GitLab

* Implemented swiglu and silu
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* Renamed nvte-*silu to nvte-*swish + generalized GetDBiasDact functions
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* Remove unused headers
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Unify the fused attn workspace size cpp code
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Reduce the skipped cases
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Rename self/cross attention to qkvpacked/kvpacked
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Update attention mask docs
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refine the attn mask implementations
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>

Support various implementations of RoPE and fix a coordinate representation bug
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Refine MHA API
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Reuse func from the flax
Signed-off-by: Reese Wang <rewang@nvidia.com>

* DPA draft
Signed-off-by: Reese Wang <rewang@nvidia.com>

* qkv packed draft
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix test_layer with fused attn
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add attn_bias_type and enhance a few code flow
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Move scale_factor from __call__ to init
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Enhance the docs
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add DPA public API and tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refine docs
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refine docs
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix conflict
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add qkv separate fused attn
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Apply BSHD_BSHD_BSHD format
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Remove debug log
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add fused attention layer tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add NVTE_FUSED_ATTN docs
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fine-grained fused attn settings
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Remove the default value of num_attetnion_head and head_dim
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add teardown for fused attn env
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Unify the Optional notation
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix Pre/Post scale bias comments
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add no_mask tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add checkpoint_name for fused attn
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix the fused attn batcher
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Adding support of sequence parallelism
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding RoPE
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix wrong batch_logical_axes
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Rnaming FSDP outer env var
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Poring RoPE to Praxis layers.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Porting GeLU + [FP8 Cast].
Signed-off-by: Ming Huang <mingh@nvidia.com>

* WAR to make XLA successfully match FP8 GEMM on FFN1 with GeLU.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Allowing arbitrary dimension of NVShape for the workspace allocation
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding checkpoint_name to fused functions of mlp.py to get better perf with nn.scan.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Modify with review feedback.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix bugs
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix typo.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fixed for lint
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Follow review feedback to modify code.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix typo.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Port SP to Praxis
Signed-off-by: Ming-Xu Huang <mingh@nvidia.com>

* Fix an issue when enabling both GQA and RoPE.
Signed-off-by: Ming-Xu Huang <mingh@nvidia.com>

* Update docs
Signed-off-by: Ming-Xu Huang <mingh@nvidia.com>

---------
Signed-off-by: Ming Huang <mingh@nvidia.com>
Signed-off-by: Ming-Xu Huang <mingh@nvidia.com>

* Support num_gqa_groups arguments
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add GQA support on the JAX bridge code
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix the kv stride of the arbitrary backend
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Complete rewrite fused attention tests and add GQA coverage
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Support unfused GQA
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Calculate seqlen before the primitive for the better perf
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add GQA layer tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Apply code style checks for te_jax
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Apply code style checks for tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add num_gqa_groups doc
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refine the qkv_type
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Correct the variable naming
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Handle Max512 CAUSAL
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add WAR for the latest jax image
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>

Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

[C/JAX] Support more mask types for the arbitrary seqlen kernels and minor changes of JAX bias (#469)

* Move bias to float32
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Enable varlen
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Increase neg infinity abs values
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Enable varlen tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Remove unnecessary code
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix lint
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Support variable sequence length after cuDNN 8.9.6
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Use unique_ptr instead of shared_ptr
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add a new mask type: PADDING_CAUSAL_MASK
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Support flash padding mask after 8.9.6
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Enhance the Max512 handling for causal masking and add the related tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Update the fused attn support lists
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Remove padding_aware from the caching
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix libtransformer.so issue
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Reduce the pad ratio tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix a bug with cuDNN 8.9.5
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Release backend resource after the module level unit test
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Clean the jax live arrays before running the unit tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix too-few-public-methods lint
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>

* [JAX] Enhance Dropout in TransformerLayer.

1. Fixed missing setup of dropout RNG key in TransformerLayer and
   LayerNormMLP.
2. Allowing seperated dropout rate for FC1's output and other hiddens.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix wrong fp8 scale in _update_fp8_metas_impl
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix typo
Signed-off-by: Ming Huang <mingh@nvidia.com>

---------
Signed-off-by: Ming Huang <mingh@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Cast Flax collections to FrozenDict as WAR to adapt Flax 0.7.1.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding min version of flax to requirements.txt in examples
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix praxis tests and rename compare_frozen_dict to compare_dict
Signed-off-by: Ming Huang <mingh@nvidia.com>

* [Paddle] Refactor FP8 state (#350)

Refactor fp8 state
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Store FP8 checkpointing data in CPU (#351)
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Make test_layer be able to run on both Flax >=0.7.1 and <=0.7.0
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Update Flax version
Signed-off-by: Tim Moon <tmoon@nvidia.com>

---------
Signed-off-by: Ming Huang <mingh@nvidia.com>
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: Tian Zheng <tizheng@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Tim Moon <tmoon@nvidia.com>

* Fully remove attn_type and set self_attn_mask_type default to 'causal'
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix tests with new arguments
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Explicit self_attn_mask_type for examples
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Update transformer_engine/jax/flax/transformer.py
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: zlsh80826 <rewang@nvidia.com>

* Update transformer_engine/jax/flax/transformer.py
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: zlsh80826 <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>
Signed-off-by: zlsh80826 <rewang@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Unfused scale+softmax if bias is present
Signed-off-by: Reese Wang <rewang@nvidia.com>

* WAR a causal masking + no_bias bug and add the unittests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix the optional args (bias) sharding
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Disable fused attn in JAX by default, enable it with NVTE_USE_FUSED_ATTN
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add thread local for the plan cache
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Rename dbeta to dbias for the readability
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add scaled softmax with dropout test cases
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Updated NVTE_FUSED_ATTN variable name
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>

* add mp example
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* refactor
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* update doc-string
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* better FP8 checker
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* update readme
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* replace te.* with te.flax* to remove deprecated warning
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* remove nouse os.environ
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* remove nouse
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* fix typo
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Update examples/jax/encoder/README.md
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Update examples/jax/encoder/README.md
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Update examples/jax/encoder/README.md
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Update examples/jax/encoder/README.md
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Update examples/jax/encoder/test_multiprocessing_encoder.py
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* remove cuda-python
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* adjust readme
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Update examples/jax/encoder/README.md
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* fix cpp lint

fix issue of "Could not find a newline character at the end of the file."
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* fix AssertionError: 1 GPU per process
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* replace tfds with datasets

The Flax application crash if it use TensorFlow Dataset (tfds) in NVIDIA JAX container.
The tfds is very useful for downloading well-knwon dataset (e.g., MNIST, GLUE) and commonly used by TF/JAX community.
However, it seems like that it is NOT compatible with NVIDIA TensorFlow in NVIDIA JAX container and somehow affects JAX.
It triggers random errors at JAX initialization depending on different versions, and make CI unstable.
Thus, this commit replaces tfds with "huggingface datasets" to download needed datasets.
See "nvbugs 4039266" for more details.
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* fix input sharding

Unlike SPMD mode, in multiprocessing mode, the input tensor must be sharded manually.
Using DP=4, TP=2 as an example, the device mesh looks like:

mesh.device_ids = [[0, 1],
                   [2, 3],
                   [4, 5],
                   [6, 7]]

Assume that the process ID is mapped to GPU ID.
The process 0 and process 1 are grouped for model parallelism,
process 2 and process 3 are grouped together too, and so on.

The process 0 and process 1 need to share the same micro-batch in the training step,
process 0 and process 2, 4, and 6 have different micro-batch.

Thus, `shard_array_wrapper` partitions inputs to 4 parts (and setup
needed arguments for jax.make_array_from_single_device_arrays).
The process 0 and process 1 take the first quarter,
process 2 and process 3 take the second quarter, and so on.
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* refactor UT for multiprocess example

Use Python `multiprocessing` to test the multiprocessing example,
if the system has multiple GPU. 1 GPU per process.

Because `jax.distributed.initialize` must be called before any other JAX or Flax API,
GPU info cannot be queried by calling jax.local_devices() in TestEncoder.
Thus, `unittest_query_gpu()` forks another process to query number of GPUs and
FP8 capability.
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* remove nouse arg `--num-gpu`

JAX doesn't have an API to setup number of GPU used in SPMD mode.
The only way is to use `CUDA_VISIBLE_DEVICES` for now.
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* fix typo
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* fix ut
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* simplify the mask setting
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* increase batch-size for multigpu example

The batch-size 64 is too small to be partitioned for 8xH100.
If batch-size is 64, the GEMM shape is 256x8192x8 per GPU.
The 8 is too small for FP8 GEMM kernel, and
cuBLASLt will throw "Failed to query heuristics".
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* fix downloading mnist error

To download MNIST via `huggingface datasets`, it requires Pillow.
Otherwise, it throws `An error occurred while generating the
dataset`
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

---------
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* Add fused attention unit tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Use NVTE_* enums
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Use NVTE_Mask_Type and remove FMHADescriptor
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Move common functions to utils
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Change namespace to fused_attn
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Move fused_attn_max_512_fwd_qkvpacked under the general APIs
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add fused_attn_max_512_bwd_qkvpacked
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Move fused_attn_max_512_bwd_qkvpacked under the general APIs
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Remove redundant blank line
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix a potential bug for cu_seqlen converter
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Reformat fused_attn_max_512
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refine the unfused attention warning message
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Rename to fused_attn_max_512
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Remove the deprecated header
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix flax import
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Rename to fused attn
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add attention related mask
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add attn_mask_type and attn_bias_type
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refactor jax primitive API
* Merge q_cu_seqlen and kv_cu_seqlen
* Remove is_causal_masking
* Replace seed with rng_state
* Add is_training argument
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Remove dsoftmax from the customcall
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add None guard for bias and dropout_rng
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add version guard
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add is_fused_attn_kernel_available() to correctly dispatch the attention impl
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix the merge conflict
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Adjust the code style
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add the missing blank lines
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Change the order of FADescriptor members
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Enhance the readability of fused_attn_max_512.cu
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Generalize the input dimension unpacking
Signed-off-by: Reese Wang <rewang@nvidia.com>

* 16 bits fused attention requires 8.9.1
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Update fused attention support matrix
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Handle None type when sharding
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Change to the padding ratio
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Performance optimization for non-bias cases
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Revert the cudnn-frontend PRIVATE keyword which was used for debugging
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Revert "Update fused attention support matrix"

This reverts commit 4effe67d0f08f733919a329ce5ab421958740f4a.
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Treat b * s as total_seqs to align ragged cases
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add FP16/BF16 max_seqlen <= 512 fused attention to the support matrix
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refine test_fused_attn.py

* Replace reference code with flax.linen
* Remove unnecessary comments
* Use AttnMaskType
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Unify the cuDNN compile version
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add dropout to the support matrix
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Slightly adjust the headers
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Typo fix: remove redundant either
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Consolidating fused attention requirements
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Replace cudnn_frontend::throw_if with NVTE_CHECK for the better error line report
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Rename to fused_attn_fp16_bf16_max_seqlen_512 for the better readability
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Remove CUDNN_FRONTEND_UNUSED
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add more annotations to the custom calls
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Adjust Module Structure.

1. Collect Flax related modules to a sub-folder, flax.
2. Add a function to unify scale_init for zero-centered-gamma LN.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Make changes be compatible to previous versions.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adapt jax/examples to the new module structure.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Update jax/docs and Add deprecated warning.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Update README
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding deprecated_wrapper
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding deprecated warning to flax modules which imported via transformer_engine.jax
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix CI errors and update docs.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Removing unnecessary deprecated warning in docs.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Implementing __iter__ to DeprecatedEnum.
Signed-off-by: Ming Huang <mingh@nvidia.com>

---------
Signed-off-by: Ming Huang <mingh@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Add zero_center_gamma/functional pass
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add zero_centered_gamma for fp8_ln_mlp
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add zero_centered_gamma to modules
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add zero_centered_gamma to TransformerLayer
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refactored code style for improved readability and consistency
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Docs enhancement for zero_centered_gamma
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add escape for line break and remove some bad if conditions
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Revise scale_init docs
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* Rename enable_fp8 to is_fp8_enabled.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding an API to get an instance of  DelayedScaling which is set via fp8_autocast.
Signed-off-by: Ming Huang <mingh@nvidia.com>

---------
Signed-off-by: Ming Huang <mingh@nvidia.com>

* add transformer module , unittests and examples
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Update tests/jax/test_sharding.py
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>

* Update transformer_engine/jax/transformer.py
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>

* remove pylint: disable=line-too-long
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* remove pylint: disable=too-many-func-args
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Fix the wrong broadcasting dim to dropout masks when enable transpose_bs.
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Enable 2xACC for WGRAD and DGRAD by default
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* rename LayerNormMlpBlock as LayerNormMLP
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* refactor to avoid line-too-long
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* rename amax_history_size to amax_history_len
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* align dropout mask to TE/PyTorch as default
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* enlarge atol for decoder unittests

Two decoder unittests can pass in old JAX container(e.g., 23.02)
but can't in latest container (devel).

1. The actual(-0.020264) and desired(-0.020386) are very close.
2. The TE kernels are not changed, the diff should come from
   new codegen behavior of XLA.

Thus, it is a common floating-point accumulated error.
Enlarge atol to avoid unittest failures.
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Adding Amax History Support

1. hide amax update in custom_vjp
2. replace amax indexing with roll(using circular buffer)
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* move kernel_init to __post_init__
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* refactor encoder examples
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* Update transformer_engine/jax/fp8.py
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>

* Update transformer_engine/jax/fp8.py
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>

* remove envvar regarding 2xACC
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

* remove unused import
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>

---------
Signed-off-by: Ryan Jeng <rjeng@nvidia.com>
Signed-off-by: Jeng Bai-Cheng <jeng1220@users.noreply.github.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Co-authored-by: Ming-Xu Huang <mingh@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>