fix save method of adapter_model.bin

src/llmtuner/tuner/dpo/collator.py

@@ -16,7 +16,7 @@ class DPODataCollatorWithPadding(DataCollatorForSeq2Seq):
             if self.tokenizer.padding_side == "left":
                 start, end = feature.size(0) - answer_len, feature.size(0)
             else:
-                start, end = prompt_len, answer_len
+                start, end = prompt_len, prompt_len + answer_len
             padded_tensor = self.label_pad_token_id * torch.ones_like(feature)
             padded_tensor[start:end] = feature[start:end]
             padded_labels.append(padded_tensor)

src/llmtuner/tuner/dpo/trainer.py

 import torch
 from collections import defaultdict
-from peft import PeftModel
 from typing import TYPE_CHECKING, Dict, Optional, Tuple, Union
 from transformers import BatchEncoding, Trainer
 from trl import DPOTrainer
+# from trl.trainer.utils import disable_dropout_in_model
 
 from llmtuner.extras.constants import IGNORE_INDEX
-from llmtuner.tuner.core.trainer import PeftModelMixin
 
 if TYPE_CHECKING:
     from transformers import PreTrainedModel
-    from llmtuner.hparams import FinetuningArguments
 
 
-class DPOPeftTrainer(PeftModelMixin, DPOTrainer):
+class CustomDPOTrainer(DPOTrainer):
 
     def __init__(
         self,
-        finetuning_args: "FinetuningArguments",
+        beta: float,
+        model: Union["PreTrainedModel", torch.nn.Module],
         ref_model: Optional[Union["PreTrainedModel", torch.nn.Module]] = None,
+        disable_dropout: Optional[bool] = True,
         **kwargs
     ):
-        self.finetuning_args = finetuning_args
+        # if disable_dropout:
+        #     disable_dropout_in_model(model)
+        #     if ref_model is not None:
+        #         disable_dropout_in_model(ref_model)
+
         self.ref_model = ref_model
         self.use_dpo_data_collator = True # hack to avoid warning
         self.label_pad_token_id = IGNORE_INDEX
         self.padding_value = 0
-        self.beta = finetuning_args.dpo_beta
+        self.beta = beta
         self._stored_metrics = defaultdict(lambda: defaultdict(list))
 
-        Trainer.__init__(self, **kwargs)
+        Trainer.__init__(self, model=model, **kwargs)
         if not hasattr(self, "accelerator"):
             raise AttributeError("Please update `transformers`.")
 
         if ref_model is not None:
+            if self.is_deepspeed_enabled:
+                self.ref_model, = self.accelerator._prepare_deepspeed(self.ref_model)
+                self.ref_model.eval()
+            else:
                 self.ref_model = self.accelerator.prepare_model(self.ref_model, evaluation_mode=True)
 
     def concatenated_forward(
@@ -42,28 +50,13 @@ class DPOPeftTrainer(PeftModelMixin, DPOTrainer):
         batch: Optional[Dict[str, torch.Tensor]] = None
     ) -> Tuple[torch.FloatTensor, torch.FloatTensor, torch.FloatTensor, torch.FloatTensor]:
         batch_copied = BatchEncoding({k: v.detach().clone() for k, v in batch.items()}) # avoid error
-        unwrapped_model: "PreTrainedModel" = self.accelerator.unwrap_model(self.model)
-
-        if not torch.is_grad_enabled():
-            unwrapped_model.gradient_checkpointing_disable()
 
-        if model is None and isinstance(unwrapped_model, PeftModel): # peft model has no ref_model
-            with unwrapped_model.disable_adapter():
-                all_logits = self.model(
-                    input_ids=batch_copied["input_ids"],
-                    attention_mask=batch_copied["attention_mask"],
-                    return_dict=True
-                ).logits.to(torch.float32)
-        else:
         all_logits = model(
             input_ids=batch_copied["input_ids"],
             attention_mask=batch_copied["attention_mask"],
             return_dict=True
         ).logits.to(torch.float32)
 
-        if not torch.is_grad_enabled():
-            unwrapped_model.gradient_checkpointing_enable()
-
         all_logps = self._get_batch_logps(
             all_logits,
             batch["labels"],

src/llmtuner/tuner/dpo/workflow.py

@@ -10,7 +10,7 @@ from llmtuner.extras.constants import IGNORE_INDEX
 from llmtuner.extras.ploting import plot_loss
 from llmtuner.tuner.core import load_model_and_tokenizer
 from llmtuner.tuner.dpo.collator import DPODataCollatorWithPadding
-from llmtuner.tuner.dpo.trainer import DPOPeftTrainer
+from llmtuner.tuner.dpo.trainer import CustomDPOTrainer
 
 if TYPE_CHECKING:
     from transformers import TrainerCallback
@@ -37,10 +37,10 @@ def run_dpo(
     training_args = Seq2SeqTrainingArguments(**training_args_dict)
 
     # Initialize our Trainer
-    trainer = DPOPeftTrainer(
-        finetuning_args=finetuning_args,
-        ref_model=deepcopy(model) if not isinstance(model, PeftModel) else None,
+    trainer = CustomDPOTrainer(
+        beta=finetuning_args.dpo_beta,
         model=model,
+        ref_model=deepcopy(model) if not isinstance(model, PeftModel) else None,
         args=training_args,
         tokenizer=tokenizer,
         data_collator=data_collator,

src/llmtuner/tuner/ppo/trainer.py

@@ -2,29 +2,27 @@ import os
 import math
 import torch
 from tqdm import tqdm
-from typing import TYPE_CHECKING, Callable, Dict, List, Optional, Tuple
+from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, Tuple
 
-from transformers import TrainerState, TrainerControl
+from transformers import GenerationConfig, Trainer, TrainerState, TrainerControl
 
 from trl import PPOTrainer
 from trl.core import LengthSampler, PPODecorators, logprobs_from_logits
 
 from llmtuner.extras.logging import get_logger
 from llmtuner.extras.misc import AverageMeter, count_parameters, get_logits_processor
-from llmtuner.tuner.core.trainer import PeftTrainer
 from llmtuner.tuner.ppo.utils import cast_layernorm_dtype, replace_model
 
 if TYPE_CHECKING:
-    from transformers import Seq2SeqTrainingArguments
+    from transformers import Seq2SeqTrainingArguments, TrainerCallback
     from trl import AutoModelForCausalLMWithValueHead
-    from llmtuner.extras.callbacks import LogCallback
-    from llmtuner.hparams import FinetuningArguments, GeneratingArguments
+    from llmtuner.hparams import GeneratingArguments
 
 
 logger = get_logger(__name__)
 
 
-class PPOPeftTrainer(PPOTrainer, PeftTrainer):
+class CustomPPOTrainer(PPOTrainer, Trainer):
     r"""
     Inherits PPOTrainer.
     """
@@ -32,17 +30,18 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
     def __init__(
         self,
         training_args: "Seq2SeqTrainingArguments",
-        finetuning_args: "FinetuningArguments",
         generating_args: "GeneratingArguments",
-        callbacks: List["LogCallback"],
+        callbacks: List["TrainerCallback"],
         compute_dtype: torch.dtype,
         **kwargs
     ):
         PPOTrainer.__init__(self, **kwargs)
+        if getattr(self.accelerator.state, "deepspeed_plugin", None) is not None:
+            raise ValueError("PPOTrainer is incompatible with DeepSpeed.")
+
         self.args = training_args
-        self.finetuning_args = finetuning_args
         self.generating_args = generating_args
-        self.log_callback = callbacks[0]
+        self.log_callback, self.save_callback = callbacks[0], callbacks[1]
         self.compute_dtype = compute_dtype
         self.state = TrainerState()
         self.control = TrainerControl()
@@ -75,10 +74,11 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
             logger.info(f"  Number of trainable parameters = {count_parameters(self.model)[0]}")
 
         # Keyword arguments for `model.generate`
-        gen_kwargs = self.generating_args.to_dict()
-        gen_kwargs["eos_token_id"] = list(set([self.tokenizer.eos_token_id] + self.tokenizer.additional_special_tokens_ids))
-        gen_kwargs["pad_token_id"] = self.tokenizer.pad_token_id
-        gen_kwargs["logits_processor"] = get_logits_processor()
+        generating_args = self.generating_args.to_dict()
+        generating_args.update(dict(
+            eos_token_id=[self.tokenizer.eos_token_id] + self.tokenizer.additional_special_tokens_ids,
+            pad_token_id=self.tokenizer.pad_token_id
+        ))
 
         length_sampler = LengthSampler(max_target_length // 2, max_target_length)
         unwrapped_model: "AutoModelForCausalLMWithValueHead" = self.accelerator.unwrap_model(self.model)
@@ -96,17 +96,23 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
             # Cast to inference mode
             unwrapped_model.gradient_checkpointing_disable()
             unwrapped_model.config.use_cache = True
+            unwrapped_model, layer_norm_params = cast_layernorm_dtype(unwrapped_model, self.compute_dtype)
+            self.model.eval()
 
             # Get inputs
-            queries, responses = self.get_inputs(batch, length_sampler, **gen_kwargs)
+            queries, responses = self.get_inputs(batch, length_sampler, generating_args)
+            self.tokenizer.padding_side = "right" # change padding side
             rewards = self.get_rewards(queries, responses, unwrapped_model)
 
             # Cast to training mode
             unwrapped_model.gradient_checkpointing_enable()
             unwrapped_model.config.use_cache = False
+            unwrapped_model, _ = cast_layernorm_dtype(unwrapped_model, self.compute_dtype, layer_norm_params)
+            self.model.train()
 
             # Run PPO step
             stats = self.step(queries, responses, rewards)
+            self.tokenizer.padding_side = "left" # restore padding side
             loss_meter.update(stats["ppo/loss/total"], n=len(rewards))
             reward_meter.update(torch.stack(rewards).mean().item(), n=len(rewards))
 
@@ -137,36 +143,44 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
                 dataiter = iter(self.dataloader)
                 steps_trained = 0
 
-        self.log_callback.on_train_end(self.args, self.state, self.control)
+        self.log_callback.on_train_end(
+            self.args, self.state, self.control, model=self.accelerator.unwrap_model(self.model)
+        )
 
     @torch.no_grad()
     def get_inputs(
         self,
         batch: Dict[str, torch.Tensor],
-        length_sampler: Optional[Callable] = None,
-        **generation_kwargs
+        length_sampler: Callable,
+        generating_args: Dict[str, Any]
     ) -> Tuple[List[torch.Tensor], List[torch.Tensor]]:
         r"""
         Generates model's responses given queries.
         """
-        if length_sampler is not None:
-            generation_kwargs["max_new_tokens"] = length_sampler()
+        generating_args["max_new_tokens"] = length_sampler()
+        gen_kwargs = dict(
+            generation_config=GenerationConfig(**generating_args),
+            logits_processor=get_logits_processor(),
+            **batch
+        )
 
-        self.model, layer_norm_params = cast_layernorm_dtype(self.model, self.compute_dtype)
+        input_ids = batch["input_ids"]
         unwrapped_model: "AutoModelForCausalLMWithValueHead" = self.accelerator.unwrap_model(self.model)
-        response: torch.Tensor = unwrapped_model.generate(**batch, **generation_kwargs)
-        self.model, _ = cast_layernorm_dtype(self.model, self.compute_dtype, layer_norm_params)
-
-        # Temporary hack to ensure the generation config is not initialized for each iteration of the evaluation loop
-        # Inspired by: https://github.com/huggingface/transformers/blob/v4.28.1/src/transformers/trainer_seq2seq.py#L273
-        if unwrapped_model.pretrained_model.generation_config._from_model_config:
-            unwrapped_model.pretrained_model.generation_config._from_model_config = False
+        response: torch.Tensor = unwrapped_model.generate(**gen_kwargs)
+        query, response = input_ids.detach().cpu(), response[:, input_ids.size(-1):].detach().cpu()
 
         queries, responses = [], []
-        query, response = batch["input_ids"].detach().cpu(), response[:, batch["input_ids"].size(-1):].detach().cpu()
         for i in range(len(query)):
             query_length = (query[i] != self.tokenizer.pad_token_id).nonzero()[0]
-            response_length = (response[i] != self.tokenizer.pad_token_id).nonzero()[-1] + 1
+            response_index = (response[i] != self.tokenizer.pad_token_id).nonzero()
+
+            if len(response_index) == 0:
+                response_length = 1 # allow empty response
+            elif self.tokenizer.pad_token_id == self.tokenizer.eos_token_id:
+                response_length = response_index[-1] + 2 # save the EOS token
+            else:
+                response_length = response_index[-1] + 1
+
             queries.append(query[i, query_length:]) # remove padding from left
             responses.append(response[i, :response_length]) # remove padding from right
 
@@ -191,7 +205,11 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
         if values.size(0) != batch["input_ids"].size(0): # adapt to chatglm2
             values = torch.transpose(values, 0, 1)
 
-        rewards = [reward for reward in values[:, -1].float().detach().cpu()] # use fp32 type
+        rewards = []
+        for i in range(values.size(0)):
+            end_index = batch["attention_mask"][i].nonzero()[-1] # use the score on the EOS token
+            rewards.append(values[i, end_index].float().detach().cpu()) # use fp32 type
+
         replace_model(unwrapped_model, target="default")
         return rewards
 
@@ -202,7 +220,8 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
         queries: torch.Tensor,
         responses: torch.Tensor,
         model_inputs: dict,
-        return_logits: Optional[bool] = False
+        return_logits: Optional[bool] = False,
+        response_masks: Optional[torch.Tensor] = None
     ):
         r"""
         Calculates model outputs in multiple batches.
@@ -220,6 +239,8 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
             input_kwargs = {key: value[i * fbs : (i + 1) * fbs] for key, value in model_inputs.items()}
             query_batch = queries[i * fbs : (i + 1) * fbs]
             response_batch = responses[i * fbs : (i + 1) * fbs]
+            if response_masks is not None:
+                response_masks_batch = response_masks[i * fbs : (i + 1) * fbs]
             input_ids = input_kwargs["input_ids"]
             attention_mask = input_kwargs["attention_mask"]
 
@@ -239,8 +260,15 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
                     start += attention_mask[j, :].nonzero()[0]
                 end = start + len(response_batch[j])
 
+                if response_masks is not None:
+                    response_masks_batch = torch.cat(
+                        (torch.zeros_like(query_batch[j]), response_masks_batch[j])
+                    )[1:]
+
                 masks[j, :start] = 0
                 masks[j, end:] = 0
+                if response_masks is not None:
+                    masks[j, start:end] = masks[j, start:end] * response_masks_batch[j][start:end]
 
             if return_logits:
                 all_logits.append(logits)
@@ -266,3 +294,6 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
         """
         if self.args.should_save:
             self._save(output_dir)
+            self.save_callback.on_save(
+                self.args, self.state, self.control, model=self.accelerator.unwrap_model(self.model)
+            )

src/llmtuner/tuner/ppo/workflow.py

@@ -4,14 +4,14 @@ import math
 from trl import PPOConfig
 from torch.optim import AdamW
 from typing import TYPE_CHECKING, Optional, List
-from transformers import DataCollatorForSeq2Seq
+from transformers import DataCollatorWithPadding
 from transformers.optimization import get_scheduler
-from transformers.utils.versions import require_version
 
 from llmtuner.dsets import get_dataset, preprocess_dataset
+from llmtuner.extras.callbacks import SavePeftModelCallback
 from llmtuner.extras.ploting import plot_loss
 from llmtuner.tuner.core import load_model_and_tokenizer
-from llmtuner.tuner.ppo.trainer import PPOPeftTrainer
+from llmtuner.tuner.ppo.trainer import CustomPPOTrainer
 
 if TYPE_CHECKING:
     from transformers import Seq2SeqTrainingArguments, TrainerCallback
@@ -29,7 +29,9 @@ def run_ppo(
     dataset = get_dataset(model_args, data_args)
     model, tokenizer = load_model_and_tokenizer(model_args, finetuning_args, training_args.do_train, stage="ppo")
     dataset = preprocess_dataset(dataset, tokenizer, data_args, training_args, stage="ppo")
-    data_collator = DataCollatorForSeq2Seq(tokenizer=tokenizer, label_pad_token_id=tokenizer.pad_token_id)
+
+    tokenizer.padding_side = "left" # use left-padding in generation while using right-padding in training
+    data_collator = DataCollatorWithPadding(tokenizer=tokenizer)
 
     ppo_config = PPOConfig(
         model_name=model_args.model_name_or_path,
@@ -44,7 +46,6 @@ def run_ppo(
     )
 
     if finetuning_args.ppo_score_norm:
-        require_version("trl>=0.5.1.dev0", "To fix: pip install git+https://github.com/huggingface/trl.git")
         ppo_config.use_score_scaling = True
         ppo_config.use_score_norm = True
 
@@ -61,11 +62,10 @@ def run_ppo(
     )
 
     # Initialize our Trainer
-    ppo_trainer = PPOPeftTrainer(
+    ppo_trainer = CustomPPOTrainer(
         training_args=training_args,
-        finetuning_args=finetuning_args,
         generating_args=generating_args,
-        callbacks=callbacks,
+        callbacks=callbacks + [SavePeftModelCallback()],
         compute_dtype=model_args.compute_dtype,
         config=ppo_config,
         model=model,

src/llmtuner/tuner/pt/workflow.py

@@ -2,12 +2,11 @@
 
 import math
 from typing import TYPE_CHECKING, Optional, List
-from transformers import DataCollatorForLanguageModeling
+from transformers import DataCollatorForLanguageModeling, Trainer
 
 from llmtuner.dsets import get_dataset, preprocess_dataset, split_dataset
 from llmtuner.extras.ploting import plot_loss
 from llmtuner.tuner.core import load_model_and_tokenizer
-from llmtuner.tuner.core.trainer import PeftTrainer
 
 if TYPE_CHECKING:
     from transformers import Seq2SeqTrainingArguments, TrainerCallback
@@ -27,8 +26,7 @@ def run_pt(
     data_collator = DataCollatorForLanguageModeling(tokenizer=tokenizer, mlm=False)
 
     # Initialize our Trainer
-    trainer = PeftTrainer(
-        finetuning_args=finetuning_args,
+    trainer = Trainer(
         model=model,
         args=training_args,
         tokenizer=tokenizer,