Merge pull request #32 from parea-ai/feat-log-endpoint

Feat log endpoint

README.md

@@ -38,7 +38,7 @@ from parea.schemas.models import Completion, UseDeployedPrompt, CompletionRespon
 
 load_dotenv()
 
-p = Parea(api_key=os.getenv("API_KEY"))
+p = Parea(api_key=os.getenv("PAREA_API_KEY"))
 
 # You will find this deployment_id in the Parea dashboard
 deployment_id = '<DEPLOYMENT_ID>'
@@ -78,6 +78,72 @@ async def main_async():
   print(deployed_prompt)
 ```    
 
+
+### Logging results from LLM providers
+
+```python
+import os
+
+from dotenv import load_dotenv
+
+import openai
+from parea import Parea
+from parea.schemas.models import LogRequest
+
+load_dotenv()
+
+openai.api_key = os.getenv("OPENAI_API_KEY")
+p = Parea(api_key=os.getenv("PAREA_API_KEY"))
+
+
+# define your OpenAI call as you would normally
+x = "Golang"
+y = "Fiber"
+inputs = {"x": x, "y": y}
+messages = [
+  {"role": "user", "content": f"Write a hello world program using {x} and the {y} framework."},
+]
+model = "gpt-3.5-turbo"
+model_params = {
+  "temperature": 0.7,
+  "top_p": 1.0,
+}
+completion = openai.ChatCompletion.create(
+  model=model,
+  messages=messages,
+  **model_params
+)
+output = completion.choices[0].message['content']
+
+# the LogRequest schema
+log_request: LogRequest = LogRequest(
+  status="success",
+  name='Test Log',
+  llm_inputs={
+    "x": x,
+    "y": y,
+  },
+  llm_configuration={
+    'model': model,
+    'messages': messages,
+    'model_params': model_params,
+  },
+  output=output,
+  input_tokens=completion.usage['prompt_tokens'],
+  output_tokens=completion.usage['completion_tokens'],
+  total_tokens=completion.usage['total_tokens'],
+)
+
+
+def main():
+  p.log(data=log_request)
+
+
+async def main_async():
+  await p.alog(data=log_request)
+```
+
+
 ### Open source community features
 
 Ready-to-use [Pull Requests templates](https://github.com/parea-ai/parea-sdk/blob/master/.github/PULL_REQUEST_TEMPLATE.md)

parea/client.py

@@ -3,11 +3,12 @@
 from attrs import asdict, define, field
 
 from parea.api_client import HTTPClient
-from parea.schemas.models import Completion, CompletionResponse, FeedbackRequest, UseDeployedPrompt, UseDeployedPromptResponse
+from parea.schemas.models import Completion, CompletionResponse, FeedbackRequest, LogRequest, UseDeployedPrompt, UseDeployedPromptResponse
 
 COMPLETION_ENDPOINT = "/completion"
 DEPLOYED_PROMPT_ENDPOINT = "/deployed-prompt"
 RECORD_FEEDBACK_ENDPOINT = "/feedback"
+LOG_ENDPOINT = "/log"
 
 
 @define
@@ -64,6 +65,20 @@ async def arecord_feedback(self, data: FeedbackRequest) -> None:
             data=asdict(data),
         )
 
+    def log(self, data: LogRequest) -> None:
+        self._client.request(
+            "POST",
+            LOG_ENDPOINT,
+            data=asdict(data),
+        )
+
+    async def alog(self, data: LogRequest) -> None:
+        await self._client.request_async(
+            "POST",
+            LOG_ENDPOINT,
+            data=asdict(data),
+        )
+
 
 def gen_trace_id() -> str:
     """Generate a unique trace id for each chain of requests"""

parea/schemas/models.py

@@ -21,7 +21,6 @@ class Message:
 
 @define
 class ModelParams:
-    model: str
     temp: float = 0.5
     top_p: float = 1.0
     frequency_penalty: float = 0.0
@@ -108,3 +107,36 @@ class FeedbackRequest:
     inference_id: Optional[str] = None
     name: Optional[str] = None
     target: Optional[str] = None
+
+
+@define
+class LogRequest:
+    inference_id: Optional[str] = None
+    trace_id: Optional[str] = None
+    trace_name: Optional[str] = None
+    end_user_identifier: Optional[str] = None
+    error: Optional[str] = None
+    status: Optional[str] = None
+
+    start_timestamp: str = ""
+    end_timestamp: str = ""
+    duration: float = 0.0
+
+    deployment_id: Optional[str] = None
+    name: Optional[str] = None
+    evaluation_metrics_ids: Optional[list[int]] = None
+    metadata: Optional[dict] = None
+    cache_hit: bool = False
+    target: Optional[str] = None
+    tags: Optional[list[str]] = None
+
+    llm_inputs: Optional[dict[str, Any]] = None
+    llm_configuration: LLMInputs = LLMInputs()
+
+    output: Optional[str] = ""
+    latency: Optional[float] = 0.0
+    input_tokens: Optional[int] = 0
+    output_tokens: Optional[int] = 0
+    total_tokens: Optional[int] = 0
+    cost: Optional[float] = None
+    feedback_score: Optional[float] = None