add smoke test training

.github/workflows/base.yml

@@ -40,7 +40,7 @@ jobs:
             cuda_version: 12.4.1
             cudnn_version: ""
             python_version: "3.11"
-            pytorch: 2.5.1
+            pytorch: 2.5.0
             torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
     steps:
       - name: Checkout

.github/workflows/tests-nightly.yml

@@ -82,6 +82,13 @@ jobs:
             num_gpus: 1
             axolotl_extras: mamba-ssm
             nightly_build: "true"
+          - cuda: 121
+            cuda_version: 12.1.1
+            python_version: "3.11"
+            pytorch: 2.3.1
+            num_gpus: 1
+            axolotl_extras: mamba-ssm
+            nightly_build: "true"
           - cuda: 124
             cuda_version: 12.4.1
             python_version: "3.11"

.github/workflows/tests.yml

@@ -72,52 +72,12 @@ jobs:
         run: |
           find "$(pip cache dir)/http-v2" -type f -mtime +14 -exec rm {} \;
 
-  docker-e2e-tests-1st:
-    if: github.repository_owner == 'axolotl-ai-cloud'
-    # this job needs to be run on self-hosted GPU runners...
-    runs-on: [self-hosted, modal]
-    timeout-minutes: 90
-    needs: [pre-commit, pytest]
-
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - cuda: 124
-            cuda_version: 12.4.1
-            python_version: "3.11"
-            pytorch: 2.4.1
-            num_gpus: 1
-            axolotl_extras:
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v4
-      - name: Install Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: "3.10"
-      - name: Install Modal
-        run: |
-          python -m pip install --upgrade pip
-          pip install modal==0.63.64 jinja2
-      - name: Update env vars
-        run: |
-          echo "BASE_TAG=main-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}" >> $GITHUB_ENV
-          echo "PYTORCH_VERSION=${{ matrix.pytorch}}" >> $GITHUB_ENV
-          echo "AXOLOTL_ARGS=${{ matrix.axolotl_args}}" >> $GITHUB_ENV
-          echo "AXOLOTL_EXTRAS=${{ matrix.axolotl_extras}}" >> $GITHUB_ENV
-          echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
-          echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
-      - name: Run tests job on Modal
-        run: |
-          modal run cicd.tests
-
   docker-e2e-tests:
     if: github.repository_owner == 'axolotl-ai-cloud'
     # this job needs to be run on self-hosted GPU runners...
     runs-on: [self-hosted, modal]
     timeout-minutes: 90
-    needs: [pre-commit, pytest, docker-e2e-tests-1st]
+    needs: [pre-commit, pytest]
 
     strategy:
       fail-fast: false
@@ -129,6 +89,18 @@ jobs:
             pytorch: 2.3.1
             num_gpus: 1
             axolotl_extras: mamba-ssm
+          - cuda: 121
+            cuda_version: 12.1.1
+            python_version: "3.11"
+            pytorch: 2.3.1
+            num_gpus: 1
+            axolotl_extras: mamba-ssm
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.4.1
+            num_gpus: 1
+            axolotl_extras:
           - cuda: 124
             cuda_version: 12.4.1
             python_version: "3.11"

docs/config.qmd

@@ -183,8 +183,6 @@ test_datasets:
 
 # use RL training: 'dpo', 'ipo', 'kto'
 rl:
-# whether to perform weighting if doing DPO training. Boolean.
-dpo_use_weighting:
 
 # The name of the chat template to use for training, following values are supported:
 # - tokenizer_default: Uses the chat template that is available in the tokenizer_config.json. If the chat template is not available in the tokenizer, it will raise an error. This is the default value.

requirements.txt

@@ -43,7 +43,7 @@ s3fs>=2024.5.0
 gcsfs>=2024.5.0
 # adlfs
 
-trl @ git++https://github.com/huggingface/trl.git@5e90682836969310e16ed8aa711dd429f85863b7
+trl @ git+https://github.com/huggingface/trl.git@31d02cfb795284591a084416b9dcb7bef5d08924
 zstandard==0.22.0
 fastcore
 

src/axolotl/core/trainer_builder.py

@@ -48,7 +48,6 @@ from trl import (
 )
 from trl.trainer.utils import RewardDataCollatorWithPadding, pad_to_length
 
-from axolotl.integrations.base import PluginManager
 from axolotl.monkeypatch.multipack import SUPPORTED_MULTIPACK_MODEL_TYPES
 from axolotl.monkeypatch.relora import ReLoRACallback, ReLoRAScheduler
 from axolotl.utils import is_comet_available, is_mlflow_available
@@ -436,7 +435,13 @@ class AxolotlTrainer(SchedulerMixin, Trainer):
         if (
             self.args.loraplus_lr_ratio is None
             and self.args.alternate_optimizer
-            not in ["optimi_adamw", "ao_adamw_8bit", "ao_adamw_4bit", "ao_adamw_fp8"]
+            not in [
+                "optimi_adamw",
+                "ao_adamw_8bit",
+                "ao_adamw_4bit",
+                "ao_adamw_fp8",
+                "soap",
+            ]
         ):
             return super().create_optimizer()
 
@@ -479,6 +484,25 @@ class AxolotlTrainer(SchedulerMixin, Trainer):
                     loraplus_lr_embedding=loraplus_lr_embedding,
                     **optimizer_kwargs,
                 )
+            elif self.args.alternate_optimizer == "soap":
+                from axolotl.utils.optimizers.soap import SOAP
+
+                optim_args = {
+                    "lr": optimizer_kwargs.pop("lr"),
+                    "eps": optimizer_kwargs.pop("eps"),
+                }
+
+                if self.cfg.optim_args:
+                    optim_args.update(self.cfg.optim_args)
+
+                optim_args["betas"] = (
+                    self.args.optim_soap_beta1,
+                    self.args.optim_soap_beta2,
+                )
+                self.optimizer = SOAP(  # pylint: disable=attribute-defined-outside-init
+                    optimizer_grouped_parameters,
+                    **optim_args,
+                )
             elif self.args.alternate_optimizer == "optimi_adamw":
                 from optimi import AdamW
 
@@ -1148,12 +1172,6 @@ class TrainerBuilderBase(abc.ABC):
 
     def get_callbacks(self) -> List[TrainerCallback]:
         callbacks = []
-
-        plugin_manager = PluginManager.get_instance()
-        callbacks.extend(
-            plugin_manager.add_callbacks_pre_trainer(cfg=self.cfg, model=self.model)
-        )
-
         if self.cfg.use_wandb:
             callbacks.append(
                 SaveAxolotlConfigtoWandBCallback(self.cfg.axolotl_config_path)
@@ -1180,17 +1198,11 @@ class TrainerBuilderBase(abc.ABC):
 
         return callbacks
 
+    @abstractmethod
     def get_post_trainer_create_callbacks(self, trainer):
         """
         Callbacks added after the trainer is created, usually b/c these need access to the trainer
         """
-        callbacks = []
-
-        plugin_manager = PluginManager.get_instance()
-        callbacks.extend(
-            plugin_manager.add_callbacks_post_trainer(cfg=self.cfg, trainer=trainer)
-        )
-        return callbacks
 
     def hook_pre_create_training_args(self, training_arguments_kwargs):
         # TODO
@@ -1236,7 +1248,7 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
         return callbacks
 
     def get_post_trainer_create_callbacks(self, trainer):
-        callbacks = super().get_post_trainer_create_callbacks(trainer=trainer)
+        callbacks = []
         if self.cfg.use_wandb and self.cfg.eval_table_size > 0:
             LogPredictionCallback = log_prediction_callback_factory(
                 trainer, self.tokenizer, "wandb"
@@ -1626,10 +1638,12 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
             trainer_kwargs["max_length"] = self.cfg.sequence_len
 
         if self.cfg.optimizer in [
+            # pylint: disable=duplicate-code
             "optimi_adamw",
             "ao_adamw_4bit",
             "ao_adamw_8bit",
             "ao_adamw_fp8",
+            "soap",
         ]:
             # Set default so transformers doesn't throw
             training_arguments_kwargs["optim"] = "adamw_hf"
@@ -1804,7 +1818,7 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
         return callbacks
 
     def get_post_trainer_create_callbacks(self, trainer):
-        callbacks = super().get_post_trainer_create_callbacks(trainer=trainer)
+        callbacks = []
         return callbacks
 
     def build_training_arguments(self, total_num_steps):
@@ -1890,18 +1904,17 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
             # default to saving each epoch if not defined
             training_args_kwargs["save_strategy"] = "epoch"
 
-        training_args_kwargs["dataset_num_proc"] = self.cfg.dataset_processes
-
         if self.cfg.rl_beta:
             training_args_kwargs["beta"] = self.cfg.rl_beta
         if self.cfg.orpo_alpha:
             # trl does some odd mapping of alpha to beta to reuse the beta parameter ???
             training_args_kwargs["beta"] = self.cfg.orpo_alpha
 
+        training_args_kwargs["dataset_num_proc"] = self.cfg.dataset_processes
+        training_args_cls = AxolotlDPOConfig
         if self.cfg.rpo_alpha is not None:
             training_args_kwargs["rpo_alpha"] = self.cfg.rpo_alpha
 
-        training_args_cls = None
         if self.cfg.rl == "simpo":
             training_args_cls = AxolotlCPOConfig
             training_args_kwargs["loss_type"] = "simpo"
@@ -1910,13 +1923,13 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
             if self.cfg.cpo_alpha is not None:
                 training_args_kwargs["cpo_alpha"] = self.cfg.cpo_alpha
 
-        elif self.cfg.rl == "orpo":
+        if self.cfg.rl == "orpo":
             training_args_cls = AxolotlORPOConfig
             training_args_kwargs["max_length"] = self.cfg.sequence_len
             if self.cfg.max_prompt_len:
                 training_args_kwargs["max_prompt_length"] = self.cfg.max_prompt_len
 
-        elif self.cfg.rl == "kto":
+        if self.cfg.rl == "kto":
             training_args_cls = AxolotlKTOConfig
 
             training_args_kwargs["desirable_weight"] = (
@@ -1926,32 +1939,11 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
                 self.cfg.kto_undesirable_weight or 1.0
             )
 
+            training_args_kwargs["dataset_num_proc"] = self.cfg.dataset_processes
             training_args_kwargs["max_length"] = self.cfg.sequence_len
             if self.cfg.max_prompt_len:
                 training_args_kwargs["max_prompt_length"] = self.cfg.max_prompt_len
 
-        else:
-            training_args_cls = AxolotlDPOConfig
-
-            training_args_kwargs["max_length"] = self.cfg.sequence_len
-
-            training_args_kwargs["max_target_length"] = None
-            if self.cfg.max_prompt_len is not None:
-                training_args_kwargs["max_prompt_length"] = self.cfg.max_prompt_len
-
-            if self.cfg.dpo_use_weighting is not None:
-                training_args_kwargs["use_weighting"] = self.cfg.dpo_use_weighting
-
-            if self.cfg.rl == "ipo":
-                training_args_kwargs["loss_type"] = "ipo"
-                if self.cfg.dpo_label_smoothing:
-                    training_args_kwargs["label_smoothing"] = self.cfg.dpo_label_smoothing
-            
-            if self.cfg.precompute_ref_log_probs is not None:
-                training_args_kwargs["precompute_ref_log_probs"] = self.cfg.precompute_ref_log_probs
-
-            training_args_kwargs["generate_during_eval"] = self.cfg.use_wandb
-
         training_args = training_args_cls(  # pylint: disable=unexpected-keyword-arg
             output_dir=self.cfg.output_dir,
             per_device_train_batch_size=self.cfg.micro_batch_size,
@@ -1971,16 +1963,27 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
     def build(self, total_num_steps):
         training_args = self.build_training_arguments(total_num_steps)
         dpo_trainer_kwargs = {}
-        
+        if self.cfg.rl == "ipo":
+            dpo_trainer_kwargs["loss_type"] = "ipo"
+            if self.cfg.dpo_label_smoothing:
+                dpo_trainer_kwargs["label_smoothing"] = self.cfg.dpo_label_smoothing
         if self.eval_dataset:
             dpo_trainer_kwargs["eval_dataset"] = self.eval_dataset
         if self.cfg.adapter and self.peft_config:
             dpo_trainer_kwargs["peft_config"] = self.peft_config
-  
+        if self.cfg.precompute_ref_log_probs is not None:
+            dpo_trainer_kwargs[
+                "precompute_ref_log_probs"
+            ] = self.cfg.precompute_ref_log_probs
         if self.cfg.rl in ["dpo", "ipo"]:
             trainer_cls = AxolotlDPOTrainer
             trainer_cls_args = [self.model, self.model_ref]
 
+            # these aren't used for the ORPO trainer
+            dpo_trainer_kwargs["max_length"] = self.cfg.sequence_len
+            dpo_trainer_kwargs["max_target_length"] = None
+            dpo_trainer_kwargs["max_prompt_length"] = self.cfg.sequence_len
+            dpo_trainer_kwargs["generate_during_eval"] = self.cfg.use_wandb
         elif self.cfg.rl == "orpo":
             trainer_cls = AxolotlORPOTrainer
             trainer_cls_args = [self.model]
@@ -2024,11 +2027,11 @@ class HFPPOTrainerBuilder(TrainerBuilderBase):
     """
 
     def get_callbacks(self):
-        callbacks = super().get_callbacks()
+        callbacks = []
         return callbacks
 
     def get_post_trainer_create_callbacks(self, trainer):
-        callbacks = super().get_post_trainer_create_callbacks(trainer=trainer)
+        callbacks = []
         return callbacks
 
     def build(self, total_num_steps):

src/axolotl/integrations/base.py

@@ -18,10 +18,9 @@ Plugins can be used to integrate third-party models, modify the training process
 
 To create a new plugin, you need to inherit from the BasePlugin class and implement the required methods.
 """
-import collections
 import importlib
 import logging
-from typing import OrderedDict
+from typing import List
 
 
 class BasePlugin:
@@ -48,7 +47,7 @@ class BasePlugin:
         Initializes the BasePlugin.
         """
 
-    def register(self, cfg):  # pylint: disable=unused-argument
+    def register(self, cfg):
         """
         Registers the plugin with the given configuration.
 
@@ -64,7 +63,7 @@ class BasePlugin:
         Returns a pydantic model for the plugin's input arguments.
         """
 
-    def pre_model_load(self, cfg):  # pylint: disable=unused-argument
+    def pre_model_load(self, cfg):
         """
         Performs actions before the model is loaded.
 
@@ -75,7 +74,7 @@ class BasePlugin:
         None
         """
 
-    def post_model_load(self, cfg, model):  # pylint: disable=unused-argument
+    def post_model_load(self, cfg, model):
         """
         Performs actions after the model is loaded.
 
@@ -87,7 +86,7 @@ class BasePlugin:
         None
         """
 
-    def pre_lora_load(self, cfg, model):  # pylint: disable=unused-argument
+    def pre_lora_load(self, cfg, model):
         """
         Performs actions before LoRA weights are loaded.
 
@@ -99,7 +98,7 @@ class BasePlugin:
         None
         """
 
-    def post_lora_load(self, cfg, model):  # pylint: disable=unused-argument
+    def post_lora_load(self, cfg, model):
         """
         Performs actions after LoRA weights are loaded.
 
@@ -111,7 +110,7 @@ class BasePlugin:
         None
         """
 
-    def create_optimizer(self, cfg, trainer):  # pylint: disable=unused-argument
+    def create_optimizer(self, cfg, trainer):
         """
         Creates and returns an optimizer for training.
 
@@ -123,9 +122,7 @@ class BasePlugin:
         object: The created optimizer.
         """
 
-    def create_lr_scheduler(
-        self, cfg, trainer, optimizer
-    ):  # pylint: disable=unused-argument
+    def create_lr_scheduler(self, cfg, trainer, optimizer):
         """
         Creates and returns a learning rate scheduler.
 
@@ -138,7 +135,7 @@ class BasePlugin:
         object: The created learning rate scheduler.
         """
 
-    def add_callbacks_pre_trainer(self, cfg, model):  # pylint: disable=unused-argument
+    def add_callbacks_pre_trainer(self, cfg, model):
         """
         Adds callbacks to the trainer before training.
 
@@ -149,11 +146,8 @@ class BasePlugin:
         Returns:
         List[callable]: A list of callback functions to be added to the TrainingArgs
         """
-        return []
 
-    def add_callbacks_post_trainer(
-        self, cfg, trainer
-    ):  # pylint: disable=unused-argument
+    def add_callbacks_post_trainer(self, cfg, trainer):
         """
         Adds callbacks to the trainer after training.
 
@@ -164,9 +158,8 @@ class BasePlugin:
         Returns:
         List[callable]: A list of callback functions to be added to the TrainingArgs
         """
-        return []
 
-    def post_train(self, cfg, model):  # pylint: disable=unused-argument
+    def post_train(self, cfg, model):
         """
         Performs actions after training is complete.
 
@@ -178,7 +171,7 @@ class BasePlugin:
         None
         """
 
-    def post_train_unload(self, cfg):  # pylint: disable=unused-argument
+    def post_train_unload(self, cfg):
         """
         Performs actions after training is complete and the model is unloaded.
 
@@ -234,7 +227,7 @@ class PluginManager:
     pre_model_load(cfg): Calls the pre_model_load method of all registered plugins.
     """
 
-    plugins: OrderedDict[str, BasePlugin] = collections.OrderedDict()
+    plugins: List[BasePlugin] = []
 
     _instance = None
 
@@ -244,7 +237,7 @@ class PluginManager:
         """
         if cls._instance is None:
             cls._instance = super(PluginManager, cls).__new__(cls)
-            cls._instance.plugins = collections.OrderedDict()
+            cls._instance.plugins: List[BasePlugin] = []
         return cls._instance
 
     @staticmethod
@@ -272,7 +265,7 @@ class PluginManager:
         """
         try:
             plugin = load_plugin(plugin_name)
-            self.plugins[plugin_name] = plugin
+            self.plugins.append(plugin)
         except ImportError:
             logging.error(f"Failed to load plugin: {plugin_name}")
 
@@ -284,7 +277,7 @@ class PluginManager:
         list[str]: A list of Pydantic classes for all registered plugins' input arguments.'
         """
         input_args = []
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             input_args_from_plugin = plugin.get_input_args()
             if input_args_from_plugin is not None:
                 input_args.append(input_args_from_plugin)
@@ -300,7 +293,7 @@ class PluginManager:
         Returns:
         None
         """
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             plugin.pre_model_load(cfg)
 
     def post_model_load(self, cfg, model):
@@ -314,7 +307,7 @@ class PluginManager:
         Returns:
         None
         """
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             plugin.post_model_load(cfg, model)
 
     def pre_lora_load(self, cfg, model):
@@ -328,7 +321,7 @@ class PluginManager:
         Returns:
         None
         """
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             plugin.pre_lora_load(cfg, model)
 
     def post_lora_load(self, cfg, model):
@@ -342,7 +335,7 @@ class PluginManager:
         Returns:
         None
         """
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             plugin.post_lora_load(cfg, model)
 
     def create_optimizer(self, cfg, trainer):
@@ -356,7 +349,7 @@ class PluginManager:
         Returns:
         object: The created optimizer, or None if none was found.
         """
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             optimizer = plugin.create_optimizer(cfg, trainer)
             if optimizer is not None:
                 return optimizer
@@ -374,7 +367,7 @@ class PluginManager:
         Returns:
         object: The created learning rate scheduler, or None if none was found.
         """
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             scheduler = plugin.create_lr_scheduler(cfg, trainer, optimizer)
             if scheduler is not None:
                 return scheduler
@@ -392,7 +385,7 @@ class PluginManager:
         List[callable]: A list of callback functions to be added to the TrainingArgs.
         """
         callbacks = []
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             callbacks.extend(plugin.add_callbacks_pre_trainer(cfg, model))
         return callbacks
 
@@ -408,7 +401,7 @@ class PluginManager:
         List[callable]: A list of callback functions to be added to the TrainingArgs.
         """
         callbacks = []
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             callbacks.extend(plugin.add_callbacks_post_trainer(cfg, trainer))
         return callbacks
 
@@ -423,5 +416,5 @@ class PluginManager:
         Returns:
         None
         """
-        for plugin in self.plugins.values():
+        for plugin in self.plugins:
             plugin.post_train_unload(cfg)

src/axolotl/utils/config/models/input/v0_4_1/__init__.py

@@ -427,6 +427,7 @@ class HyperparametersConfig(BaseModel):
                 "ao_adamw_4bit",
                 "ao_adamw_8bit",
                 "ao_adamw_fp8",
+                "soap",
             ],
         ]
     ] = OptimizerNames.ADAMW_HF.value
@@ -439,6 +440,10 @@ class HyperparametersConfig(BaseModel):
             "help": "The target modules to optimize, i.e. the module names that you would like to train."
         },
     )
+
+    optim_soap_beta1: Optional[float] = None
+    optim_soap_beta2: Optional[float] = None
+
     torchdistx_path: Optional[str] = None
     lr_scheduler: Optional[Union[SchedulerType, Literal["one_cycle"]]] = "cosine"
     lr_scheduler_kwargs: Optional[Dict[str, Any]] = None
@@ -588,9 +593,6 @@ class AxolotlInputConfig(
 
     rl: Optional[RLType] = None
     reward_model: Optional[bool] = None
-    dpo_use_weighting: Optional[
-        bool
-    ] = None  # whether to use weighting in DPO trainer. If none, default is false in the trainer.
 
     datasets: Optional[conlist(Union[SFTDataset, DPODataset, KTODataset], min_length=1)] = None  # type: ignore
     test_datasets: Optional[conlist(Union[SFTDataset, DPODataset, KTODataset], min_length=1)] = None  # type: ignore

src/axolotl/utils/data/sft.py

@@ -2,11 +2,9 @@
 
 import functools
 import logging
-import time
 from pathlib import Path
 from typing import List, Optional, Tuple, Union
 
-import requests
 from datasets import (
     Dataset,
     DatasetDict,
@@ -55,28 +53,6 @@ from axolotl.utils.trainer import (
 LOG = logging.getLogger("axolotl")
 
 
-def retry_on_request_exceptions(max_retries=3, delay=1):
-    def decorator(func):
-        @functools.wraps(func)
-        def wrapper(*args, **kwargs):  # pylint: disable=inconsistent-return-statements
-            for attempt in range(max_retries):
-                try:
-                    return func(*args, **kwargs)
-                except (
-                    requests.exceptions.ReadTimeout,
-                    requests.exceptions.ConnectionError,
-                ) as exc:
-                    if attempt < max_retries - 1:
-                        time.sleep(delay)
-                    else:
-                        raise exc
-
-        return wrapper
-
-    return decorator
-
-
-@retry_on_request_exceptions(max_retries=3, delay=5)
 def prepare_dataset(cfg, tokenizer, processor=None):
     prompters = []
     if not cfg.pretraining_dataset:

src/axolotl/utils/optimizers/soap/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2024 Nikhil Vyas
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

src/axolotl/utils/optimizers/soap/__init__.py

@@ -0,0 +1,475 @@
+# pylint: skip-file
+# Copied from https://github.com/nikhilvyas/SOAP
+from itertools import chain
+
+import torch
+import torch.optim as optim
+
+# Parts of the code are modifications of Pytorch's AdamW optimizer
+# Parts of the code are modifications of code from https://github.com/jiaweizzhao/GaLore/blob/master/galore_torch/galore_projector.py
+
+
+class SOAP(optim.Optimizer):
+    """
+    Implements SOAP algorithm (https://arxiv.org/abs/2409.11321).
+
+    Parameters:
+        params (`Iterable[nn.parameter.Parameter]`):
+            Iterable of parameters to optimize or dictionaries defining parameter groups.
+        lr (`float`, *optional*, defaults to 0.003):
+            The learning rate to use.
+        betas (`Tuple[float,float]`, *optional*, defaults to `(0.95, 0.95)`):
+            Adam's betas parameters (b1, b2).
+        shampoo_beta (`float`, *optional*, defaults to -1):
+            If >= 0, use this beta for the preconditioner (L and R in paper, state['GG'] below) moving average instead of betas[1].
+        eps (`float`, *optional*, defaults to 1e-08):
+            Adam's epsilon for numerical stability.
+        weight_decay (`float`, *optional*, defaults to 0.01): weight decay coefficient.
+        precondition_frequency (`int`, *optional*, defaults to 10):
+            How often to update the preconditioner.
+        max_precond_dim (`int`, *optional*, defaults to 10000):
+            Maximum dimension of the preconditioner.
+            Set to 10000, so that we exclude most common vocab sizes while including layers.
+        merge_dims (`bool`, *optional*, defaults to `False`):
+            Whether or not to merge dimensions of the preconditioner.
+        precondition_1d (`bool`, *optional*, defaults to `False`):
+            Whether or not to precondition 1D gradients.
+        normalize_grads (`bool`, *optional*, defaults to `False`):
+            Whether or not to normalize gradients per layer.
+            Helps at large precondition_frequency (~100 in our experiments),
+            but hurts performance at small precondition_frequency (~10 in our experiments).
+        data_format (`str`, *optional*, defaults to `channels_first`):
+            Data format of the input for convolutional layers.
+            Should be "channels_last" for data_format of NHWC and "channels_first" for NCHW.
+        correct_bias (`bool`, *optional*, defaults to `True`):
+            Whether or not to use bias correction in Adam.
+    """
+
+    def __init__(
+        self,
+        params,
+        lr: float = 3e-3,
+        betas=(0.95, 0.95),
+        shampoo_beta: float = -1,
+        eps: float = 1e-8,
+        weight_decay: float = 0.01,
+        precondition_frequency: int = 10,
+        max_precond_dim: int = 10000,  #
+        merge_dims: bool = False,  # Merge dimensions till the product of the dimensions is less than or equal to max_precond_dim.
+        precondition_1d: bool = False,
+        normalize_grads: bool = False,
+        data_format: str = "channels_first",
+        correct_bias: bool = True,
+    ):
+        defaults = {
+            "lr": lr,
+            "betas": betas,
+            "shampoo_beta": shampoo_beta,
+            "eps": eps,
+            "weight_decay": weight_decay,
+            "precondition_frequency": precondition_frequency,
+            "max_precond_dim": max_precond_dim,
+            "merge_dims": merge_dims,
+            "precondition_1d": precondition_1d,
+            "normalize_grads": normalize_grads,
+            "correct_bias": correct_bias,
+        }
+        super().__init__(params, defaults)
+        self._data_format = data_format
+
+    def merge_dims(self, grad, max_precond_dim):
+        """
+        Merges dimensions of the gradient tensor till the product of the dimensions is less than or equal to max_precond_dim.
+        """
+        assert self._data_format in ["channels_first", "channels_last"]
+        if self._data_format == "channels_last" and grad.dim() == 4:
+            grad = grad.permute(0, 3, 1, 2)
+        shape = grad.shape
+        new_shape = []
+
+        curr_shape = 1
+        for sh in shape:
+            temp_shape = curr_shape * sh
+            if temp_shape > max_precond_dim:
+                if curr_shape > 1:
+                    new_shape.append(curr_shape)
+                    curr_shape = sh
+                else:
+                    new_shape.append(sh)
+                    curr_shape = 1
+            else:
+                curr_shape = temp_shape
+
+        if curr_shape > 1 or len(new_shape) == 0:
+            new_shape.append(curr_shape)
+
+        new_grad = grad.reshape(new_shape)
+        return new_grad
+
+    @torch.no_grad()
+    def step(self):
+        """
+        Performs a single optimization step.
+
+        Arguments:
+            closure (`Callable`, *optional*): A closure that reevaluates the model and returns the loss.
+        """
+        loss = None
+
+        for group in self.param_groups:
+            for p in group["params"]:
+                if p.grad is None:
+                    continue
+                grad = p.grad
+
+                state = self.state[p]
+
+                if "step" not in state:
+                    state["step"] = 0
+
+                    # State initialization
+                if "exp_avg" not in state:
+                    # Exponential moving average of gradient values
+                    state["exp_avg"] = torch.zeros_like(grad)
+                    # Exponential moving average of squared gradient values
+                    state["exp_avg_sq"] = torch.zeros_like(grad)
+
+                if "Q" not in state:
+                    self.init_preconditioner(
+                        grad,
+                        state,
+                        precondition_frequency=group["precondition_frequency"],
+                        precondition_1d=group["precondition_1d"],
+                        shampoo_beta=(
+                            group["shampoo_beta"]
+                            if group["shampoo_beta"] >= 0
+                            else group["betas"][1]
+                        ),
+                        max_precond_dim=group["max_precond_dim"],
+                        merge_dims=group["merge_dims"],
+                    )
+                    self.update_preconditioner(
+                        grad,
+                        state,
+                        max_precond_dim=group["max_precond_dim"],
+                        merge_dims=group["merge_dims"],
+                        precondition_1d=group["precondition_1d"],
+                    )
+                    continue  # first step is skipped so that we never use the current gradients in the projection.
+
+                # Projecting gradients to the eigenbases of Shampoo's preconditioner
+                # i.e. projecting to the eigenbases of matrices in state['GG']
+                grad_projected = self.project(
+                    grad,
+                    state,
+                    merge_dims=group["merge_dims"],
+                    max_precond_dim=group["max_precond_dim"],
+                )
+
+                exp_avg, exp_avg_sq = state["exp_avg"], state["exp_avg_sq"]
+                beta1, beta2 = group["betas"]
+
+                state["step"] += 1
+
+                # Decay the first and second moment running average coefficient
+                # In-place operations to update the averages at the same time
+                exp_avg.mul_(beta1).add_(grad, alpha=(1.0 - beta1))
+                exp_avg_sq.mul_(beta2).add_(
+                    grad_projected.square(), alpha=(1.0 - beta2)
+                )
+
+                denom = exp_avg_sq.sqrt().add_(group["eps"])
+
+                # Projecting the exponential moving average of gradients to the eigenbases of Shampoo's preconditioner
+                # i.e. projecting to the eigenbases of matrices in state['GG']
+                exp_avg_projected = self.project(
+                    exp_avg,
+                    state,
+                    merge_dims=group["merge_dims"],
+                    max_precond_dim=group["max_precond_dim"],
+                )
+
+                step_size = group["lr"]
+                if group["correct_bias"]:
+                    bias_correction1 = 1.0 - beta1 ** (state["step"])
+                    bias_correction2 = 1.0 - beta2 ** (state["step"])
+                    step_size = step_size * (bias_correction2**0.5) / bias_correction1
+
+                # Projecting back the preconditioned (by Adam) exponential moving average of gradients
+                # to the original space
+                norm_grad = self.project_back(
+                    exp_avg_projected / denom,
+                    state,
+                    merge_dims=group["merge_dims"],
+                    max_precond_dim=group["max_precond_dim"],
+                )
+
+                if group["normalize_grads"]:
+                    norm_grad = norm_grad / (1e-30 + torch.mean(norm_grad**2) ** 0.5)
+
+                p.add_(norm_grad, alpha=-step_size)
+
+                # From AdamW code: Just adding the square of the weights to the loss function is *not*
+                # the correct way of using L2 regularization/weight decay with Adam,
+                # since that will interact with the m and v parameters in strange ways.
+                #
+                # Instead we want to decay the weights in a manner that doesn't interact
+                # with the m/v parameters. This is equivalent to adding the square
+                # of the weights to the loss with plain (non-momentum) SGD.
+                # Add weight decay at the end (fixed version)
+                if group["weight_decay"] > 0.0:
+                    p.add_(p, alpha=(-group["lr"] * group["weight_decay"]))
+
+                # Update is done after the gradient step to avoid using current gradients in the projection.
+                self.update_preconditioner(
+                    grad,
+                    state,
+                    max_precond_dim=group["max_precond_dim"],
+                    merge_dims=group["merge_dims"],
+                    precondition_1d=group["precondition_1d"],
+                )
+
+        return loss
+
+    def init_preconditioner(
+        self,
+        grad,
+        state,
+        precondition_frequency=10,
+        shampoo_beta=0.95,
+        max_precond_dim=10000,
+        precondition_1d=False,
+        merge_dims=False,
+    ):
+        """
+        Initializes the preconditioner matrices (L and R in the paper).
+        """
+        state[
+            "GG"
+        ] = []  # Will hold all the preconditioner matrices (L and R in the paper).
+        if grad.dim() == 1:
+            if not precondition_1d or grad.shape[0] > max_precond_dim:
+                state["GG"].append([])
+            else:
+                state["GG"].append(
+                    torch.zeros(grad.shape[0], grad.shape[0], device=grad.device)
+                )
+        else:
+            if merge_dims:
+                grad = self.merge_dims(grad, max_precond_dim)
+
+            for sh in grad.shape:
+                if sh > max_precond_dim:
+                    state["GG"].append([])
+                else:
+                    state["GG"].append(torch.zeros(sh, sh, device=grad.device))
+
+        state["Q"] = None  # Will hold all the eigenbases of the preconditioner.
+        state["precondition_frequency"] = precondition_frequency
+        state["shampoo_beta"] = shampoo_beta
+
+    def project(self, grad, state, merge_dims=False, max_precond_dim=10000):
+        """
+        Projects the gradient to the eigenbases of the preconditioner.
+        """
+        original_shape = grad.shape
+        if merge_dims:
+            if grad.dim() == 4 and self._data_format == "channels_last":
+                permuted_shape = grad.permute(0, 3, 1, 2).shape
+            grad = self.merge_dims(grad, max_precond_dim)
+
+        for mat in state["Q"]:
+            if len(mat) > 0:
+                grad = torch.tensordot(
+                    grad,
+                    mat,
+                    dims=[[0], [0]],
+                )
+            else:
+                permute_order = list(range(1, len(grad.shape))) + [0]
+                grad = grad.permute(permute_order)
+
+        if merge_dims:
+            if self._data_format == "channels_last" and len(original_shape) == 4:
+                grad = grad.reshape(permuted_shape).permute(0, 2, 3, 1)
+            else:
+                grad = grad.reshape(original_shape)
+        return grad
+
+    def update_preconditioner(
+        self,
+        grad,
+        state,
+        max_precond_dim=10000,
+        merge_dims=False,
+        precondition_1d=False,
+    ):
+        """
+        Updates the preconditioner matrices and the eigenbases (L, R, Q_L, Q_R in the paper).
+        """
+        if grad.dim() == 1:
+            if precondition_1d and grad.shape[0] <= max_precond_dim:
+                state["GG"][0].lerp_(
+                    grad.unsqueeze(1) @ grad.unsqueeze(0), 1 - state["shampoo_beta"]
+                )
+        else:
+            if merge_dims:
+                new_grad = self.merge_dims(grad, max_precond_dim)
+                for idx, sh in enumerate(new_grad.shape):
+                    if sh <= max_precond_dim:
+                        outer_product = torch.tensordot(
+                            new_grad,
+                            new_grad,
+                            dims=[
+                                [
+                                    *chain(
+                                        range(idx), range(idx + 1, len(new_grad.shape))
+                                    )
+                                ]
+                            ]
+                            * 2,
+                        )
+                        state["GG"][idx].lerp_(outer_product, 1 - state["shampoo_beta"])
+            else:
+                for idx, sh in enumerate(grad.shape):
+                    if sh <= max_precond_dim:
+                        outer_product = torch.tensordot(
+                            grad,
+                            grad,
+                            # Contracts across all dimensions except for k.
+                            dims=[[*chain(range(idx), range(idx + 1, len(grad.shape)))]]
+                            * 2,
+                        )
+                        state["GG"][idx].lerp_(outer_product, 1 - state["shampoo_beta"])
+
+        if state["Q"] is None:
+            state["Q"] = self.get_orthogonal_matrix(state["GG"])
+        if state["step"] > 0 and state["step"] % state["precondition_frequency"] == 0:
+            state["Q"] = self.get_orthogonal_matrix_QR(
+                state, max_precond_dim, merge_dims
+            )
+
+    def project_back(self, grad, state, merge_dims=False, max_precond_dim=10000):
+        """
+        Projects the gradient back to the original space.
+        """
+        original_shape = grad.shape
+        if merge_dims:
+            if self._data_format == "channels_last" and grad.dim() == 4:
+                permuted_shape = grad.permute(0, 3, 1, 2).shape
+            grad = self.merge_dims(grad, max_precond_dim)
+        for mat in state["Q"]:
+            if len(mat) > 0:
+                grad = torch.tensordot(
+                    grad,
+                    mat,
+                    dims=[[0], [1]],
+                )
+            else:
+                permute_order = list(range(1, len(grad.shape))) + [0]
+                grad = grad.permute(permute_order)
+
+        if merge_dims:
+            if self._data_format == "channels_last" and len(original_shape) == 4:
+                grad = grad.reshape(permuted_shape).permute(0, 2, 3, 1)
+            else:
+                grad = grad.reshape(original_shape)
+        return grad
+
+    def get_orthogonal_matrix(self, mat):
+        """
+        Computes the eigenbases of the preconditioner using torch.linalg.eigh decomposition.
+        """
+        matrix = []
+        for m in mat:
+            if len(m) == 0:
+                matrix.append([])
+                continue
+            if m.data.dtype != torch.float:
+                float_data = False
+                original_type = m.data.dtype
+                original_device = m.data.device
+                matrix.append(m.data.float())
+            else:
+                float_data = True
+                matrix.append(m.data)
+
+        final = []
+        for m in matrix:
+            if len(m) == 0:
+                final.append([])
+                continue
+            try:
+                _, Q = torch.linalg.eigh(
+                    m + 1e-30 * torch.eye(m.shape[0], device=m.device)
+                )
+            except:  # pylint: disable=bare-except # noqa: E722
+                _, Q = torch.linalg.eigh(
+                    m.to(torch.float64) + 1e-30 * torch.eye(m.shape[0], device=m.device)
+                )
+                Q = Q.to(m.dtype)
+            Q = torch.flip(Q, [1])
+
+            if not float_data:
+                Q = Q.to(original_device).type(original_type)
+            final.append(Q)
+        return final
+
+    def get_orthogonal_matrix_QR(self, state, max_precond_dim=10000, merge_dims=False):
+        """
+        Computes the eigenbases of the preconditioner using one round of power iteration
+        followed by torch.linalg.qr decomposition.
+        """
+        precond_list = state["GG"]
+        orth_list = state["Q"]
+
+        matrix = []
+        orth_matrix = []
+        for m, o in zip(precond_list, orth_list):
+            if len(m) == 0:
+                matrix.append([])
+                orth_matrix.append([])
+                continue
+            if m.data.dtype != torch.float:
+                float_data = False
+                original_type = m.data.dtype
+                original_device = m.data.device
+                matrix.append(m.data.float())
+                orth_matrix.append(o.data.float())
+            else:
+                float_data = True
+                matrix.append(m.data.float())
+                orth_matrix.append(o.data.float())
+
+        orig_shape = state["exp_avg_sq"].shape
+        if self._data_format == "channels_last" and len(orig_shape) == 4:
+            permuted_shape = state["exp_avg_sq"].permute(0, 3, 1, 2).shape
+        if merge_dims:
+            exp_avg_sq = self.merge_dims(state["exp_avg_sq"], max_precond_dim)
+        else:
+            exp_avg_sq = state["exp_avg_sq"]
+
+        final = []
+        for ind, (m, o) in enumerate(zip(matrix, orth_matrix)):
+            if len(m) == 0:
+                final.append([])
+                continue
+            est_eig = torch.diag(o.T @ m @ o)
+            sort_idx = torch.argsort(est_eig, descending=True)
+            exp_avg_sq = exp_avg_sq.index_select(ind, sort_idx)
+            o = o[:, sort_idx]
+            power_iter = m @ o
+            Q, _ = torch.linalg.qr(power_iter)
+
+            if not float_data:
+                Q = Q.to(original_device).type(original_type)
+            final.append(Q)
+
+        if merge_dims:
+            if self._data_format == "channels_last" and len(orig_shape) == 4:
+                exp_avg_sq = exp_avg_sq.reshape(permuted_shape).permute(0, 2, 3, 1)
+            else:
+                exp_avg_sq = exp_avg_sq.reshape(orig_shape)
+
+        state["exp_avg_sq"] = exp_avg_sq
+        return final

test.yml

@@ -1,59 +0,0 @@
-base_model: JackFram/llama-68m
-
-load_in_8bit: false
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: arcee-ai/distilabel-intel-orca-dpo-pairs-binarized
-    type: chatml.ultra
-    split: train
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.1
-output_dir: ./outputs/out
-
-sequence_len: 2048
-pad_to_sequence_len: true
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 1
-num_epochs: 1
-optimizer: paged_adamw_8bit
-lr_scheduler: cosine
-learning_rate: 2e-5
-
-train_on_inputs: false
-group_by_length: false
-bf16: auto
-fp16:
-tf32: false
-
-gradient_checkpointing: true
-gradient_checkpointing_kwargs:
-  use_reentrant: false
-early_stopping_patience:
-resume_from_checkpoint:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-rl: dpo
-dpo_use_weighting: true
-
-warmup_steps: 10
-evals_per_epoch: 2
-eval_table_size:
-saves_per_epoch: 1
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  pad_token: <|end_of_text|>

test2.yml

@@ -1,43 +0,0 @@
-base_model: JackFram/llama-68m
-
-load_in_8bit: true
-
-datasets:
-  - path: arcee-ai/distilabel-intel-orca-dpo-pairs-binarized
-    type: chatml.ultra
-    split: train
-output_dir: ./outputs/lora-out
-
-sequence_len: 1024
-
-adapter: lora
-lora_r: 64
-lora_alpha: 32
-lora_dropout: 0.1
-lora_target_linear: true
-
-rl: dpo
-dpo_use_weighting: true
-
-wandb_project: check_dpotrainer
-wandb_entity: axolotl-ai
-wandb_watch:
-wandb_name: baseline/dpo_base/dpo_use_weighting
-wandb_log_model:
-
-
-num_epochs: 1
-micro_batch_size: 4
-gradient_accumulation_steps: 1
-learning_rate: 0.00001
-optimizer: paged_adamw_8bit
-lr_scheduler: cosine
-max_steps": 20
-save_steps: 10
-warmup_steps: 5
-gradient_checkpointing: True
-gradient_checkpointing_kwargs:
-  use_reentrant: false
-#special_tokens:
-#  pad_token: <|end_of_text|>
-

tests/e2e/test_dpo.py

@@ -115,51 +115,6 @@ class TestDPOLlamaLora(unittest.TestCase):
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
         assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
 
-    @with_temp_dir
-    def test_dpo_use_weighting(self, temp_dir):
-        # pylint: disable=duplicate-code
-        cfg = DictDefault(
-            {
-                "base_model": "JackFram/llama-68m",
-                "tokenizer_type": "LlamaTokenizer",
-                "sequence_len": 1024,
-                "load_in_8bit": True,
-                "adapter": "lora",
-                "lora_r": 64,
-                "lora_alpha": 32,
-                "lora_dropout": 0.1,
-                "lora_target_linear": True,
-                "special_tokens": {},
-                "rl": "dpo",
-                "dpo_use_weighting": True,
-                "datasets": [
-                    {
-                        "path": "arcee-ai/distilabel-intel-orca-dpo-pairs-binarized",
-                        "type": "chatml.ultra",
-                        "split": "train",
-                    },
-                ],
-                "num_epochs": 1,
-                "micro_batch_size": 4,
-                "gradient_accumulation_steps": 1,
-                "output_dir": temp_dir,
-                "learning_rate": 0.00001,
-                "optimizer": "paged_adamw_8bit",
-                "lr_scheduler": "cosine",
-                "max_steps": 20,
-                "save_steps": 10,
-                "warmup_steps": 5,
-                "gradient_checkpointing": True,
-                "gradient_checkpointing_kwargs": {"use_reentrant": True},
-            }
-        )
-        normalize_config(cfg)
-        cli_args = TrainerCliArgs()
-        dataset_meta = load_rl_datasets(cfg=cfg, cli_args=cli_args)
-
-        train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
-
     @pytest.mark.skip("kto_pair no longer supported in trl")
     @with_temp_dir
     def test_kto_pair_lora(self, temp_dir):

tests/e2e/test_optimizers.py

@@ -65,3 +65,44 @@ class TestCustomOptimizers(unittest.TestCase):
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
         assert (Path(temp_dir) / "adapter_model.bin").exists()
+
+    @with_temp_dir
+    def test_soap(self, temp_dir):
+        # pylint: disable=duplicate-code
+        cfg = DictDefault(
+            {
+                "base_model": "HuggingFaceTB/SmolLM-135M",
+                "sequence_len": 1024,
+                "load_in_8bit": True,
+                "adapter": "lora",
+                "lora_r": 8,
+                "lora_alpha": 16,
+                "lora_dropout": 0.05,
+                "lora_target_linear": True,
+                "val_set_size": 0.1,
+                "special_tokens": {
+                    "pad_token": "<|endoftext|>",
+                },
+                "datasets": [
+                    {
+                        "path": "vicgalle/alpaca-gpt4",
+                        "type": "alpaca",
+                    },
+                ],
+                "num_epochs": 1,
+                "micro_batch_size": 8,
+                "gradient_accumulation_steps": 1,
+                "output_dir": temp_dir,
+                "learning_rate": 0.00001,
+                "optimizer": "soap",
+                "optim_soap_beta1": 0.95,
+                "optim_soap_beta2": 0.95,
+                "lr_scheduler": "cosine",
+            }
+        )
+        normalize_config(cfg)
+        cli_args = TrainerCliArgs()
+        dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
+
+        train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
+        assert (Path(temp_dir) / "adapter_model.bin").exists()