Merge pull request #91 from eosphoros-ai/refactor

Refactor
eosphoros-ai · Oct 10, 2023 · fbb8513 · fbb8513
2 parents 0d5ae50 + a2128db
commit fbb8513
Show file tree

Hide file tree

Showing 13 changed files with 45 additions and 852 deletions.
diff --git a/dbgpt_hub/configs/__init__.py b/dbgpt_hub/configs/__init__.py
@@ -1,7 +1,6 @@
 from .model_args import (
     ModelArguments,
     TrainingArguments,
-    ModelInferenceArguments,
 )
 from .data_args import(
     DataArguments,
@@ -13,5 +12,4 @@
     "Llama2Template",
     "ModelArguments",
     "TrainingArguments",
-    "ModelInferenceArguments",
 ]
diff --git a/dbgpt_hub/configs/data_args.py b/dbgpt_hub/configs/data_args.py
@@ -1,16 +1,12 @@
 import os
 import json
-from typing import List, Literal, Optional
-from dataclasses import dataclass, field
 import tiktoken
-from dataclasses import dataclass
-from typing import TYPE_CHECKING, Dict, List, Optional, Tuple, Union
-
-# from dbgpt_hub.llm_base.loggings import get_logger
-
+from dataclasses import dataclass, field
+from typing import TYPE_CHECKING, Dict, List, Literal, Optional, Tuple, Union
 if TYPE_CHECKING:
     from transformers import PreTrainedTokenizer
 
+
 DEFAULT_PROMPT_DICT = {
     "prompt_input": ("{instruction}\n\n{input}\n\n"),
     "prompt_no_input": ("{instruction}\n\n"),

diff --git a/dbgpt_hub/configs/model_args.py b/dbgpt_hub/configs/model_args.py
@@ -1,14 +1,10 @@
 import json
-import os
-import yaml
 import torch
 from dataclasses import dataclass, field, asdict
-from typing import Optional, Any, Dict, List,Literal
+from typing import Optional, Any, Dict,Literal
 from transformers import Seq2SeqTrainingArguments
 from dbgpt_hub.configs.config import (
     MODEL_PATH,
-    DEFAULT_FT_MODEL_NAME,
-    DATA_PATH,
     ADAPTER_PATH,
 )
 
@@ -118,28 +114,6 @@ def __post_init__(self):
             HfFolder.save_token(self.hf_auth_token)
 
 
-@dataclass
-class ModelInferenceArguments:
-    cache_dir: Optional[str] = field(default=None)
-    model_name_or_path: Optional[str] = field(
-        default=MODEL_PATH, metadata={"help": "Path to pre-trained model"}
-    )
-    model_max_length: int = field(
-        default=1024,
-        metadata={
-            "help": "Maximum sequence length. Sequences will be right padded (and possibly truncated)."
-        },
-    )
-    prompt_template: str = field(
-        default="default",
-        metadata={
-            "help": "Prompt template name. Such as vanilla, alpaca, llama2, vicuna..., etc."
-        },
-    )
-    source_prefix: Optional[str] = field(
-        default=None, metadata={"help": "Prefix to prepend to every source text."}
-    )
-
 
 
 @dataclass