v2.1.1

What's Changed

Dependency changes

• Pull in new versions of fms-acceleration-peft, fms-acceleration-foak with fixes for AutoGPTQ and gradient accumulation hooks and adds granite GPTQ model
• build(deps): set transformers below 4.46, waiting on fixes by @anhuong in #384

Additional changes

• docs: Update Supported Models List in README by @tharapalanivel in #382

Full Changelog: v2.1.0...v2.1.1

v2.1.1-rc.2

deps: set transformers below 4.46, waiting on fixes (#384)

Signed-off-by: Anh Uong <anh.uong@ibm.com>

v2.1.1-rc.1

docs: Update supported models list in README (#382)

Signed-off-by: Thara Palanivel <130496890+tharapalanivel@users.noreply.github.com>

v2.1.0

New Major Feature

• Support for GraniteForCausalLM model architecture

Dependency upgrades

• Upgraded transformers to version 4.45.2, now supports GraniteForCausalLM models. Note that if a model is trained with transformers v4.45, you need the same version transformers>=4.45 to load the trained model, prior versions of transformers will not be compatible.
• Upgraded accelerate to version 1.0.1.
• Limit upper bound of torch to be under v2.5.0, not including v2.5.0, so it's compatible with flash-attention-2.
• Upgraded fms_acceleration_peft to v0.3.1 which includes disabling offloading state dict which caused ephemeral storage issues when loading large models with QLoRA. Also includes setting defaults when target_modules=None.

Additional bug fix

• Fix for crash when running a multi GPU training with a non-existent output dir.

Full list of Changes

• ci: run unit tests, fmt, image build on release branch by @anhuong in #361
• chore: update code owners by @anhuong in #363
• fix: crash when output directory doesn't exist by @HarikrishnanBalagopal in #364
• refactor: move tokenizer_data_utils with the rest of utils, add further unit testing. by @willmj in #348
• build(deps): update transformers and accelerate deps by @anhuong in #355
• build(deps): Update peft requirement from <0.13,>=0.8.0 to >=0.8.0,<0.14 by @dependabot in #354
• build(deps): Upgrade accelerate requirement to allow version 1.0.0 by @willmj in #371
• build: Set triton environment variables by @willmj in #370
• build(deps): torch<2.5 due to FA2 error with new version by @anhuong in #375
• chore: merge set of changes for v2.1.0 by @aluu317 in #376

Full Changelog: v2.0.1...v2.1.0

v2.1.0-rc.1

What's Changed

• ci: run unit tests, fmt, image build on release branch by @anhuong in #361
• chore: update code owners by @anhuong in #363
• fix: crash when output directory doesn't exist by @HarikrishnanBalagopal in #364
• refactor: move tokenizer_data_utils with the rest of utils, add further unit testing. by @willmj in #348
• build(deps): update transformers and accelerate deps by @anhuong in #355
• build(deps): Update peft requirement from <0.13,>=0.8.0 to >=0.8.0,<0.14 by @dependabot in #354
• build(deps): Upgrade accelerate requirement to allow version 1.0.0 by @willmj in #371
• build: Set triton environment variables by @willmj in #370

Full Changelog: v2.0.1...v2.0.2-rc.1

v2.0.1

New major features:

1. Support for LoRA for the following model architectures - llama3, llama3.1, granite (GPTBigCode and LlamaForCausalLM), mistral, mixtral, and allam
2. Support for QLora for the following model architectures - llama3, granite (GPTBigCode and LlamaForCausalLM), mistral, mixtral
3. Addition of post-processing function to format tuned adapters as required by vLLM for inference. Refer to README on how to run as a script. When tuning on image, post-processing can be enabled using the flag lora_post_process_for_vllm. See build README for details on how to set this flag.
4. Enablement of new flags for throughput improvements: padding_free to process multiple examples without adding padding tokens, multipack for multi-GPU training to balance the number of tokens processed on each device, and fast_kernels for optimized tuning with fused operations and triton kernels. See README for details on how to set these flags and use cases.

Dependency upgrades:

1. Upgraded transformers to version 4.44.2 needed for tuning of all models
2. Upgraded accelerate to version 0.33 needed for tuning of all models. Version 0.34.0 has a bug for FSDP.

API /interface changes:

1. train() API now returns a tuple of trainer instance and additional metadata as a dict

Additional features and fixes

1. Support of resume tuning from the existing checkpoint. Refer to README on how to use it as a flag. Flag resume_training defaults to True.
2. Addition of default pad token in tokenizer when EOS and PAD tokens are equal to improve training quality.
3. JSON compatability for input datasets. See docs for details on data formats.
4. Fix to not resize embedding layer by default, embedding layer can continue to be resized as needed using flag embedding_size_multiple_of.

Full List of what's Changed

• fix: do not resize embedding layer by default by @kmehant in #310
• fix: logger is unbound error by @HarikrishnanBalagopal in #308
• feat: Enable JSON dataset compatibility by @willmj in #297
• doc: How to tune LoRA lm_head by @aluu317 in #305
• docs: Add findings from exploration into model tuning performance degradation by @willmj in #315
• fix: warnings about casing when building the Docker image by @HarikrishnanBalagopal in #318
• fix: need to pass skip_prepare_dataset for pretokenized dataset due to breaking change in HF SFTTrainer by @HarikrishnanBalagopal in #326
• feat: install fms-acceleration to enable qlora by @anhuong in #284
• feat: Migrating the trainer controller to python logger by @seshapad in #309
• fix: remove fire ported from Hari's PR #303 by @HarikrishnanBalagopal in #324
• dep: cap transformers version due to FSDP bug by @anhuong in #335
• deps: Add protobuf to support aLLaM models by @willmj in #336
• fix: add enable_aim build args in all stages needed by @anhuong in #337
• fix: remove lm_head post processing by @Abhishek-TAMU in #333
• doc: Add qLoRA README by @aluu317 in #322
• feat: Add deps to evaluate qLora tuned model by @aluu317 in #312
• feat: Add support for smoothly resuming training from a saved checkpoint by @Abhishek-TAMU in #300
• ci: add a github workflow to label pull requests based on their title by @HarikrishnanBalagopal in #298
• fix: Addition of default pad token in tokenizer when EOS and PAD token are equal by @Abhishek-TAMU in #343
• feat: Add DataClass Arguments to Activate Padding-Free and MultiPack Plugin and FastKernels by @achew010 in #280
• fix: cap transformers at v4.44 by @anhuong in #349
• fix: utilities to post process checkpoint for LoRA by @Ssukriti in #338
• feat: Add post processing logic to accelerate launch by @willmj in #351
• build: install additional fms-acceleration plugins by @anhuong in #350
• fix: unable to find output_dir in multi-GPU during resume_from_checkpoint check by @Abhishek-TAMU in #352
• fix: check for wte.weight along with embed_tokens.weight by @willmj in #356
• release: merge set of changes for v2.0.0 by @Abhishek-TAMU in #357
• build(deps): unset hardcoded trl version to get latest updates by @anhuong in #358

New Contributors

• @achew010 made their first contribution in #280

Full Changelog: v1.2.2...v2.0.0

v2.0.0

This version has old dependency and users should move to v2.0.1 instead

v2.0.0-rc.2

What's Changed

• fix: check for wte.weight along with embed_tokens.weight by @willmj in #356

Full Changelog: v2.0.0-rc.1...v2.0.0-rc.2

v2.0.0-rc.1

What's Changed

• fix: do not resize embedding layer by default by @kmehant in #310
• fix: logger is unbound error by @HarikrishnanBalagopal in #308
• feat: Enable JSON dataset compatibility by @willmj in #297
• doc: How to tune LoRA lm_head by @aluu317 in #305
• docs: Add findings from exploration into model tuning performance degradation by @willmj in #315
• fix: warnings about casing when building the Docker image by @HarikrishnanBalagopal in #318
• fix: need to pass skip_prepare_dataset for pretokenized dataset due to breaking change in HF SFTTrainer by @HarikrishnanBalagopal in #326
• feat: install fms-acceleration to enable qlora by @anhuong in #284
• feat: Migrating the trainer controller to python logger by @seshapad in #309
• fix: remove fire ported from Hari's PR #303 by @HarikrishnanBalagopal in #324
• dep: cap transformers version due to FSDP bug by @anhuong in #335
• deps: Add protobuf to support aLLaM models by @willmj in #336
• fix: add enable_aim build args in all stages needed by @anhuong in #337
• fix: remove lm_head post processing by @Abhishek-TAMU in #333
• doc: Add qLoRA README by @aluu317 in #322
• feat: Add deps to evaluate qLora tuned model by @aluu317 in #312
• feat: Add support for smoothly resuming training from a saved checkpoint by @Abhishek-TAMU in #300
• ci: add a github workflow to label pull requests based on their title by @HarikrishnanBalagopal in #298
• fix: Addition of default pad token in tokenizer when EOS and PAD token are equal by @Abhishek-TAMU in #343
• feat: Add DataClass Arguments to Activate Padding-Free and MultiPack Plugin and FastKernels by @achew010 in #280
• fix: cap transformers at v4.44 by @anhuong in #349
• fix: utilities to post process checkpoint for LoRA by @Ssukriti in #338
• feat: Add post processing logic to accelerate launch by @willmj in #351
• build: install additional fms-acceleration plugins by @anhuong in #350
• fix: unable to find output_dir in multi-GPU during resume_from_checkpoint check by @Abhishek-TAMU in #352

New Contributors

• @achew010 made their first contribution in #280

Full Changelog: v1.2.1...v2.0.0-rc.1

v1.2.2

What's Changed

• deps: Add protobuf to support ALLaM models by @willmj in #328
• deps: set previous versions for accelerate and trl for patch release by @willmj in #329

Full Changelog: v1.2.1...v1.2.2