Kumar Deepak

Not able to export ONNX from llama2-7b 8 bit quantized pytorch model

4

it is a torch.onnx.export issue. Filed here to track it. TBD: File on PyTorch. STR: Set in your shell: export HF_HOME=/HF_HOME Save following as model.py: ``` import sys, argparse import...

Fx importer does not support bfloat16

1

**Error seen:** ``` Traceback (most recent call last): File "/proj/gdba/kumar/nod/SHARK-TestSuite/e2eshark/t-r-bf16-direct-fx-importer/pytorch/combinations/mlp/runmodel.py", line 131, in torch_mlir_model = export_and_import(model, test_input) File "/proj/gdba/kumar/nod/SHARK-TestSuite/e2eshark/t-r-bf16-direct-fx-importer/pytorch/combinations/mlp/runmodel.py", line 59, in export_and_import fx_importer.import_frozen_exported_program(prog) File "/proj/gdba/kumar/anaconda3/envs/e2e/lib/python3.10/site-packages/torch_mlir/extras/fx_importer.py", line 351, in import_frozen_exported_program...

Save following as model.py ``` module { func.func @main_graph(%arg0: !torch.vtensor) -> !torch.vtensor attributes {torch.onnx_meta.ir_version = 8 : si64, torch.onnx_meta.opset_version = 17 : si64, torch.onnx_meta.producer_name = "pytorch", torch.onnx_meta.producer_version = "2.3.0"} {...

Importing an ONNX OPT model runs into error while lowering to torch MLIR

1

Import of ONNX model into torch MLIR using onnx importer is successful. But hits an error while lowering to torch MLIR: But, now I hit opt-125M.fp32.torch-onnx.mlir:244:12: error: failed to legalize...