Merge branch 'main' into main

Former-commit-id: 870d2c7bf74d0da5a927bef4b8b01d15cc66a3e9

src/llmtuner/tuner/sft/workflow.py

@@ -16,7 +16,7 @@ from llmtuner.extras.logging import reset_logging, get_logger
 
 if TYPE_CHECKING:
     from transformers import Seq2SeqTrainingArguments, TrainerCallback
-    from llmtuner.hparams import ModelArguments, DataArguments, FinetuningArguments
+    from llmtuner.hparams import ModelArguments, DataArguments, FinetuningArguments, GeneratingArguments
 
 logger = get_logger(__name__)
 
@@ -25,6 +25,7 @@ def run_sft(
     data_args: "DataArguments",
     training_args: "Seq2SeqTrainingArguments",
     finetuning_args: "FinetuningArguments",
+    generating_args: "GeneratingArguments",
     callbacks: Optional[List["TrainerCallback"]] = None
 ):
     dataset = get_dataset(model_args, data_args)
@@ -50,31 +51,15 @@ def run_sft(
         data_collator=data_collator,
         callbacks=callbacks,
         compute_metrics=ComputeMetrics(tokenizer) if training_args.predict_with_generate else None,
-        **split_dataset(dataset, data_args.dev_ratio, training_args.do_train)
+        **split_dataset(dataset, data_args, training_args)
     )
 
     # Keyword arguments for `model.generate`
-    gen_kwargs = {
-        "do_sample": True,
-        "top_p": 0.7,
-        "max_new_tokens": data_args.max_target_length + 1,
-        "temperature": 0.95,
-        "logits_processor": get_logits_processor()
-    }
-    # Detecting last checkpoint.
-    last_checkpoint = None
-    if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
-        last_checkpoint = get_last_checkpoint(training_args.output_dir)
-        if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
-            raise ValueError(
-                f"Output directory ({training_args.output_dir}) already exists and is not empty. "
-                "Use --overwrite_output_dir to overcome."
-            )
-        elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
-            logger.info(
-                f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
-                "the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
-            )
+    gen_kwargs = generating_args.to_dict()
+    gen_kwargs["eos_token_id"] = list(set([tokenizer.eos_token_id] + tokenizer.additional_special_tokens_ids))
+    gen_kwargs["pad_token_id"] = tokenizer.pad_token_id
+    gen_kwargs["logits_processor"] = get_logits_processor()
+
     # Training
     if training_args.do_train:
         checkpoint = None