Skip to content

Commit

Permalink
update docs for FastBert (#1496)
Browse files Browse the repository at this point in the history
  • Loading branch information
jingxu10 authored Mar 22, 2023
1 parent 99a70a6 commit 9878e4f
Show file tree
Hide file tree
Showing 3 changed files with 4 additions and 6 deletions.
4 changes: 1 addition & 3 deletions docs/tutorials/features.rst
Original file line number Diff line number Diff line change
Expand Up @@ -201,9 +201,7 @@ For more detailed information, check `HyperTune <features/hypertune.md>`_.
Fast BERT Optimization (Experimental, *NEW feature from 2.0.0*)
---------------------------------------------------------------

Intel proposed a technique, Tensor Processing Primitives (TPP), a programming abstraction striving for efficient, portable implementation of DL workloads with high-productivity. TPPs define a compact, yet versatile set of 2D-tensor operators (or a virtual Tensor ISA), which subsequently can be utilized as building-blocks to construct complex operators on high-dimensional tensors.

Implementation of TPP is integrated into Intel® Extension for PyTorch\*. BERT could benefit from this new technique. An API `ipex.fast_bert` is provided for a simple usage.
Intel proposed a technique to speed up BERT workloads. Implementation is integrated into Intel® Extension for PyTorch\*. An API `ipex.fast_bert` is provided for a simple usage.

For more detailed information, check `Fast BERT <features/fast_bert.md>`_.

Expand Down
4 changes: 2 additions & 2 deletions docs/tutorials/features/fast_bert.md
Original file line number Diff line number Diff line change
Expand Up @@ -3,9 +3,9 @@ Fast BERT (Experimental)

### Feature Description

Intel proposed a technique, Tensor Processing Primitives (TPP), a programming abstraction striving for efficient, portable implementation of DL workloads with high-productivity. TPPs define a compact, yet versatile set of 2D-tensor operators (or a virtual Tensor ISA), which subsequently can be utilized as building-blocks to construct complex operators on high-dimensional tensors. Detailed contents are available at [*Tensor Processing Primitives: A Programming Abstraction for Efficiency and Portability in Deep Learning & HPC Workloads*](https://arxiv.org/pdf/2104.05755.pdf).
Intel proposed a technique to speed up BERT workloads. Implementation leverages the idea from [*Tensor Processing Primitives: A Programming Abstraction for Efficiency and Portability in Deep Learning & HPC Workloads*](https://arxiv.org/pdf/2104.05755.pdf).

Implementation of TPP is integrated into Intel® Extension for PyTorch\*. BERT could benefit from this new technique, for both training and inference.
The Implementation is integrated into Intel® Extension for PyTorch\*. BERT could benefit from this new technique, for both training and inference.

### Prerequisite

Expand Down
2 changes: 1 addition & 1 deletion docs/tutorials/releases.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ We are pleased to announce the release of Intel® Extension for PyTorch\* 2.0.0-

### Highlights

- **Fast BERT optimization (Experimental)**: Intel introduced a new technique, Tensor Processing Primitives (TPP), a programming abstraction striving for efficient, portable implementation of DL workloads with high-productivity. Intel® Extension for PyTorch\* integrated this TPP implementation, which benefits BERT model especially training. A new API `ipex.fast_bert` is provided to try this new optimization. More detailed information can be found at [Fast Bert Feature](./features/fast_bert.md).
- **Fast BERT optimization (Experimental)**: Intel introduced a new technique to speed up BERT workloads. Intel® Extension for PyTorch\* integrated this implementation, which benefits BERT model especially training. A new API `ipex.fast_bert` is provided to try this new optimization. More detailed information can be found at [Fast Bert Feature](./features/fast_bert.md).

- **Work with torch.compile as an backend (Experimental)**: PyTorch 2.0 introduces a new feature, `torch.compile`, to speed up PyTorch execution. We've enabled Intel® Extension for PyTorch as a backend of torch.compile, which can leverage this new PyTorch API's power of graph capture and provide additional optimization based on these graphs.
The usage of this new feature is quite simple as below:
Expand Down

0 comments on commit 9878e4f

Please sign in to comment.