fix #2282 and update tool prompt

data/alpaca_data_zh_51k.json

@@ -11192,7 +11192,7 @@
   {
     "instruction": "从给定数据中找出两个事件的概率",
     "input": "事件A：会下雨\n事件B：会下雪\n数据：\n- 明日天气预报：下雨的概率为80％，下雪的概率为20％",
-    "输出": "事件A和事件B同时发生的概率为0.16或16％。这是通过将事件A的概率（0.8）和事件B的概率（0.2）相乘来计算的。"
+    "output": "事件A和事件B同时发生的概率为0.16或16％。这是通过将事件A的概率（0.8）和事件B的概率（0.2）相乘来计算的。"
   },
   {
     "instruction": "对于给定情境给予建议",

src/llmtuner/data/aligner.py

@@ -15,7 +15,7 @@ def convert_alpaca(examples: Dict[str, List[Any]], dataset_attr: "DatasetAttr")
     outputs = {"prompt": [], "response": [], "system": [], "tools": []}
     for i in range(len(examples[dataset_attr.prompt])):
         prompt = []
-        if dataset_attr.history:
+        if dataset_attr.history and isinstance(examples[dataset_attr.history][i], list):
             for old_prompt, old_response in examples[dataset_attr.history][i]:
                 prompt.append({"role": Role.USER, "content": old_prompt})
                 prompt.append({"role": Role.ASSISTANT, "content": old_response})
@@ -25,12 +25,9 @@ def convert_alpaca(examples: Dict[str, List[Any]], dataset_attr: "DatasetAttr")
             instruction += "\n" + examples[dataset_attr.query][i]
         prompt.append({"role": Role.USER, "content": instruction})
 
-        if dataset_attr.response:
+        if dataset_attr.response and isinstance(examples[dataset_attr.response][i], list):
-            if isinstance(examples[dataset_attr.response][i], list):
+            response = [{"role": Role.ASSISTANT, "content": content} for content in examples[dataset_attr.response][i]]
-                response = [
+        elif dataset_attr.response and isinstance(examples[dataset_attr.response][i], str):
-                    {"role": Role.ASSISTANT, "content": content} for content in examples[dataset_attr.response][i]
-                ]
-            else:
             response = [{"role": Role.ASSISTANT, "content": examples[dataset_attr.response][i]}]
         else:
             response = []

src/llmtuner/data/formatter.py

@@ -15,10 +15,10 @@ JSON_FORMAT_PROMPT = (
 
 TOOL_SYSTEM_PROMPT = (
     "You have access to the following tools:\n{tool_text}"
-    "Use the following format to answer the question:\n"
+    "Use the following format if using a tool:\n"
     "```\n"
-    "Action: the action to take, should be one of [{tool_names}] if using a tool.\n"
+    "Action: tool name (one of [{tool_names}]).\n"
-    "Action Input: the input to the action{format_prompt}.\n"
+    "Action Input: the input to the tool{format_prompt}.\n"
     "```\n"
 )
 
@@ -95,12 +95,15 @@ class StringFormatter(Formatter):
         for slot in self.slots:
             if isinstance(slot, str):
                 for name, value in kwargs.items():
+                    if not isinstance(value, str):
+                        raise RuntimeError("Expected a string, got {}".format(value))
+
                     slot = slot.replace("{{" + name + "}}", value, 1)
                 elements.append(slot)
             elif isinstance(slot, (dict, set)):
                 elements.append(slot)
             else:
-                raise ValueError("Input must be string, set[str] or dict[str, str], got {}".format(type(slot)))
+                raise RuntimeError("Input must be string, set[str] or dict[str, str], got {}".format(type(slot)))
 
         return elements
 
@@ -124,7 +127,7 @@ class FunctionFormatter(Formatter):
             elif isinstance(slot, (dict, set)):
                 elements.append(slot)
             else:
-                raise ValueError("Input must be string, set[str] or dict[str, str], got {}".format(type(slot)))
+                raise RuntimeError("Input must be string, set[str] or dict[str, str], got {}".format(type(slot)))
 
         return elements
 

src/llmtuner/data/preprocess.py

@@ -4,6 +4,7 @@ from typing import TYPE_CHECKING, Any, Callable, Dict, List, Literal, Tuple
 
 from ..extras.constants import IGNORE_INDEX
 from ..extras.logging import get_logger
+from .utils import Role
 
 
 if TYPE_CHECKING:
@@ -51,7 +52,7 @@ def preprocess_supervised_dataset(
     model_inputs = {"input_ids": [], "attention_mask": [], "labels": []}
 
     for i in range(len(examples["prompt"])):
-        if len(examples["prompt"][i]) == 0 or len(examples["response"][i]) != 1:
+        if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) != 1:
             continue
 
         messages = examples["prompt"][i] + examples["response"][i]
@@ -93,7 +94,7 @@ def preprocess_packed_supervised_dataset(
     model_inputs = {"input_ids": [], "attention_mask": [], "labels": []}
     input_ids, labels = [], []
     for i in range(len(examples["prompt"])):
-        if len(examples["prompt"][i]) == 0 or len(examples["response"][i]) != 1:
+        if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) != 1:
             continue
 
         messages = examples["prompt"][i] + examples["response"][i]
@@ -137,10 +138,14 @@ def preprocess_unsupervised_dataset(
     model_inputs = {"input_ids": [], "attention_mask": [], "labels": []}
 
     for i in range(len(examples["prompt"])):
-        if len(examples["prompt"][i]) == 0 or len(examples["response"][i]) != 1:
+        if len(examples["prompt"][i]) % 2 != 1:
             continue
 
+        if len(examples["response"][i]) == 1:
             messages = examples["prompt"][i] + examples["response"][i]
+        else:
+            messages = examples["prompt"][i] + [{"role": Role.ASSISTANT, "content": ""}]
+
         input_ids, labels = template.encode_oneturn(
             tokenizer, messages, examples["system"][i], examples["tools"][i], data_args.cutoff_len
         )
@@ -164,7 +169,7 @@ def preprocess_pairwise_dataset(
     # build input pairs with format `<bos> X`, `Y1 <eos>` and `Y2 <eos>`
     model_inputs = {"prompt_ids": [], "chosen_ids": [], "rejected_ids": []}
     for i in range(len(examples["prompt"])):
-        if len(examples["prompt"][i]) == 0 or len(examples["response"][i]) < 2:
+        if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) < 2:
             continue
 
         chosen_messages = examples["prompt"][i] + [examples["response"][i][0]]

src/llmtuner/hparams/parser.py

@@ -60,7 +60,7 @@ def _verify_model_args(model_args: "ModelArguments", finetuning_args: "Finetunin
         if finetuning_args.finetuning_type != "lora":
             raise ValueError("Quantization is only compatible with the LoRA method.")
 
-        if finetuning_args.create_new_adapter:
+        if model_args.adapter_name_or_path is not None and finetuning_args.create_new_adapter:
             raise ValueError("Cannot create new adapter upon a quantized model.")
 
     if model_args.adapter_name_or_path is not None and len(model_args.adapter_name_or_path) != 1: