diff --git a/applications/Chat/coati/ray/experience_maker_holder.py b/applications/Chat/coati/ray/experience_maker_holder.py index 8551ef1eacef..07d9c3e4f396 100644 --- a/applications/Chat/coati/ray/experience_maker_holder.py +++ b/applications/Chat/coati/ray/experience_maker_holder.py @@ -205,15 +205,15 @@ def update_experience_maker(self, self.experience_maker.actor.model.load_state_dict(new_actor_state_dict, strict=False) else: new_actor_state_dict = state_dict_to(new_actor_state_dict, device=torch.cuda.current_device()) - state_dict_increasae = self.actor_lora_constructor.reconstruct_increase(new_actor_state_dict, new_actor_lora_config_dict) - self.actor_lora_constructor.load_state_dict_increase(self.experience_maker.actor.model, state_dict_increasae) + state_dict_increase = self.actor_lora_constructor.reconstruct_increase(new_actor_state_dict, new_actor_lora_config_dict) + self.actor_lora_constructor.load_state_dict_increase(self.experience_maker.actor.model, state_dict_increase) if new_critic_state_dict is not None: if not self._update_lora_weights or fully_update: self.experience_maker.critic.load_state_dict(new_critic_state_dict, strict=False) else: new_critic_state_dict = state_dict_to(new_critic_state_dict, device=torch.cuda.current_device()) - state_dict_increasae = self.critic_lora_constructor.reconstruct_increase(new_critic_state_dict, new_critic_lora_config_dict) - self.critic_lora_constructor.load_state_dict_increase(self.experience_maker.critic, state_dict_increasae) + state_dict_increase = self.critic_lora_constructor.reconstruct_increase(new_critic_state_dict, new_critic_lora_config_dict) + self.critic_lora_constructor.load_state_dict_increase(self.experience_maker.critic, state_dict_increase) # the lock must be released after both actor and critic being updated if chunk_end: diff --git a/applications/Chat/coati/ray/lora_constructor.py b/applications/Chat/coati/ray/lora_constructor.py index 599a58248728..4809617f647b 100644 --- a/applications/Chat/coati/ray/lora_constructor.py +++ b/applications/Chat/coati/ray/lora_constructor.py @@ -19,7 +19,7 @@ class LoRAConfig: class LoRAConstructor: ''' Tools for reconstructing a model from a remote LoRA model. - (Transfering only LoRA data costs much less!) + (Transferring only LoRA data costs much less!) Usage: Step 1 (Sender): filter_state_dict_lora() @@ -52,7 +52,7 @@ def reconstruct_increase(self, state_dict_lora: Dict[str, Any], lora_config_dict if lora_config_dict is not None: self.register_lora_config(lora_config_dict) - state_dict_increasae = OrderedDict() + state_dict_increase = OrderedDict() config_iter = iter(self.lora_config_dict.items()) lora_A, lora_B, layer_prefix = None, None, None for k, v in state_dict_lora.items(): @@ -65,11 +65,11 @@ def reconstruct_increase(self, state_dict_lora: Dict[str, Any], lora_config_dict assert layer_prefix_2 == layer_prefix, "unmatched (state_dict, config_dict) pair" lora_B = v weight_data_increase = self._compute(lora_A, lora_B, config) - state_dict_increasae[layer_prefix + '.weight'] = weight_data_increase + state_dict_increase[layer_prefix + '.weight'] = weight_data_increase lora_A, lora_B, layer_prefix = None, None, None else: raise ValueError('unexpected key') - return state_dict_increasae + return state_dict_increase def _compute(self, lora_A, lora_B, config=LoRAConfig()): def T(w): @@ -80,12 +80,12 @@ def T(w): return weight_data_increase return 0 - def load_state_dict_increase(self, model: nn.Module, state_dict_increasae: Dict[str, Any]): + def load_state_dict_increase(self, model: nn.Module, state_dict_increase: Dict[str, Any]): ''' The final reconstruction step ''' # naive approach - model.load_state_dict({k: v + model.state_dict()[k] for k, v in state_dict_increasae.items()}, strict=False) + model.load_state_dict({k: v + model.state_dict()[k] for k, v in state_dict_increase.items()}, strict=False) @staticmethod def filter_state_dict_lora(state_dict: Dict[str, Any], keep_non_lora=False): diff --git a/applications/Chat/coati/trainer/strategies/colossalai.py b/applications/Chat/coati/trainer/strategies/colossalai.py index fafd0918deaf..cfdab2806a25 100644 --- a/applications/Chat/coati/trainer/strategies/colossalai.py +++ b/applications/Chat/coati/trainer/strategies/colossalai.py @@ -29,7 +29,7 @@ class ColossalAIStrategy(DDPStrategy): precision(str): The precision to use. Choose in ('fp32', 'fp16'). Stage 3 only supports fp16. seed(int): The seed for the random number generator. shard_init(bool): Whether to shard the model parameters during initialization. Only for ZeRO-3. - This is not compativle with `from_pretrained()`. We temporarily disable this and will support it in the future. + This is not compatible with `from_pretrained()`. We temporarily disable this and will support it in the future. placement_policy(str): The placement policy for gemini. Choose in ('cpu', 'cuda') If it is “cpu”, parameters, gradients and optimizer states will be offloaded to CPU, If it is “cuda”, they will not be offloaded, which means max CUDA memory will be used. It is the fastest. @@ -39,7 +39,7 @@ class ColossalAIStrategy(DDPStrategy): hidden_dim(optional, int): The hidden dimension for the gemini. Only for ZeRO-3. min_chunk_size_mb(float): The minimum chunk size in MB. Only for ZeRO-3. gpu_margin_mem_ratio(float): The margin memory ratio for the GPU. Only for ZeRO-3. - reduce_bugket_size(int): The reduce bucket size in bytes. Only for ZeRO-1 and ZeRO-2. + reduce_bucket_size(int): The reduce bucket size in bytes. Only for ZeRO-1 and ZeRO-2. overlap_communication(bool): Whether to overlap communication and computation. Only for ZeRO-1 and ZeRO-2. initial_scale(float): The initial scale for the optimizer. growth_factor(float): The growth factor for the optimizer.