SimPO Trainer

Simple Preference Optimization (SimPO) by Yu Meng, Mengzhou Xia, and Danqi Chen proposes a simpler and more effective preference optimization algorithm than DPO without using a reference model. The key designs in SimPO are (1) using length-normalized log likelihood as the implicit reward, and (2) incorporating a target reward margin in the Bradley-Terry ranking objective.

The official code can be found at princeton-nlp/SimPO.

Expected dataset format

The SimPO trainer expects a format identical to the DPO trainer, which should include three entries. These entries should be named as follows:

for example:

simpo_dataset_dict = {
    "prompt": [
        "hello",
        "how are you",
        "What is your name?",
        "What is your name?",
        "Which is the best programming language?",
        "Which is the best programming language?",
        "Which is the best programming language?",
    ],
    "chosen": [
        "hi nice to meet you",
        "I am fine",
        "My name is Mary",
        "My name is Mary",
        "Python",
        "Python",
        "Java",
    ],
    "rejected": [
        "leave me alone",
        "I am not fine",
        "Whats it to you?",
        "I dont have a name",
        "Javascript",
        "C++",
        "C++",
    ],
}

where the prompt contains the context inputs, chosen contains the corresponding chosen responses and rejected contains the corresponding negative (rejected) responses. Note that a prompt can have multiple responses and this is reflected in the entries being repeated in the dictionary’s value arrays.

Expected model format

The SimPO trainer expects a model of AutoModelForCausalLM.

Using the SimPOTrainer

For a detailed example have a look at the examples/scripts/simpo.py script. At a high level we need to initialize the SimPOTrainer with a model we wish to train. Note that SimPOTrainer eliminates the need to use the reference model, simplifying the optimization process.

simpo_config = SimPOConfig(
    beta=2.5,
    gamma_beta_ratio=0.25, # ratio between target reward margin (gamma) and beta
)

simpo_trainer = SimPOTrainer(
    model,
    args=simpo_config,
    train_dataset=train_dataset,
    tokenizer=tokenizer,
)

After this, one can then call:

simpo_trainer.train()

Logging

While training and evaluating we record the following reward metrics:

SimPOTrainer

class trl.SimPOTrainer

< >

( model: Union = None args: Optional = None data_collator: Optional = None train_dataset: Optional = None eval_dataset: Union = None tokenizer: Optional = None model_init: Optional = None callbacks: Optional = None optimizers: Tuple = (None, None) preprocess_logits_for_metrics: Optional = None peft_config: Optional = None compute_metrics: Optional = None )

Parameters

  • model (transformers.PreTrainedModel) — The model to train, preferably an AutoModelForSequenceClassification.
  • args (SimPOConfig) — The SimPO config arguments to use for training.
  • data_collator (transformers.DataCollator) — The data collator to use for training. If None is specified, the default data collator (DPODataCollatorWithPadding) will be used which will pad the sequences to the maximum length of the sequences in the batch, given a dataset of paired sequences.
  • train_dataset (datasets.Dataset) — The dataset to use for training.
  • eval_dataset (datasets.Dataset) — The dataset to use for evaluation.
  • tokenizer (transformers.PreTrainedTokenizerBase) — The tokenizer to use for training. This argument is required if you want to use the default data collator.
  • model_init (Callable[[], transformers.PreTrainedModel]) — The model initializer to use for training. If None is specified, the default model initializer will be used.
  • callbacks (List[transformers.TrainerCallback]) — The callbacks to use for training.
  • optimizers (Tuple[torch.optim.Optimizer, torch.optim.lr_scheduler.LambdaLR]) — The optimizer and scheduler to use for training.
  • preprocess_logits_for_metrics (Callable[[torch.Tensor, torch.Tensor], torch.Tensor]) — The function to use to preprocess the logits before computing the metrics.
  • peft_config (Dict, defaults to None) — The PEFT configuration to use for training. If you pass a PEFT configuration, the model will be wrapped in a PEFT model.
  • compute_metrics (Callable[[EvalPrediction], Dict], optional) — The function to use to compute the metrics. Must take a EvalPrediction and return a dictionary string to metric values.

Initialize SimPOTrainer.

build_tokenized_answer

< >

( prompt answer )

Llama tokenizer does satisfy enc(a + b) = enc(a) + enc(b). It does ensure enc(a + b) = enc(a) + enc(a + b)[len(enc(a)):]. Reference: https://github.com/EleutherAI/lm-evaluation-harness/pull/531#issuecomment-1595586257

concatenated_forward

< >

( model: Module batch: Dict )

Run the given model on the given batch of inputs, concatenating the chosen and rejected inputs together.

We do this to avoid doing two forward passes, because it’s faster for FSDP.

concatenated_inputs

< >

( batch: Dict is_encoder_decoder: bool = False label_pad_token_id: int = -100 padding_value: int = 0 device: Optional = None )

Concatenate the chosen and rejected inputs into a single tensor.

evaluation_loop

< >

( dataloader: DataLoader description: str prediction_loss_only: Optional = None ignore_keys: Optional = None metric_key_prefix: str = 'eval' )

Overriding built-in evaluation loop to store metrics for each batch. Prediction/evaluation loop, shared by Trainer.evaluate() and Trainer.predict().

Works both with or without labels.

get_batch_logps

< >

( logits: FloatTensor labels: LongTensor label_pad_token_id: int = -100 is_encoder_decoder: bool = False )

Compute the log probabilities of the given labels under the given logits.

get_batch_loss_metrics

< >

( model batch: Dict train_eval: Literal = 'train' )

Compute the SimPO loss and other metrics for the given batch of inputs for train or test.

get_batch_samples

< >

( model batch: Dict )

Generate samples from the model and reference model for the given batch of inputs.

log

< >

( logs: Dict )

Parameters

  • logs (Dict[str, float]) — The values to log.

Log logs on the various objects watching training, including stored metrics.

simpo_loss

< >

( policy_chosen_logps: FloatTensor policy_rejected_logps: FloatTensor ) A tuple of three tensors

Returns

A tuple of three tensors

(losses, chosen_rewards, rejected_rewards). The losses tensor contains the SimPO loss for each example in the batch. The chosen_rewards and rejected_rewards tensors contain the rewards for the chosen and rejected responses, respectively.

Compute the SimPO loss for a batch of policy model log probabilities.

tokenize_row

< >

( feature model: Union = None )

Tokenize a single row from a SimPO specific dataset.

At this stage, we don’t convert to PyTorch tensors yet; we just handle the truncation in case the prompt + chosen or prompt + rejected responses is/are too long. First we truncate the prompt; if we’re still too long, we truncate the chosen/rejected.

We also create the labels for the chosen/rejected responses, which are of length equal to the sum of the length of the prompt and the chosen/rejected response, with label_pad_token_id for the prompt tokens.

SimPOConfig

class trl.SimPOConfig

< >

( output_dir: str overwrite_output_dir: bool = False do_train: bool = False do_eval: bool = False do_predict: bool = False eval_strategy: Union = 'no' prediction_loss_only: bool = False per_device_train_batch_size: int = 8 per_device_eval_batch_size: int = 8 per_gpu_train_batch_size: Optional = None per_gpu_eval_batch_size: Optional = None gradient_accumulation_steps: int = 1 eval_accumulation_steps: Optional = None eval_delay: Optional = 0 torch_empty_cache_steps: Optional = None learning_rate: float = 5e-05 weight_decay: float = 0.0 adam_beta1: float = 0.9 adam_beta2: float = 0.999 adam_epsilon: float = 1e-08 max_grad_norm: float = 1.0 num_train_epochs: float = 3.0 max_steps: int = -1 lr_scheduler_type: Union = 'linear' lr_scheduler_kwargs: Union = <factory> warmup_ratio: float = 0.0 warmup_steps: int = 0 log_level: Optional = 'passive' log_level_replica: Optional = 'warning' log_on_each_node: bool = True logging_dir: Optional = None logging_strategy: Union = 'steps' logging_first_step: bool = False logging_steps: float = 500 logging_nan_inf_filter: bool = True save_strategy: Union = 'steps' save_steps: float = 500 save_total_limit: Optional = None save_safetensors: Optional = True save_on_each_node: bool = False save_only_model: bool = False restore_callback_states_from_checkpoint: bool = False no_cuda: bool = False use_cpu: bool = False use_mps_device: bool = False seed: int = 42 data_seed: Optional = None jit_mode_eval: bool = False use_ipex: bool = False bf16: bool = False fp16: bool = False fp16_opt_level: str = 'O1' half_precision_backend: str = 'auto' bf16_full_eval: bool = False fp16_full_eval: bool = False tf32: Optional = None local_rank: int = -1 ddp_backend: Optional = None tpu_num_cores: Optional = None tpu_metrics_debug: bool = False debug: Union = '' dataloader_drop_last: bool = False eval_steps: Optional = None dataloader_num_workers: int = 0 dataloader_prefetch_factor: Optional = None past_index: int = -1 run_name: Optional = None disable_tqdm: Optional = None remove_unused_columns: Optional = True label_names: Optional = None load_best_model_at_end: Optional = False metric_for_best_model: Optional = None greater_is_better: Optional = None ignore_data_skip: bool = False fsdp: Union = '' fsdp_min_num_params: int = 0 fsdp_config: Union = None fsdp_transformer_layer_cls_to_wrap: Optional = None accelerator_config: Union = None deepspeed: Union = None label_smoothing_factor: float = 0.0 optim: Union = 'adamw_torch' optim_args: Optional = None adafactor: bool = False group_by_length: bool = False length_column_name: Optional = 'length' report_to: Union = None ddp_find_unused_parameters: Optional = None ddp_bucket_cap_mb: Optional = None ddp_broadcast_buffers: Optional = None dataloader_pin_memory: bool = True dataloader_persistent_workers: bool = False skip_memory_metrics: bool = True use_legacy_prediction_loop: bool = False push_to_hub: bool = False resume_from_checkpoint: Optional = None hub_model_id: Optional = None hub_strategy: Union = 'every_save' hub_token: Optional = None hub_private_repo: bool = False hub_always_push: bool = False gradient_checkpointing: bool = False gradient_checkpointing_kwargs: Union = None include_inputs_for_metrics: bool = False eval_do_concat_batches: bool = True fp16_backend: str = 'auto' evaluation_strategy: Union = None push_to_hub_model_id: Optional = None push_to_hub_organization: Optional = None push_to_hub_token: Optional = None mp_parameters: str = '' auto_find_batch_size: bool = False full_determinism: bool = False torchdynamo: Optional = None ray_scope: Optional = 'last' ddp_timeout: Optional = 1800 torch_compile: bool = False torch_compile_backend: Optional = None torch_compile_mode: Optional = None dispatch_batches: Optional = None split_batches: Optional = None include_tokens_per_second: Optional = False include_num_input_tokens_seen: Optional = False neftune_noise_alpha: Optional = None optim_target_modules: Union = None batch_eval_metrics: bool = False eval_on_start: bool = False use_liger_kernel: Optional = False eval_use_gather_object: Optional = False beta: float = 2.0 gamma_beta_ratio: float = 0.25 sft_weight: float = 0.0 label_smoothing: float = 0.0 loss_type: Literal = 'sigmoid' label_pad_token_id: int = -100 padding_value: Optional = None truncation_mode: str = 'keep_end' max_length: Optional = None max_prompt_length: Optional = None max_completion_length: Optional = None max_target_length: Optional = None is_encoder_decoder: Optional = None disable_dropout: bool = True generate_during_eval: bool = False dataset_num_proc: Optional = None model_init_kwargs: Optional = None )

Parameters

  • beta (float, optional, defaults to 2.0) — Beta factor in SimPO loss.
  • gamma_beta_ratio (float, optional, defaults to 0.25) — Ratio between the target reward margin (gamma) and beta in SimPO loss.
  • sft_weight (float, optional, defaults to 0.0) — SFT loss weight added to the SimPO loss (0.0 is not using SFT).
  • label_smoothing (float, optional, defaults to 0.0) — Label smoothing factor.
  • loss_type (str, optional, defaults to "sigmoid") — Type of loss to use. Possible values are:

    • "sigmoid": sigmoid loss from the original DPO paper.
    • "hinge": hinge loss on the normalized likelihood from the SLiC paper.
  • label_pad_token_id (int, optional, defaults to -100) — Label pad token id. This argument is required if you want to use the default data collator.
  • padding_value (Optional[int], optional, defaults to None) — Padding value if it is different to the tokenizer’s pad_token_id.
  • truncation_mode (str, optional, defaults to "keep_end") — Truncation mode to use, either "keep_end" or "keep_start". This argument is required if you want to use the default data collator.
  • max_length (Optional[int], optional, defaults to None) — Maximum length of the sequences in the batch. This argument is required if you want to use the default data collator.
  • max_prompt_length (Optional[int], optional, defaults to None) — Maximum length of the prompt. This argument is required if you want to use the default data collator.
  • max_completion_length (Optional[int], optional, defaults to None) — Maximum length of the target. This argument is required if you want to use the default data collator and your model is an encoder-decoder.
  • max_target_length (Optional[int], optional, defaults to None) — Maximum length of the target. This argument is required if you want to use the default data collator and your model is an encoder-decoder.
  • is_encoder_decoder (Optional[bool], optional, defaults to None) — If no model is provided, we need to know if the model_init returns an encoder-decoder.
  • disable_dropout (bool, optional, defaults to True) — Whether or not to disable dropouts in model.
  • generate_during_eval (bool, optional, defaults to False) — Whether to sample and log generations during evaluation step.
  • dataset_num_proc (Optional[int], optional, defaults to None) — Number of processes to use for processing the datasets.
  • model_init_kwargs (Optional[Dict[str, Any]], optional, defaults to None) — Dict of optional kwargs to pass when instantiating the model from a string

SimPOConfig collects all training arguments related to the SimPOTrainer class.

Using HfArgumentParser we can turn this class into argparse arguments that can be specified on the command line.

< > Update on GitHub