Merge branch 'main' into dr_log10feedback_documentation

examples/logging/litellm_async_stream_completion.py

@@ -0,0 +1,23 @@
+import asyncio
+
+import litellm
+
+from log10.litellm import Log10LitellmLogger
+
+
+log10_handler = Log10LitellmLogger(tags=["litellm_acompletion"])
+litellm.callbacks = [log10_handler]
+
+model_name = "claude-3-haiku-20240307"
+
+
+async def completion():
+    response = await litellm.acompletion(
+        model=model_name, messages=[{"role": "user", "content": "count to 10"}], stream=True
+    )
+    async for chunk in response:
+        if chunk.choices[0].delta.content:
+            print(chunk.choices[0].delta.content, end="", flush=True)
+
+
+asyncio.run(completion())

examples/logging/litellm_completion_stream.py

@@ -0,0 +1,13 @@
+import litellm
+
+from log10.litellm import Log10LitellmLogger
+
+
+log10_handler = Log10LitellmLogger(tags=["litellm_completion", "stream"])
+litellm.callbacks = [log10_handler]
+response = litellm.completion(
+    model="gpt-3.5-turbo", messages=[{"role": "user", "content": "Count to 10."}], stream=True
+)
+for chunk in response:
+    if chunk.choices[0].delta.content:
+        print(chunk.choices[0].delta.content, end="", flush=True)

examples/logging/litellm_image.py

@@ -0,0 +1,34 @@
+import base64
+
+import httpx
+import litellm
+
+from log10.litellm import Log10LitellmLogger
+
+
+log10_handler = Log10LitellmLogger(tags=["litellm_image"])
+litellm.callbacks = [log10_handler]
+
+image_url = "https://upload.wikimedia.org/wikipedia/commons/e/e8/Log10.png"
+image_media_type = "image/png"
+image_data = base64.b64encode(httpx.get(image_url).content).decode("utf-8")
+
+
+model_name = ("gpt-4-vision-preview",)
+model_name = "claude-3-haiku-20240307"
+resp = litellm.completion(
+    model=model_name,
+    messages=[
+        {
+            "role": "user",
+            "content": [
+                {
+                    "type": "image_url",
+                    "image_url": {"url": f"data:{image_media_type};base64,{image_data}"},
+                },
+                {"type": "text", "text": "What's the red curve in the figure, is it log2 or log10? Be concise."},
+            ],
+        }
+    ],
+)
+print(resp.choices[0].message.content)

examples/logging/litellm_image_async_stream.py

@@ -0,0 +1,42 @@
+import asyncio
+import base64
+
+import httpx
+import litellm
+
+from log10.litellm import Log10LitellmLogger
+
+
+log10_handler = Log10LitellmLogger(tags=["litellm_image", "stream", "async"])
+litellm.callbacks = [log10_handler]
+
+image_url = "https://upload.wikimedia.org/wikipedia/commons/e/e8/Log10.png"
+image_media_type = "image/png"
+image_data = base64.b64encode(httpx.get(image_url).content).decode("utf-8")
+
+model_name = "claude-3-haiku-20240307"
+
+
+async def completion():
+    resp = litellm.completion(
+        model=model_name,
+        messages=[
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "image_url",
+                        "image_url": {"url": f"data:{image_media_type};base64,{image_data}"},
+                    },
+                    {"type": "text", "text": "What's the red curve in the figure, is it log2 or log10? Be concise."},
+                ],
+            }
+        ],
+        stream=True,
+    )
+    for chunk in resp:
+        if chunk.choices[0].delta.content:
+            print(chunk.choices[0].delta.content, end="", flush=True)
+
+
+asyncio.run(completion())

examples/logging/litellm_image_stream.py

@@ -0,0 +1,35 @@
+import base64
+
+import httpx
+import litellm
+
+from log10.litellm import Log10LitellmLogger
+
+
+log10_handler = Log10LitellmLogger(tags=["litellm_image", "stream"])
+litellm.callbacks = [log10_handler]
+
+image_url = "https://upload.wikimedia.org/wikipedia/commons/e/e8/Log10.png"
+image_media_type = "image/png"
+image_data = base64.b64encode(httpx.get(image_url).content).decode("utf-8")
+
+model_name = "claude-3-haiku-20240307"
+resp = litellm.completion(
+    model=model_name,
+    messages=[
+        {
+            "role": "user",
+            "content": [
+                {
+                    "type": "image_url",
+                    "image_url": {"url": f"data:{image_media_type};base64,{image_data}"},
+                },
+                {"type": "text", "text": "What's the red curve in the figure, is it log2 or log10? Be concise."},
+            ],
+        }
+    ],
+    stream=True,
+)
+for chunk in resp:
+    if chunk.choices[0].delta.content:
+        print(chunk.choices[0].delta.content, end="", flush=True)

examples/logging/mistralai_chat_no_streaming.py

@@ -0,0 +1,24 @@
+import mistralai
+from mistralai.client import MistralClient
+from mistralai.models.chat_completion import ChatMessage
+
+from log10.load import log10
+
+
+log10(mistralai)
+
+
+def main():
+    model = "mistral-tiny"
+
+    client = MistralClient()
+
+    chat_response = client.chat(
+        model=model,
+        messages=[ChatMessage(role="user", content="10 + 2 * 3=?")],
+    )
+    print(chat_response.choices[0].message.content)
+
+
+if __name__ == "__main__":
+    main()

examples/logging/mistralai_chat_with_streaming.py

@@ -0,0 +1,27 @@
+import mistralai
+from mistralai.client import MistralClient
+from mistralai.models.chat_completion import ChatMessage
+
+from log10.load import log10
+
+
+log10(mistralai)
+
+
+def main():
+    model = "mistral-tiny"
+
+    client = MistralClient()
+
+    response = client.chat_stream(
+        model=model,
+        messages=[ChatMessage(role="user", content="count the odd numbers from 1 to 20.")],
+    )
+    # import ipdb; ipdb.set_trace()
+    for chunk in response:
+        if chunk.choices[0].delta.content is not None:
+            print(chunk.choices[0].delta.content, end="")
+
+
+if __name__ == "__main__":
+    main()

log10/_httpx_utils.py

@@ -2,6 +2,7 @@
 import logging
 import time
 import traceback
+import uuid
 from datetime import datetime, timezone
 
 import httpx
@@ -17,6 +18,7 @@
 _log10_config = Log10Config()
 base_url = _log10_config.url
 httpx_client = httpx.Client()
+httpx_async_client = httpx.AsyncClient()
 
 
 def _get_time_diff(created_at):
@@ -83,19 +85,35 @@ def _try_post_request(url: str, payload: dict = {}) -> httpx.Response:
         logger.error(f"Failed to insert in log10: {payload} with error {err}")
 
 
+async def _try_post_request_async(url: str, payload: dict = {}) -> httpx.Response:
+    headers = {
+        "x-log10-token": _log10_config.token,
+        "x-log10-organization-id": _log10_config.org_id,
+        "Content-Type": "application/json",
+    }
+    payload["organization_id"] = _log10_config.org_id
+    try:
+        res = await httpx_async_client.post(url, headers=headers, json=payload)
+        res.raise_for_status()
+        return res
+    except httpx.HTTPStatusError as http_err:
+        if "401" in str(http_err):
+            logger.error(
+                "Failed authorization. Please verify that LOG10_TOKEN and LOG10_ORG_ID are set correctly and try again."
+                + "\nSee https://github.com/log10-io/log10#%EF%B8%8F-setup for details"
+            )
+        else:
+            logger.error(f"Failed with error: {http_err}")
+    except Exception as err:
+        logger.error(f"Failed to insert in log10: {payload} with error {err}")
+
+
 async def get_completion_id(request: Request):
     if "v1/chat/completions" not in str(request.url):
         logger.warning("Currently logging is only available for v1/chat/completions.")
         return
 
-    completion_url = "/api/completions"
-    res = _try_post_request(url=f"{base_url}{completion_url}")
-    try:
-        completion_id = res.json().get("completionID")
-    except Exception as e:
-        logger.error(f"Failed to get completion ID. Error: {e}. Skipping completion recording.")
-    else:
-        request.headers["x-log10-completion-id"] = completion_id
+    request.headers["x-log10-completion-id"] = str(uuid.uuid4())
 
 
 async def log_request(request: Request):
@@ -125,7 +143,7 @@ async def log_request(request: Request):
     }
     if get_log10_session_tags():
         log_row["tags"] = get_log10_session_tags()
-    _try_post_request(url=f"{base_url}/api/completions/{completion_id}", payload=log_row)
+    await _try_post_request_async(url=f"{base_url}/api/completions/{completion_id}", payload=log_row)
 
 
 class _LogResponse(Response):
@@ -205,7 +223,7 @@ async def aiter_bytes(self, *args, **kwargs):
             }
             if get_log10_session_tags():
                 log_row["tags"] = get_log10_session_tags()
-            _try_post_request(url=f"{base_url}/api/completions/{completion_id}", payload=log_row)
+            await _try_post_request_async(url=f"{base_url}/api/completions/{completion_id}", payload=log_row)
 
 
 class _LogTransport(httpx.AsyncBaseTransport):
@@ -246,7 +264,7 @@ async def handle_async_request(self, request: httpx.Request) -> httpx.Response:
             }
             if get_log10_session_tags():
                 log_row["tags"] = get_log10_session_tags()
-            _try_post_request(url=f"{base_url}/api/completions/{completion_id}", payload=log_row)
+            await _try_post_request_async(url=f"{base_url}/api/completions/{completion_id}", payload=log_row)
             return response
         elif response.headers.get("content-type") == "text/event-stream":
             return _LogResponse(

log10/feedback/feedback.py

@@ -59,7 +59,7 @@ def create(
         res = self._post_request(self.feedback_create_url, json_payload)
         return res
 
-    def list(self, offset: int = 0, limit: int = 50, task_id: str = None) -> httpx.Response:
+    def list(self, offset: int = 0, limit: int = 50, task_id: str = "") -> httpx.Response:
         base_url = self._log10_config.url
         api_url = "/api/v1/feedback"
         url = f"{base_url}{api_url}?organization_id={self._log10_config.org_id}&offset={offset}&limit={limit}&task_id={task_id}"