Jack
Jack
Reviewed By: helunwencser Differential Revision: D60319977
The v0.4.0 release will be cut from the "[release/0.4](https://github.com/pytorch/executorch/tree/release/0.4)" branch for critical fixes to the [v0.4.0](https://github.com/pytorch/executorch/releases/tag/v0.4.0) release. Branch Cut Date: 09/15/2024 Intended Release Date: 10/17/2024 This issue is for tracking...
For situations where the forward has non-position arguments, such as https://github.com/pytorch/torchtune/blob/3c450ef5f1fbe8237f899e942fd5222491a47ca7/torchtune/modules/transformer.py#L519
### Summary - Removes redundant steps in the Llama2 export - Factors out checkpointing to be shared with future Llama models (namely 3.2 multimodal) - Comments and orders code more...
### Summary Adds a default pull request template. Follows how PyTorch tracks changelogs for release notes, which is adding "Release notes: " labels, e.g. [this](https://github.com/pytorch/pytorch/pull/137289) PyTorch PR. Fixes #5793 ###...
### 🚀 The feature, motivation and pitch Create a pull request template to encourage more descriptive PRs and facilitate release note aggregation. ### Alternatives N/A ### Additional context N/A ###...
#### Context This PR factors out the optimizable portions of SDPA (namely the kv cache update, the transpose, the expand, and the actual sdpa). This allows a module containing optimized...
Refactor `_to_edge_and_lower_llama_xnnpack` and `_to_edge_and_lower_llama` to remove args
Refactor `_get_source_transforsm` to remove args