diff --git a/trlx/data/configs.py b/trlx/data/configs.py index a94073107..0f3610789 100644 --- a/trlx/data/configs.py +++ b/trlx/data/configs.py @@ -231,6 +231,8 @@ class TrainConfig: minibatch_size: Optional[int] = None + reward_only_in_main_process: bool = True + @classmethod def from_dict(cls, config: Dict[str, Any]): return cls(**config) diff --git a/trlx/trainer/accelerate_base_trainer.py b/trlx/trainer/accelerate_base_trainer.py index e15fb06da..3a8456427 100644 --- a/trlx/trainer/accelerate_base_trainer.py +++ b/trlx/trainer/accelerate_base_trainer.py @@ -387,19 +387,23 @@ def evaluate(self): # noqa: C901 if self.config.model.model_arch_type == "seq2seq": samples = samples[:, 1:].contiguous() - prompt_sizes = torch.tensor(prompts.input_ids.shape[1]).repeat(len(prompts.input_ids)) - prompts, samples, prompt_sizes = self.accelerator.gather_for_metrics( - self.accelerator.pad_across_processes( - [prompts.input_ids, samples, prompt_sizes.to(samples.device)], - dim=1, - pad_index=self.tokenizer.pad_token_id, - ) + prompt_sizes = torch.tensor(prompts.input_ids.shape[1], device=samples.device).repeat( + len(prompts.input_ids) ) + if self.config.train.reward_only_in_main_process: + prompts, samples, prompt_sizes = self.accelerator.gather_for_metrics( + self.accelerator.pad_across_processes( + [prompts.input_ids, samples, prompt_sizes], + dim=1, + pad_index=self.tokenizer.pad_token_id, + ) + ) + metadata = gather_dict(metadata, self.accelerator.gradient_state) + else: + prompts = prompts.input_ids all_samples.extend(samples.tolist()) all_prompts.extend(prompts.tolist()) all_prompt_sizes.extend(prompt_sizes.tolist()) - - metadata = gather_dict(metadata, self.accelerator.gradient_state) all_metadata.append(metadata) desc = [ @@ -412,11 +416,16 @@ def evaluate(self): # noqa: C901 stats["time/generate"] = time() - generate_time - if self.accelerator.is_main_process: + if not self.config.train.reward_only_in_main_process or self.accelerator.is_main_process: str_samples, str_prompts, str_outputs = self.decode(all_prompts, all_samples, all_prompt_sizes) - columns = ["prompt", "output"] + if self.accelerator.is_main_process: + columns = ["prompt", "output"] + + # gather should be invoked in every process, not just the main process columns_data = [str_prompts, str_outputs] + if not self.config.train.reward_only_in_main_process: + columns_data = self.accelerator.gather_for_metrics(columns_data) metadata, *xs = all_metadata for k in metadata: @@ -439,41 +448,54 @@ def evaluate(self): # noqa: C901 rewards = torch.tensor([sum(reward) for reward in rewards], dtype=float) else: rewards = torch.tensor(rewards, dtype=float) - mean_reward = rewards.mean().item() - columns.append("reward") - if not isinstance(rewards, list): - rewards = rewards.tolist() - columns_data.append(rewards) - stats[f"reward/mean{sweep_suffix}"] = mean_reward + + # gather should be invoked in every process, not just the main process + if not self.config.train.reward_only_in_main_process: + rewards = self.accelerator.gather(rewards) + + if self.accelerator.is_main_process: + mean_reward = rewards.mean().item() + + columns.append("reward") + if not isinstance(rewards, list): + rewards = rewards.tolist() + columns_data.append(rewards) + stats[f"reward/mean{sweep_suffix}"] = mean_reward # additionally log any other metrics if self.metric_fn: logger.info("Computing metrics") metric_time = time() metrics = self.metric_fn(samples=str_samples, prompts=str_prompts, outputs=str_outputs, **metadata) - stats["time/metric"] = time() - metric_time + if not self.config.train.reward_only_in_main_process: + metrics = self.accelerator.gather_for_metrics(metrics) - mean_metrics = { - f"metrics/{k}{sweep_suffix}": torch.as_tensor(xs).mean(-1).item() for k, xs in metrics.items() - } + if self.accelerator.is_main_process: + stats["time/metric"] = time() - metric_time - stats.update(mean_metrics) + mean_metrics = { + f"metrics/{k}{sweep_suffix}": torch.as_tensor(xs).mean(-1).item() + for k, xs in metrics.items() + } - for metric, values in metrics.items(): - # Skip metrics that are scalers since they represent aggregated values - if isinstance(values, float): - continue - columns.append(metric) - if not isinstance(values, list): - values = values.tolist() - columns_data.append(values) + stats.update(mean_metrics) + + for metric, values in metrics.items(): + # Skip metrics that are scalers since they represent aggregated values + if isinstance(values, float): + continue + columns.append(metric) + if not isinstance(values, list): + values = values.tolist() + columns_data.append(values) # Prepend the sweep argument along with samples - if self.generate_sweep_kwarg: - columns.insert(0, gen_sweep_arg) - columns_data.insert(0, [gen_sweep_value] * len(samples)) + if self.accelerator.is_main_process: + if self.generate_sweep_kwarg: + columns.insert(0, gen_sweep_arg) + columns_data.insert(0, [gen_sweep_value] * len(samples)) - table.append(list(zip(*columns_data))) + table.append(list(zip(*columns_data))) # Log and display evaluation metrics logger.info("Summarizing evaluation") diff --git a/trlx/trainer/accelerate_ppo_trainer.py b/trlx/trainer/accelerate_ppo_trainer.py index 1a4801aaf..1060582e9 100644 --- a/trlx/trainer/accelerate_ppo_trainer.py +++ b/trlx/trainer/accelerate_ppo_trainer.py @@ -289,18 +289,25 @@ def make_experience(self, num_rollouts: int = 1024, iter_count: int = 0): # noq device = samples.device prompt_sizes = torch.tensor([prompt_tensors.shape[1]] * len(prompt_tensors), device=device) - padded_samples = self.accelerator.pad_across_processes( - samples, dim=1, pad_index=self.tokenizer.eos_token_id, pad_first=False - ) - padded_prompts = self.accelerator.pad_across_processes( - prompt_tensors, dim=1, pad_index=self.tokenizer.eos_token_id, pad_first=False - ) - gathered_samples = self.accelerator.gather(padded_samples) - gathered_prompts = self.accelerator.gather(padded_prompts) - gathered_prompt_sizes = self.accelerator.gather(prompt_sizes) - metadata = gather_dict({k: v for k, v in batch.items() if k != "input_ids" and k != "attention_mask"}) + metadata = {k: v for k, v in batch.items() if k != "input_ids" and k != "attention_mask"} + + if self.config.train.reward_only_in_main_process: + padded_samples = self.accelerator.pad_across_processes( + samples, dim=1, pad_index=self.tokenizer.eos_token_id, pad_first=False + ) + padded_prompts = self.accelerator.pad_across_processes( + prompt_tensors, dim=1, pad_index=self.tokenizer.eos_token_id, pad_first=False + ) + gathered_samples = self.accelerator.gather(padded_samples) + gathered_prompts = self.accelerator.gather(padded_prompts) + gathered_prompt_sizes = self.accelerator.gather(prompt_sizes) + metadata = gather_dict(metadata) + else: + gathered_samples = samples + gathered_prompts = prompt_tensors + gathered_prompt_sizes = prompt_sizes - if self.accelerator.is_main_process: + if not self.config.train.reward_only_in_main_process or self.accelerator.is_main_process: all_str_samples, all_str_prompts, all_str_outputs = self.decode( gathered_prompts, gathered_samples, gathered_prompt_sizes, append_eos_token=True ) @@ -316,9 +323,9 @@ def make_experience(self, num_rollouts: int = 1024, iter_count: int = 0): # noq **metadata, ) all_scores = [ - torch.tensor(score, dtype=torch.float, device=device).view( - -1, - ) + score.view(-1) + if isinstance(score, torch.Tensor) + else torch.tensor(score, dtype=torch.float, device=device).view(-1) for score in all_scores ] # Pad 0 reward on the ends @@ -327,20 +334,29 @@ def make_experience(self, num_rollouts: int = 1024, iter_count: int = 0): # noq stats["time/rollout_score"] = time() - rollout_score_time - all_scores = list(all_scores.reshape(self.accelerator.num_processes, -1, max_len).unbind()) + if self.config.train.reward_only_in_main_process: + all_scores = list(all_scores.reshape(self.accelerator.num_processes, -1, max_len).unbind()) else: all_scores = None max_len = torch.tensor(0, dtype=torch.long, device=device) - if torch.distributed.is_initialized(): - torch.distributed.broadcast(max_len, 0) - scores = torch.empty((len(samples), max_len), device=device) - torch.distributed.scatter(scores, all_scores) + if self.config.train.reward_only_in_main_process: + if torch.distributed.is_initialized(): + torch.distributed.broadcast(max_len, 0) + scores = torch.empty((len(samples), max_len), device=device) + torch.distributed.scatter(scores, all_scores) # scores is one shard of one process after scatter + else: + scores = all_scores[0].clone().detach() # shard of one process else: - scores = all_scores[0].clone().detach() + scores = all_scores.clone().detach() # shard of one process + # `all_scores` no longer used, no need to gather it scores_mask = scores != -np.inf - str_samples, str_prompts, str_outputs = self.decode(prompt_tensors, samples, append_eos_token=True) + if self.config.train.reward_only_in_main_process: + _, _, str_outputs = self.decode(prompt_tensors, samples, append_eos_token=True) + else: + str_outputs = all_str_outputs + # `all_str_outputs` no longer used, no need to gather it # Pad the sample outputs outputs = self.tokenizer(str_outputs).input_ids