Nash-MD Trainer

Overview

Nash-MD was proposed in the paper Nash Learning from Human Feedback by Rémi Munos, Michal Valko, Daniele Calandriello, Mohammad Gheshlaghi Azar, Mark Rowland, Daniel Guo, Yunhao Tang, Matthieu Geist, Thomas Mésnard, and Andrea Michi.

The abstract from the paper is the following:

Reinforcement learning from human feedback (RLHF) has emerged as the main paradigm for aligning large language models (LLMs) with human preferences. Typically, RLHF involves the initial step of learning a reward model from human feedback, often expressed as preferences between pairs of text generations produced by a pre-trained LLM. Subsequently, the LLM’s policy is fine-tuned by optimizing it to maximize the reward model through a reinforcement learning algorithm. However, an inherent limitation of current reward models is their inability to fully represent the richness of human preferences and their dependency on the sampling distribution. In this study, we introduce an alternative pipeline for the fine-tuning of LLMs using pairwise human feedback. Our approach entails the initial learning of a preference model, which is conditioned on two inputs given a prompt, followed by the pursuit of a policy that consistently generates responses preferred over those generated by any competing policy, thus defining the Nash equilibrium of this preference model. We term this approach Nash learning from human feedback (NLHF). In the context of a tabular policy representation, we present a novel algorithmic solution, Nash-MD, founded on the principles of mirror descent. This algorithm produces a sequence of policies, with the last iteration converging to the regularized Nash equilibrium. Additionally, we explore parametric representations of policies and introduce gradient descent algorithms for deep-learning architectures. To demonstrate the effectiveness of our approach, we present experimental results involving the fine-tuning of a LLM for a text summarization task. We believe NLHF offers a compelling avenue for preference learning and policy optimization with the potential of advancing the field of aligning LLMs with human preferences.

This post-training method was contributed by Kashif Rasul and Daniil Tiapkin, Pierre Ménard, Daniele Calandriello and Quentin Gallouédec.

Quick start

This example demonstrates how to train a model using the Nash-MD method. We use the Qwen 0.5B model as the base model and PairRMJudge as a judge. We use the prompts from the UltraFeedback dataset. You can view the prompts in the dataset here:

Below is the script to train the model:

# train_nash_md.py
from datasets import load_dataset
from trl import NashMDConfig, NashMDTrainer, PairRMJudge
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2-0.5B-Instruct")
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-0.5B-Instruct")
judge = PairRMJudge()
train_dataset = load_dataset("trl-lib/ultrafeedback-prompt", split="train")

training_args = NashMDConfig(output_dir="Qwen2-0.5B-NashMD", logging_steps=10)
trainer = NashMDTrainer(
    model=model, judge=judge, args=training_args, processing_class=tokenizer, train_dataset=train_dataset
)
trainer.train()

Execute the script using the following command:

accelerate launch train_nash_md.py

Distributed across 8 GPUs, the training takes approximately 3 hours.

To see how the trained model performs, you can use the TRL Chat CLI.

$ trl chat --model_name_or_path trl-lib/Qwen2-0.5B-NashMD
<quentin_gallouedec>:
What is the best programming language?

<trl-lib/Qwen2-0.5B-NashMD>:
The best programming language depends on personal preference, the complexity of the project, and the specific requirements of the task. Some programming languages that are often recommended include Python, Java, and JavaScript, and there are many other languages to choose from depending on individual needs.

Expected dataset type

Nash-MD requires a prompt-only dataset. The NashMDTrainer supports both conversational and standard dataset format. When provided with a conversational dataset, the trainer will automatically apply the chat template to the dataset.

Usage tips

Use a reward model

Instead of a judge, you can chose to use a reward model — see Reward Bench for a leaderboard of public models you can use. Below is a code example showing how to replace a judge with the trl-lib/Qwen2-0.5B-Reward model:

- from trl import PairRMJudge
+ from transformers import AutoModelForSequenceClassification

- judge = PairRMJudge()
+ reward_model = AutoModelForSequenceClassification.from_pretrained("trl-lib/Qwen2-0.5B-Reward", num_labels=1)

  trainer = NashMDTrainer(
      ...
-     judge=judge,
+     reward_model=reward_model,
  )

Make sure that the SFT model and reward model use the same chat template and the same tokenizer. Otherwise, you may find the model completions are scored incorrectly during training.

Encourage EOS token generation

We may want the model to generate completions within a given length. During training, the model will generate completions up to the maximum length specified in the max_new_tokens argument of NashMDConfig. If you want to penalize the model for not generating an EOS token before reaching the maximum length, you can use the missing_eos_penalty argument of NashMDConfig:

training_args = NashMDConfig(..., max_new_tokens=128, missing_eos_penalty=1.0)

Logging Completions

To better understand your model’s behavior during training, you can log sample completions periodically using the LogCompletionsCallback.

trainer = NashMDTrainer(..., eval_dataset=eval_dataset)
completions_callback = LogCompletionsCallback(trainer, num_prompts=8)
trainer.add_callback(completions_callback)

This callback logs the model’s generated completions directly to Weights & Biases.

Logged Completions

Example script

We provide an example script to train a model using the Nash-MD method. The script is available in examples/scripts/nash_md.py

To test the online DPO script with the Qwen2.5 0.5B model on the UltraFeedback dataset, run the following command:

python examples/scripts/nash_md.py \
    --model_name_or_path Qwen/Qwen2.5-0.5B-Instruct \
    --judge pair_rm \
    --dataset_name trl-lib/ultrafeedback-prompt \
    --learning_rate 5.0e-7 \
    --logging_steps 25 \
    --output_dir Qwen2.5-0.5B-NashMD-PairRM \
    --warmup_ratio 0.1 \
    --push_to_hub

Logged metrics

The logged metrics are as follows:

loss/kl: The mean KL divergence between the model and reference data.
objective/entropy: The mean entropy of the model and reference data.
loss/score: The mean reinforce score loss.
rewards/chosen: The mean scores (according to the reward model) of the model completions.
rewards/rejected: The mean scores (according to the reward model) of the mixture completions.
rewards/probabilities: The mean probability (according to the reward model or judge) of the model completions chosen vs the mixture completion.
rewards/accuracies: The accuracies of the Nash-MD’s implicit reward model.
rewards/margins: The mean reward margin (according to reward model) between the chosen and mixture completions.
logps/chosen: The mean log probabilities of the chosen completions.
logps/rejected: The mean log probabilities of the reference completions.
val/model_contain_eos_token: The amount of times the model’s output contains the eos token.
val/ref_contain_eos_token: The amount of times the mixture’s output contains the eos token.
beta: The parameter that controls the weight of the loss term representing the deviation from the reference model. Typically fixed, but can be made dynamic by passing a list to NashMDConfig.
mixture_coef: Logit mixture coefficient for the model and reference model. Typically fixed, but can be made dynamic by passing a list to NashMDConfig.

NashMDTrainer

class trl.NashMDTrainer

< source >

( model: typing.Union[transformers.modeling_utils.PreTrainedModel, torch.nn.modules.module.Module] = None ref_model: typing.Union[transformers.modeling_utils.PreTrainedModel, torch.nn.modules.module.Module] = None reward_model: typing.Union[transformers.modeling_utils.PreTrainedModel, torch.nn.modules.module.Module, NoneType] = None judge: typing.Optional[trl.trainer.judges.BasePairwiseJudge] = None args: typing.Optional[trl.trainer.nash_md_config.NashMDConfig] = None data_collator: typing.Optional[typing.Callable] = None train_dataset: typing.Union[datasets.arrow_dataset.Dataset, datasets.iterable_dataset.IterableDataset, NoneType] = None eval_dataset: typing.Union[datasets.arrow_dataset.Dataset, dict[str, datasets.arrow_dataset.Dataset], NoneType] = None processing_class: typing.Union[transformers.tokenization_utils_base.PreTrainedTokenizerBase, transformers.image_processing_utils.BaseImageProcessor, transformers.feature_extraction_utils.FeatureExtractionMixin, transformers.processing_utils.ProcessorMixin, NoneType] = None peft_config: typing.Optional[dict] = None compute_metrics: typing.Optional[typing.Callable[[transformers.trainer_utils.EvalPrediction], dict]] = None callbacks: typing.Optional[list[transformers.trainer_callback.TrainerCallback]] = None optimizers: tuple = (None, None) preprocess_logits_for_metrics: typing.Optional[typing.Callable[[torch.Tensor, torch.Tensor], torch.Tensor]] = None )

Parameters

model (transformers.PreTrainedModel) — The model to train, preferably an AutoModelForCausalLM.
ref_model (PreTrainedModelWrapper) — Hugging Face transformer model with a casual language modelling head. Used for implicit reward computation and loss. If no reference model is provided, the trainer will create a reference model with the same architecture as the model to be optimized.
reward_model (transformers.PreTrainedModel) — The reward model to score completions with, preferably an AutoModelForSequenceClassification.
judge (BasePairwiseJudge) — The judge to use for pairwise comparison of model completions.
args (NashMDConfig) — The NashMD 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.
processing_class (PreTrainedTokenizerBase or BaseImageProcessor or FeatureExtractionMixin or ProcessorMixin, optional) — Processing class used to process the data. If provided, will be used to automatically process the inputs for the model, and it will be saved along the model to make it easier to rerun an interrupted training or reuse the fine-tuned model.
peft_config (dict) — The peft config to use for training.
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.
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.

Initialize NashMDTrainer as a subclass of OnlineDPOConfig.

NashMDConfig

class trl.NashMDConfig

< source >

( output_dir: str overwrite_output_dir: bool = False do_train: bool = False do_eval: bool = False do_predict: bool = False eval_strategy: typing.Union[transformers.trainer_utils.IntervalStrategy, str] = '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: typing.Optional[int] = None per_gpu_eval_batch_size: typing.Optional[int] = None gradient_accumulation_steps: int = 1 eval_accumulation_steps: typing.Optional[int] = None eval_delay: typing.Optional[float] = 0 torch_empty_cache_steps: typing.Optional[int] = None learning_rate: float = 5e-07 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: typing.Union[transformers.trainer_utils.SchedulerType, str] = 'linear' lr_scheduler_kwargs: typing.Union[dict, str, NoneType] = <factory> warmup_ratio: float = 0.0 warmup_steps: int = 0 log_level: typing.Optional[str] = 'passive' log_level_replica: typing.Optional[str] = 'warning' log_on_each_node: bool = True logging_dir: typing.Optional[str] = None logging_strategy: typing.Union[transformers.trainer_utils.IntervalStrategy, str] = 'steps' logging_first_step: bool = False logging_steps: float = 500 logging_nan_inf_filter: bool = True save_strategy: typing.Union[transformers.trainer_utils.SaveStrategy, str] = 'steps' save_steps: float = 500 save_total_limit: typing.Optional[int] = None save_safetensors: typing.Optional[bool] = 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: typing.Optional[int] = 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: typing.Optional[bool] = None local_rank: int = -1 ddp_backend: typing.Optional[str] = None tpu_num_cores: typing.Optional[int] = None tpu_metrics_debug: bool = False debug: typing.Union[str, typing.List[transformers.debug_utils.DebugOption]] = '' dataloader_drop_last: bool = False eval_steps: typing.Optional[float] = None dataloader_num_workers: int = 0 dataloader_prefetch_factor: typing.Optional[int] = None past_index: int = -1 run_name: typing.Optional[str] = None disable_tqdm: typing.Optional[bool] = None remove_unused_columns: typing.Optional[bool] = True label_names: typing.Optional[typing.List[str]] = None load_best_model_at_end: typing.Optional[bool] = False metric_for_best_model: typing.Optional[str] = None greater_is_better: typing.Optional[bool] = None ignore_data_skip: bool = False fsdp: typing.Union[typing.List[transformers.trainer_utils.FSDPOption], str, NoneType] = '' fsdp_min_num_params: int = 0 fsdp_config: typing.Union[dict, str, NoneType] = None fsdp_transformer_layer_cls_to_wrap: typing.Optional[str] = None accelerator_config: typing.Union[dict, str, NoneType] = None deepspeed: typing.Union[dict, str, NoneType] = None label_smoothing_factor: float = 0.0 optim: typing.Union[transformers.training_args.OptimizerNames, str] = 'adamw_torch' optim_args: typing.Optional[str] = None adafactor: bool = False group_by_length: bool = False length_column_name: typing.Optional[str] = 'length' report_to: typing.Union[NoneType, str, typing.List[str]] = None ddp_find_unused_parameters: typing.Optional[bool] = None ddp_bucket_cap_mb: typing.Optional[int] = None ddp_broadcast_buffers: typing.Optional[bool] = 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: typing.Optional[str] = None hub_model_id: typing.Optional[str] = None hub_strategy: typing.Union[transformers.trainer_utils.HubStrategy, str] = 'every_save' hub_token: typing.Optional[str] = None hub_private_repo: typing.Optional[bool] = None hub_always_push: bool = False gradient_checkpointing: bool = False gradient_checkpointing_kwargs: typing.Union[dict, str, NoneType] = None include_inputs_for_metrics: bool = False include_for_metrics: typing.List[str] = <factory> eval_do_concat_batches: bool = True fp16_backend: str = 'auto' evaluation_strategy: typing.Union[transformers.trainer_utils.IntervalStrategy, str] = None push_to_hub_model_id: typing.Optional[str] = None push_to_hub_organization: typing.Optional[str] = None push_to_hub_token: typing.Optional[str] = None mp_parameters: str = '' auto_find_batch_size: bool = False full_determinism: bool = False torchdynamo: typing.Optional[str] = None ray_scope: typing.Optional[str] = 'last' ddp_timeout: typing.Optional[int] = 1800 torch_compile: bool = False torch_compile_backend: typing.Optional[str] = None torch_compile_mode: typing.Optional[str] = None dispatch_batches: typing.Optional[bool] = None split_batches: typing.Optional[bool] = None include_tokens_per_second: typing.Optional[bool] = False include_num_input_tokens_seen: typing.Optional[bool] = False neftune_noise_alpha: typing.Optional[float] = None optim_target_modules: typing.Union[NoneType, str, typing.List[str]] = None batch_eval_metrics: bool = False eval_on_start: bool = False use_liger_kernel: typing.Optional[bool] = False eval_use_gather_object: typing.Optional[bool] = False average_tokens_across_devices: typing.Optional[bool] = False reward_model_path: typing.Optional[str] = None judge: typing.Optional[str] = None max_new_tokens: int = 64 temperature: float = 0.9 missing_eos_penalty: typing.Optional[float] = None beta: list = <factory> loss_type: typing.Literal['sigmoid', 'ipo'] = 'sigmoid' dataset_num_proc: typing.Optional[int] = None disable_dropout: bool = True mixture_coef: list = <factory> )

Parameters

mixture_coef (float or list[float], optional, defaults to 0.5) — Logit mixture coefficient for the model and reference model. If a list of floats is provided then the mixture coefficient is selected for each new epoch and the last coefficient is used for the rest of the epochs.

Configuration class for the NashMDTrainer.

Subclass of OnlineDPOConfig we can use all its arguments and add the following:

< > Update on GitHub