 model

1

TORCH_COMPILE=1 AIO_NUM_THREADS=80 numactl --membind=0 --cpubind=0 python natural_language_processing/text_generation/gpt/run.py -m gpt2 --lambada_path /ampere/aml/lambada_test_plain_text.txt --num_runs 10

MarcelWilnicki

the command: `python run.py --model_name gpt2 -f pytorch --lambada_path /ampere/aml/natural_language_processing/text_generation/nanogpt/lambada_test_plain_text.txt` results in following error: `Traceback (most recent call last): File "/ampere/mzd/ampere_model_library/utils/pytorch.py", line 65, in __init__ self._frozen_script = torch.jit.freeze(torch.jit.script(self._model), preserved_attrs=[func]) File...

MarcelWilnicki

Add fp16 PyTorch models

3

In Pytorch you can convert model to fp16 with `module.half()` call. I think it should be called before converting to TorchScript. See docs https://pytorch.org/docs/stable/generated/torch.nn.Module.html. I think it should be quite...

kkontny

Add NHWC Pytorch torchvision models

3

Since new version Torchvision supports NHWC version of the models, please add the support to model zoo, they can be enabled with some flag like `--nhwc`. Since we don't support...

kkontny