-
Notifications
You must be signed in to change notification settings - Fork 496
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
E2E HuggingFace Bert using LTC Backend #912
E2E HuggingFace Bert using LTC Backend #912
Commits on Jun 6, 2022
-
Configuration menu - View commit details
-
Copy full SHA for 80f61a1 - Browse repository at this point
Copy the full SHA 80f61a1View commit details -
Add ops to support bert lowering
- Add empty_strided and as_strided - Restore zeros_like to op blacklist (Without this, tensors will be unintentionally created with a CPU device rather than lazy) - Check for composite implicit ops and add device data IR - Also fix codegen for functionalization
Configuration menu - View commit details
-
Copy full SHA for 67237df - Browse repository at this point
Copy the full SHA 67237dfView commit details -
Configuration menu - View commit details
-
Copy full SHA for 31b2d33 - Browse repository at this point
Copy the full SHA 31b2d33View commit details -
Configuration menu - View commit details
-
Copy full SHA for a601b9e - Browse repository at this point
Copy the full SHA a601b9eView commit details -
Configuration menu - View commit details
-
Copy full SHA for 950e0ec - Browse repository at this point
Copy the full SHA 950e0ecView commit details -
Configuration menu - View commit details
-
Copy full SHA for 0187a26 - Browse repository at this point
Copy the full SHA 0187a26View commit details -
Apply fixes to work with latest upstream/main
- Pass importOptions into getMlirTypeFromTorchType during NodeImporter::importNode Without this, the tensor type created may have a mismatched type as ImportOptions may cause vtensor to be used instead of tensor
Configuration menu - View commit details
-
Copy full SHA for dac0269 - Browse repository at this point
Copy the full SHA dac0269View commit details -
Update shape inference functions
- Fixed compute_shape_native_batch_norm when mean and var are uninitialized Previously, the number of shapes returned would be <3 if either mean or val was didn't exist. Instead, we now initialize them with a vector matching the number of channels. - Implemented compute_shape_mul - Fixed bug in reshape shape inference error message
Configuration menu - View commit details
-
Copy full SHA for f6e3761 - Browse repository at this point
Copy the full SHA f6e3761View commit details
Commits on Jun 7, 2022
-
Get MLIR backend more consistent with TS backend
- Remove LazyNativeFunctions::_unsafe_view from autogen - Blacklist ops to make JIT graph more like output of TS backend - Print graph when SSA value has mismatch of types and results - Remove normalize_index from LazyShapeInference - Fix seeds for LTC example models
Configuration menu - View commit details
-
Copy full SHA for 38ee102 - Browse repository at this point
Copy the full SHA 38ee102View commit details -
Update and clean up shape inference functions
- Prune shape inference functions - Add shape inference function for GenerateSlice - Add shape inference function for GenerateCopy
Configuration menu - View commit details
-
Copy full SHA for ec3c9a9 - Browse repository at this point
Copy the full SHA ec3c9a9View commit details