Merge pull request #3 from pedMatias/feature/multi-model-support

[feature & bug fix] Update openai library and add support to gpt4

Author: pplonski

README.md

@@ -89,6 +89,19 @@ plot = PlotAI(df)
 plot.make("make a scatter plot")
 ```
 
+By default the library will use '*gpt-3.5-turbo*'. You can use different OpenAI models:
+
+```python
+# import PlotAI
+from plotai import PlotAI
+
+# create PlotAI object, pass pandas DataFrame as an argument
+plot = PlotAI(df, model_version="gpt-4")
+
+# make a plot, just tell what you want
+plot.make("make a scatter plot")
+```
+
 ## More examples
 
 #### Analyze the GPD dataset

plotai/llm/openai.py

@@ -2,10 +2,11 @@
 import openai
 
 from dotenv import load_dotenv
+
 load_dotenv()
 
 
-class ChatGPT():
+class ChatGPT:
 
     temperature = 0
     max_tokens = 1000
@@ -14,14 +15,15 @@ class ChatGPT():
     presence_penalty = 0.6
     model = "gpt-3.5-turbo"
 
-
-    def __init__(self):
+    def __init__(self, model: str):
         api_key = os.environ.get("OPENAI_API_KEY")
         if api_key is None:
-            raise Exception("Please set OPENAI_API_KEY environment variable."
-            "You can obtain API key from https://platform.openai.com/account/api-keys")
+            raise Exception(
+                "Please set OPENAI_API_KEY environment variable."
+                "You can obtain API key from https://platform.openai.com/account/api-keys"
+            )
         openai.api_key = api_key
-        
+        self.model = model
 
     @property
     def _default_params(self):
@@ -35,6 +37,7 @@ def _default_params(self):
         }
 
     def chat(self, prompt):
+        client = openai.OpenAI()
 
         params = {
             **self._default_params,
@@ -45,5 +48,5 @@ def chat(self, prompt):
                 }
             ],
         }
-        response = openai.ChatCompletion.create(**params)
-        return response["choices"][0]["message"]["content"]
+        response = client.chat.completions.create(**params)
+        return response.choices[0].message.content

plotai/plotai.py

@@ -5,10 +5,14 @@
 from plotai.code.executor import Executor
 from plotai.code.logger import Logger
 
+
 class PlotAI:
 
-    def __init__(self, *args, **kwargs):
+    def __init__(self, model_version: str = "gpt-3.5-turbo", *args, **kwargs):
 
+        # OpenAI Model Version
+        self.model_version = model_version
+        # DataFrame to plot
         self.df, self.x, self.y, self.z = None, None, None, None
         if len(args) > 1:
             for i in range(len(args)):
@@ -34,13 +38,11 @@ def __init__(self, *args, **kwargs):
                     setattr(self, k, kwargs[k])
 
     def make(self, prompt):
-
-        df, x, y, z = self.df, self.x, self.y, self.z
-        p = Prompt(prompt, self.df, self.x, self.y, self.z)    
+        p = Prompt(prompt, self.df, self.x, self.y, self.z)
 
         Logger().log({"title": "Prompt", "details": p.value})
 
-        response = ChatGPT().chat(p.value)
+        response = ChatGPT(model=self.model_version).chat(p.value)
 
         Logger().log({"title": "Response", "details": response})
 
@@ -49,8 +51,7 @@ def make(self, prompt):
         if error is not None:
             Logger().log({"title": "Error in code execution", "details": error})
 
-
-            # p_again = Prompt(prompt, self.df, self.x, self.y, self.z, previous_code=response, previous_error=error)  
+            # p_again = Prompt(prompt, self.df, self.x, self.y, self.z, previous_code=response, previous_error=error)
 
             # Logger().log({"title": "Prompt with fix", "details": p_again.value})
 

requirements.txt

@@ -1,5 +1,5 @@
-matplotlib
-pandas
-numpy
-openai
-python-dotenv
+matplotlib~=3.8.3
+pandas~=2.2.0
+numpy~=1.26.4
+openai~=1.12.0
+python-dotenv~=1.0.1

setup.py

@@ -10,7 +10,7 @@
 
 setup(
     name="plotai",
-    version="0.0.2",
+    version="0.0.3",
     description="Create plots in Python with AI",
     long_description=long_description,
     long_description_content_type="text/markdown",
@@ -21,7 +21,7 @@
     packages=find_packages(exclude=["*.tests", "*.tests.*", "tests.*", "tests"]),
     install_requires=open("requirements.txt").readlines(),
     include_package_data=True,
-    python_requires='>=3.7.1',
+    python_requires=">=3.7.1",
     classifiers=[
         "Programming Language :: Python",
         "Programming Language :: Python :: 3.7",
@@ -37,6 +37,6 @@
         "matplotlib",
         "llm",
         "openai",
-        "mljar"
+        "mljar",
     ],
 )

tests/test_plotai.py

@@ -17,3 +17,9 @@ def test_pass_data(self):
         df2 = pd.DataFrame({"x":np.random.rand(rows), "y": np.random.rand(rows)})
         plot = PlotAI(df=df2)
         #plot.make("Plot a scatter plot")
+    
+    def test_gpt4(self):
+        rows = 100
+        df2 = pd.DataFrame({"x":np.random.rand(rows), "y": np.random.rand(rows)})
+        plot = PlotAI(df=df2, model_version="gpt4")
+        #plot.make("Plot a scatter plot")