Merge pull request #224 from CambioML/dev

Add summary prompt and bump version to 0.0.30

Author: Cambio ML

example/toc.ipynb

@@ -13,7 +13,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!pip3 install -q pandas tabulate uniflow==0.0.29\n"
+    "!pip3 install -q pandas tabulate uniflow==0.0.30\n"
    ]
   },
   {

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "uniflow"
-version = "0.0.29"
+version = "0.0.30"
 description = "Unified interface for pre-training data augmentation and post-training evaluation of Large Language Models (LLMs)."
 authors = ["CambioML <[email protected]>"]
 maintainers = ["Rachel Hu <[email protected]>"]

uniflow/__init__.py

@@ -4,4 +4,4 @@
 
 __all__ = ["PromptTemplate", "Context"]
 
-__version__ = "0.0.29"
+__version__ = "0.0.30"

uniflow/flow/config.py

@@ -131,6 +131,32 @@ class TransformConfig:
     auto_split_long_text: bool = field(default=False)
 
 
+@dataclass
+class TransformSummaryConfig(TransformConfig):
+    """Transform Summary Config Class."""
+
+    flow_name: str = "TransformOpenAIFlow"
+    model_config: ModelConfig = field(default_factory=OpenAIModelConfig)
+    num_thread: int = 1
+    prompt_template: PromptTemplate = field(
+        default_factory=lambda: PromptTemplate(
+            instruction="""
+            Given a lengthy and detailed text, the task is to generate a concise and accurate summary that captures the main points, themes, and conclusions without losing the essence of the original content.
+            The summary should be coherent, readable, and no longer than a few sentences, making it easily understandable at a glance.
+            It must distill complex ideas and narratives into their core insights, highlighting critical information while omitting extraneous details.
+            The goal is to provide a clear and succinct overview that enables readers to grasp the significance and overarching narrative of the full text quickly.
+            """,
+            few_shot_prompt=[
+                Context(
+                    context="...",
+                    summary="...",
+                ),
+            ],
+        )
+    )
+    auto_split_long_text: bool = field(default=False)
+
+
 @dataclass
 class TransformGmailSpamConfig(TransformConfig):
     """Transform Google Config Class."""