mirror of
https://github.com/saymrwulf/pytorch.git
synced 2026-05-15 21:00:47 +00:00
851579d868
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/34991 The definition for the partition to be run on CPU is that it will contain an empty device_id list. We chose this over an op with no partitioning info because 1. Backward compatible with models that don't have partitioning info 2. Being explicit can flush out issues in earlier stage. Test Plan: ``` LD_LIBRARY_PATH=third-party-buck/platform007/build/fb-nnpi/lib ./sigrid/predictor/tests/scripts/launch_ads_test_predictor.sh -g --nnpi --force_models=175742819_0 --sigrid_force_model_dir=$HOME/models/ --smc_server_port=7447 --glow-num-devices=1 --glow_interpreter_memory=$((256<<20)) --caffe2_fbgemm_fake_fp16_clamp --glow_global_fp16 --glow_clip_fp16 --glow_global_fused_scale_offset_fp16 --fbgemm_deserialize_to_original_format --caffe2_dump_input_of_type=Onnxifi --caffe2_logging_print_tensor --caffe2_predictor_use_memonger=no --onnxifi_debug_mode=true --caffe2_dump_input_with_recordio --caffe2_predictor_onnxifi_max_batch_size=32 --caffe2_predictor_onnxifi_max_seq_size=9600 --glow_onnxifi_backend=Interpreter --onnxifi_blacklist_ops=SparseLengthsSum,SparseLengthsWeightedSum --glow_dump_graph ``` Now it hits a new error. Reviewed By: ipiszy Differential Revision: D20503167 fbshipit-source-id: 5a609760130bd1131e299ce85b7824cbcbdf1f09 |
||
|---|---|---|
| .. | ||
| contrib | ||
| core | ||
| cuda_rtc | ||
| db | ||
| distributed | ||
| experiments | ||
| ideep | ||
| image | ||
| mobile | ||
| mpi | ||
| observers | ||
| onnx | ||
| operators | ||
| opt | ||
| perfkernels | ||
| predictor | ||
| proto | ||
| python | ||
| quantization | ||
| queue | ||
| serialize | ||
| sgd | ||
| share | ||
| test | ||
| transforms | ||
| utils | ||
| video | ||
| .clang-format | ||
| __init__.py | ||
| c2_aten_srcs.bzl | ||
| CMakeLists.txt | ||
| README.md | ||
| release-notes.md | ||
| requirements.txt | ||
| VERSION_NUMBER | ||
Caffe2
Caffe2 is a lightweight, modular, and scalable deep learning framework. Building on the original Caffe, Caffe2 is designed with expression, speed, and modularity in mind.
Questions and Feedback
Please use Github issues (https://github.com/pytorch/pytorch/issues) to ask questions, report bugs, and request new features.