update llama3 example base models to use nous

* add support for optimi_adamw optimizer w kahan summation

* pydantic validator for optimi_adamw

* workaround for setting optimizer for fsdp

* make sure to install optimizer packages

* make sure to have parity for model parameters passed to optimizer

* add smoke test for optimi_adamw optimizer

* don't use foreach optimi by default

cicd/Dockerfile.jinja

@@ -24,9 +24,9 @@ RUN git fetch origin +$GITHUB_REF && \
 # If AXOLOTL_EXTRAS is set, append it in brackets
 RUN pip install causal_conv1d
 RUN if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
-        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,optimizers,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
     else \
-        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore] $AXOLOTL_ARGS; \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,optimizers] $AXOLOTL_ARGS; \
     fi
 
 # So we can test the Docker image

docker/Dockerfile

@@ -22,9 +22,9 @@ WORKDIR /workspace/axolotl
 # If AXOLOTL_EXTRAS is set, append it in brackets
 RUN pip install causal_conv1d
 RUN if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
-        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,optimizers,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
     else \
-        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore] $AXOLOTL_ARGS; \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,optimizers] $AXOLOTL_ARGS; \
     fi
 
 # So we can test the Docker image

examples/llama-3/fft-8b.yaml

@@ -1,4 +1,4 @@
-base_model: meta-llama/Meta-Llama-3-8B
+base_model: NousResearch/Meta-Llama-3-8B
 model_type: LlamaForCausalLM
 tokenizer_type: AutoTokenizer
 

examples/llama-3/lora-8b.yml

@@ -1,4 +1,4 @@
-base_model: meta-llama/Meta-Llama-3-8B
+base_model: NousResearch/Meta-Llama-3-8B
 model_type: LlamaForCausalLM
 tokenizer_type: AutoTokenizer
 

examples/llama-3/qlora-fsdp-70b.yaml

@@ -1,4 +1,4 @@
-base_model: casperhansen/llama-3-70b-fp16
+base_model: NousResearch/Meta-Llama-3-70B
 model_type: LlamaForCausalLM
 tokenizer_type: AutoTokenizer  # PreTrainedTokenizerFast
 

examples/llama-3/qlora.yml

@@ -1,4 +1,4 @@
-base_model: meta-llama/Meta-Llama-3-8B
+base_model: NousResearch/Meta-Llama-3-8B
 model_type: AutoModelForCausalLM
 tokenizer_type: AutoTokenizer
 
@@ -7,7 +7,7 @@ load_in_4bit: true
 strict: false
 
 datasets:
-  - path: aaditya/alpaca_subset_1
+  - path: tatsu-lab/alpaca
     type: alpaca
 dataset_prepared_path:
 val_set_size: 0

setup.py

@@ -104,5 +104,11 @@ setup(
         "galore": [
             "galore_torch",
         ],
+        "optimizers": [
+            "galore_torch",
+            "lion-pytorch==0.1.2",
+            "lomo-optim==0.1.1",
+            "torch-optimi==0.2.1",
+        ],
     },
 )

src/axolotl/core/trainer_builder.py

@@ -226,6 +226,12 @@ class AxolotlTrainingMixins:
         default=None,
         metadata={"help": "whether to use sequential sampling for curriculum learning"},
     )
+    alternate_optimizer: Optional[str] = field(
+        default=None,
+        metadata={
+            "help": "workaround to pass an alternate optimizer to the HF trainer"
+        },
+    )
 
 
 @dataclass
@@ -285,25 +291,59 @@ class AxolotlTrainer(Trainer):
             self.loss_fct = torch.nn.CrossEntropyLoss(reduction="none")
 
     def create_optimizer(self):
-        if self.args.loraplus_lr_ratio is None:
+        if (
+            self.args.loraplus_lr_ratio is None
+            and self.args.alternate_optimizer != "optimi_adamw"
+        ):
             return super().create_optimizer()
 
         opt_model = self.model_wrapped if is_sagemaker_mp_enabled() else self.model
         if self.optimizer is None:  # pylint: disable=access-member-before-definition
+            decay_parameters = self.get_decay_parameter_names(opt_model)
+            optimizer_grouped_parameters = [
+                {
+                    "params": [
+                        p
+                        for n, p in opt_model.named_parameters()
+                        if (n in decay_parameters and p.requires_grad)
+                    ],
+                    "weight_decay": self.args.weight_decay,
+                },
+                {
+                    "params": [
+                        p
+                        for n, p in opt_model.named_parameters()
+                        if (n not in decay_parameters and p.requires_grad)
+                    ],
+                    "weight_decay": 0.0,
+                },
+            ]
+
             optimizer_cls, optimizer_kwargs = Trainer.get_optimizer_cls_and_kwargs(
                 self.args,
                 opt_model,
             )
 
-            loraplus_lr_ratio = getattr(self.args, "loraplus_lr_ratio", None)
+            if self.args.loraplus_lr_ratio is not None:
-            loraplus_lr_embedding = getattr(self.args, "loraplus_lr_embedding", None)
+                loraplus_lr_ratio = getattr(self.args, "loraplus_lr_ratio", None)
-            self.optimizer = create_loraplus_optimizer(  # pylint: disable=attribute-defined-outside-init
+                loraplus_lr_embedding = getattr(
-                opt_model,
+                    self.args, "loraplus_lr_embedding", None
-                optimizer_cls,
+                )
-                optimizer_kwargs,
+                self.optimizer = create_loraplus_optimizer(  # pylint: disable=attribute-defined-outside-init
-                loraplus_lr_ratio,
+                    opt_model,
-                loraplus_lr_embedding,
+                    optimizer_cls,
-            )
+                    optimizer_kwargs,
+                    loraplus_lr_ratio,
+                    loraplus_lr_embedding,
+                )
+            elif self.args.alternate_optimizer == "optimi_adamw":
+                from optimi import AdamW
+
+                self.optimizer = (  # pylint: disable=attribute-defined-outside-init
+                    AdamW(
+                        optimizer_grouped_parameters, foreach=False, **optimizer_kwargs
+                    )
+                )
 
         if is_sagemaker_mp_enabled():
             self.optimizer = smp.DistributedOptimizer(  # pylint: disable=attribute-defined-outside-init
@@ -1396,6 +1436,11 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
 
         trainer_kwargs = {}
 
+        if self.cfg.optimizer == "optimi_adamw":
+            # Set default so transformers doesn't throw
+            training_arguments_kwargs["optim"] = "adamw_hf"
+            training_arguments_kwargs["alternate_optimizer"] = self.cfg.optimizer
+
         if self.cfg.optimizer == "lion_pytorch":
             from lion_pytorch import Lion
 

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

@@ -341,7 +341,7 @@ class HyperparametersConfig(BaseModel):
     learning_rate: Union[str, float]
     weight_decay: Optional[float] = 0.0
     optimizer: Optional[
-        Union[OptimizerNames, Literal["lion_pytorch"]]
+        Union[OptimizerNames, Literal["lion_pytorch", "optimi_adamw"]]
     ] = OptimizerNames.ADAMW_HF.value
     optim_args: Optional[Union[str, Dict[str, Any]]] = Field(
         default=None, metadata={"help": "Optional arguments to supply to optimizer."}

tests/e2e/test_lora_llama.py

@@ -34,8 +34,8 @@ class TestLoraLlama(unittest.TestCase):
                 "sequence_len": 1024,
                 "load_in_8bit": True,
                 "adapter": "lora",
-                "lora_r": 32,
+                "lora_r": 8,
-                "lora_alpha": 64,
+                "lora_alpha": 16,
                 "lora_dropout": 0.05,
                 "lora_target_linear": True,
                 "val_set_size": 0.1,
@@ -50,7 +50,7 @@ class TestLoraLlama(unittest.TestCase):
                         "type": "alpaca",
                     },
                 ],
-                "num_epochs": 2,
+                "num_epochs": 1,
                 "micro_batch_size": 8,
                 "gradient_accumulation_steps": 1,
                 "output_dir": temp_dir,

tests/e2e/test_optimizers.py

@@ -0,0 +1,67 @@
+"""
+E2E tests for custom optimizers using Llama
+"""
+
+import logging
+import os
+import unittest
+from pathlib import Path
+
+from axolotl.cli import load_datasets
+from axolotl.common.cli import TrainerCliArgs
+from axolotl.train import train
+from axolotl.utils.config import normalize_config
+from axolotl.utils.dict import DictDefault
+
+from .utils import with_temp_dir
+
+LOG = logging.getLogger("axolotl.tests.e2e")
+os.environ["WANDB_DISABLED"] = "true"
+
+
+class TestCustomOptimizers(unittest.TestCase):
+    """
+    Test case for Llama models using LoRA
+    """
+
+    @with_temp_dir
+    def test_optimi_adamw(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": 8,
+                "lora_alpha": 16,
+                "lora_dropout": 0.05,
+                "lora_target_linear": True,
+                "val_set_size": 0.1,
+                "special_tokens": {
+                    "unk_token": "<unk>",
+                    "bos_token": "<s>",
+                    "eos_token": "</s>",
+                },
+                "datasets": [
+                    {
+                        "path": "mhenrichsen/alpaca_2k_test",
+                        "type": "alpaca",
+                    },
+                ],
+                "num_epochs": 1,
+                "micro_batch_size": 8,
+                "gradient_accumulation_steps": 1,
+                "output_dir": temp_dir,
+                "learning_rate": 0.00001,
+                "optimizer": "optimi_adamw",
+                "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()