docs: update capturing of token counts when using the decorator

Author: marcklingen

components-mdx/get-started-python-decorator-any-llm.mdx

@@ -17,8 +17,19 @@ def anthropic_completion(**kwargs):
       metadata=kwargs_clone
   )
 
+  response = anthopic_client.messages.create(**kwargs)
+
+  # See docs for more details on token counts and usd cost in Langfuse
+  # https://langfuse.com/docs/model-usage-and-cost
+  langfuse_context.update_current_observation(
+      usage={
+          "input": response.usage.input_tokens,
+          "output": response.usage.output_tokens
+      }
+  )
+
   # return result
-  return anthopic_client.messages.create(**kwargs).content[0].text
+  return response.content[0].text
 
 @observe()
 def main():

cookbook/python_decorators.ipynb

@@ -236,8 +236,19 @@
         "      metadata=kwargs_clone\n",
         "  )\n",
         "  \n",
+        "  response = anthopic_client.messages.create(**kwargs)\n",
+        "\n",
+        "  # See docs for more details on token counts and usd cost in Langfuse\n",
+        "  # https://langfuse.com/docs/model-usage-and-cost\n",
+        "  langfuse_context.update_current_observation(\n",
+        "      usage={\n",
+        "          \"input\": response.usage.input_tokens,\n",
+        "          \"output\": response.usage.output_tokens\n",
+        "      }\n",
+        "  )\n",
+        "\n",
         "  # return result\n",
-        "  return anthopic_client.messages.create(**kwargs).content[0].text\n",
+        "  return response.content[0].text\n",
         "\n",
         "@observe()\n",
         "def main():\n",
@@ -256,7 +267,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "> **Example trace**: https://cloud.langfuse.com/project/cloramnkj0002jz088vzn1ja4/traces/ece9079d-e12c-4c0e-9dc3-8805d0bbe8ec?observation=723c04ff-cdca-4716-8143-e691129be315"
+        "> **Example trace**: https://cloud.langfuse.com/project/cloramnkj0002jz088vzn1ja4/traces/66d06dd7-eeec-40c1-9b11-aac0e9c4f2fe?observation=d48a45f8-593c-4013-8a8a-23665b94aeda"
       ]
     },
     {

pages/docs/model-usage-and-cost.mdx

@@ -38,7 +38,56 @@ If available in the LLM response, ingesting usage and/or cost is the most accura
 
 Many of the Langfuse integrations automatically capture usage and cost data from the LLM response. If this does not work as expected, please create an [issue](/issue) on GitHub.
 
-<Tabs items={["Python", "JS"]}>
+<Tabs items={["Python (Decorator)", "Python (low-level SDK)", "JS"]}>
+<Tab>
+
+```python
+@observe(as_type="generation")
+def anthropic_completion(**kwargs):
+  # optional, extract some fields from kwargs
+  kwargs_clone = kwargs.copy()
+  input = kwargs_clone.pop('messages', None)
+  model = kwargs_clone.pop('model', None)
+  langfuse_context.update_current_observation(
+      input=input,
+      model=model,
+      metadata=kwargs_clone
+  )
+
+  response = anthopic_client.messages.create(**kwargs)
+
+  langfuse_context.update_current_observation(
+      usage={
+          "input": response.usage.input_tokens,
+          "output": response.usage.output_tokens,
+          # "total": int,  # if not set, it is derived from input + output
+          "unit": "TOKENS", # any of: "TOKENS", "CHARACTERS", "MILLISECONDS", "SECONDS", "IMAGES"
+
+          # Optionally, also ingest usd cost. Alternatively, you can infer it via a model definition in Langfuse.
+          # Here we assume the input and output cost are 1 USD each.
+          "input_cost": 1,
+          "output_cost": 1,
+          # "total_cost": float, # if not set, it is derived from input_cost + output_cost
+      }
+  )
+
+  # return result
+  return response.content[0].text
+
+@observe()
+def main():
+  return anthropic_completion(
+      model="claude-3-opus-20240229",
+      max_tokens=1024,
+      messages=[
+          {"role": "user", "content": "Hello, Claude"}
+      ]
+  )
+
+main()
+```
+
+</Tab>
 <Tab>
 
 ```python

pages/docs/sdk/python/example.md

@@ -156,8 +156,19 @@ def anthropic_completion(**kwargs):
       metadata=kwargs_clone
   )
 
+  response = anthopic_client.messages.create(**kwargs)
+
+  # See docs for more details on token counts and usd cost in Langfuse
+  # https://langfuse.com/docs/model-usage-and-cost
+  langfuse_context.update_current_observation(
+      usage={
+          "input": response.usage.input_tokens,
+          "output": response.usage.output_tokens
+      }
+  )
+
   # return result
-  return anthopic_client.messages.create(**kwargs).content[0].text
+  return response.content[0].text
 
 @observe()
 def main():
@@ -172,7 +183,7 @@ def main():
 main()
 ```
 
-> **Example trace**: https://cloud.langfuse.com/project/cloramnkj0002jz088vzn1ja4/traces/ece9079d-e12c-4c0e-9dc3-8805d0bbe8ec?observation=723c04ff-cdca-4716-8143-e691129be315
+> **Example trace**: https://cloud.langfuse.com/project/cloramnkj0002jz088vzn1ja4/traces/66d06dd7-eeec-40c1-9b11-aac0e9c4f2fe?observation=d48a45f8-593c-4013-8a8a-23665b94aeda
 
 ## Customize input/output
 

pages/guides/cookbook/python_decorators.md

@@ -156,8 +156,19 @@ def anthropic_completion(**kwargs):
       metadata=kwargs_clone
   )
 
+  response = anthopic_client.messages.create(**kwargs)
+
+  # See docs for more details on token counts and usd cost in Langfuse
+  # https://langfuse.com/docs/model-usage-and-cost
+  langfuse_context.update_current_observation(
+      usage={
+          "input": response.usage.input_tokens,
+          "output": response.usage.output_tokens
+      }
+  )
+
   # return result
-  return anthopic_client.messages.create(**kwargs).content[0].text
+  return response.content[0].text
 
 @observe()
 def main():
@@ -172,7 +183,7 @@ def main():
 main()
 ```
 
-> **Example trace**: https://cloud.langfuse.com/project/cloramnkj0002jz088vzn1ja4/traces/ece9079d-e12c-4c0e-9dc3-8805d0bbe8ec?observation=723c04ff-cdca-4716-8143-e691129be315
+> **Example trace**: https://cloud.langfuse.com/project/cloramnkj0002jz088vzn1ja4/traces/66d06dd7-eeec-40c1-9b11-aac0e9c4f2fe?observation=d48a45f8-593c-4013-8a8a-23665b94aeda
 
 ## Customize input/output