chore: import upstream snapshot with attribution

examples/patching/anyscale.py

@@ -0,0 +1,37 @@
+import os
+import instructor
+
+from openai import OpenAI
+from pydantic import BaseModel
+
+
+# By default, the patch function will patch the ChatCompletion.create and ChatCompletion.acreate methods. to support response_model parameter
+client = instructor.from_openai(
+    OpenAI(
+        base_url="https://api.endpoints.anyscale.com/v1",
+        api_key=os.environ["ANYSCALE_API_KEY"],
+    ),
+    mode=instructor.Mode.JSON_SCHEMA,
+)
+
+
+# Now, we can use the response_model parameter using only a base model
+# rather than having to use the ResponseSchema class
+class UserExtract(BaseModel):
+    name: str
+    age: int
+
+
+user: UserExtract = client.chat.completions.create(
+    model="mistralai/Mixtral-8x7B-Instruct-v0.1",
+    response_model=UserExtract,
+    messages=[
+        {"role": "user", "content": "Extract jason is 25 years old"},
+    ],
+)  # type: ignore
+
+print(user)
+{
+    "name": "Jason",
+    "age": 25,
+}

examples/patching/oai.py

@@ -0,0 +1,33 @@
+import instructor
+
+from openai import OpenAI
+from pydantic import BaseModel
+
+
+# By default, the patch function will patch the ChatCompletion.create and ChatCompletion.acreate methods. to support response_model parameter
+client = instructor.from_openai(
+    OpenAI(),
+    mode=instructor.Mode.TOOLS,
+)
+
+
+# Now, we can use the response_model parameter using only a base model
+# rather than having to use the ResponseSchema class
+class UserExtract(BaseModel):
+    name: str
+    age: int
+
+
+user: UserExtract = client.chat.completions.create(
+    model="gpt-3.5-turbo",
+    response_model=UserExtract,
+    messages=[
+        {"role": "user", "content": "Extract jason is 25 years old"},
+    ],
+)  # type: ignore
+
+print(user)
+{
+    "name": "Jason",
+    "age": 25,
+}

examples/patching/pcalls.py

@@ -0,0 +1,87 @@
+from typing import Literal, Union
+from collections.abc import Iterable
+from pydantic import BaseModel
+from instructor import ResponseSchema
+
+import time
+import openai
+import instructor
+
+
+client = openai.OpenAI()
+
+
+class Weather(ResponseSchema):
+    location: str
+    units: Literal["imperial", "metric"]
+
+
+class GoogleSearch(ResponseSchema):
+    query: str
+
+
+if __name__ == "__main__":
+
+    class Query(BaseModel):
+        query: list[Union[Weather, GoogleSearch]]
+
+    client = instructor.from_openai(client, mode=instructor.Mode.PARALLEL_TOOLS)
+
+    start = time.perf_counter()
+    resp = client.chat.completions.create(
+        model="gpt-4-turbo-preview",
+        messages=[
+            {"role": "system", "content": "You must always use tools"},
+            {
+                "role": "user",
+                "content": "What is the weather in toronto and dallas and who won the super bowl?",
+            },
+        ],
+        response_model=Iterable[Union[Weather, GoogleSearch]],
+    )
+    print(f"# Time: {time.perf_counter() - start:.2f}")
+
+    print("# Instructor: Question with Toronto and Super Bowl")
+    print([model for model in resp])
+
+    start = time.perf_counter()
+    resp = client.chat.completions.create(
+        model="gpt-4-turbo-preview",
+        messages=[
+            {
+                "role": "user",
+                "content": "What is the weather in toronto and dallas?",
+            },
+        ],
+        tools=[
+            {"type": "function", "function": Weather.openai_schema},
+            {"type": "function", "function": GoogleSearch.openai_schema},
+        ],
+        tool_choice="auto",
+    )
+    print(f"# Time: {time.perf_counter() - start:.2f}")
+
+    print("# Question with Toronto and Dallas")
+    for tool_call in resp.choices[0].message.tool_calls:
+        print(tool_call.model_dump_json(indent=2))
+
+    start = time.perf_counter()
+    resp = client.chat.completions.create(
+        model="gpt-4-turbo-preview",
+        messages=[
+            {
+                "role": "user",
+                "content": "What is the weather in toronto? and who won the super bowl?",
+            },
+        ],
+        tools=[
+            {"type": "function", "function": Weather.openai_schema},
+            {"type": "function", "function": GoogleSearch.openai_schema},
+        ],
+        tool_choice="auto",
+    )
+    print(f"# Time: {time.perf_counter() - start:.2f}")
+
+    print("# Question with Toronto and Super Bowl")
+    for tool_call in resp.choices[0].message.tool_calls:
+        print(tool_call.model_dump_json(indent=2))

examples/patching/together.py

@@ -0,0 +1,35 @@
+import os
+import openai
+from pydantic import BaseModel
+import instructor
+
+client = openai.OpenAI(
+    base_url="https://api.together.xyz/v1",
+    api_key=os.environ["TOGETHER_API_KEY"],
+)
+
+
+# By default, the patch function will patch the ChatCompletion.create and ChatCompletion.acreate methods. to support response_model parameter
+client = instructor.from_openai(client, mode=instructor.Mode.TOOLS)
+
+
+# Now, we can use the response_model parameter using only a base model
+# rather than having to use the ResponseSchema class
+class UserExtract(BaseModel):
+    name: str
+    age: int
+
+
+user: UserExtract = client.chat.completions.create(
+    model="mistralai/Mixtral-8x7B-Instruct-v0.1",
+    response_model=UserExtract,
+    messages=[
+        {"role": "user", "content": "Extract jason is 25 years old"},
+    ],
+)  # type: ignore
+
+print(user.model_dump_json(indent=2))
+{
+    "name": "Jason",
+    "age": 25,
+}