The Writer API supports streaming responses from the API. Streams use server-side events so that you can display content as the API generates it in real time.
Streaming improves user experience by reducing latency from the time the user submits a request to the time the response is available.
The code below shows a streaming request and response from a text generation request using curl. The response is a stream of server-sent events.
Copy
curl --location 'https://api.writer.com/v1/completions' \--header 'Content-Type: application/json' \--header "Authorization: Bearer $WRITER_API_KEY" \--data '{ "model": "palmyra-x5", "prompt": "Respond to a customer chat request about a delayed shipment with a message that apologizes for the delay, offers a tracking number, and provides a new estimated delivery date.", "stream": true}'
When you stream a response using a Writer SDK, the SDK creates an iterator that yields chunks of the response. You can iterate over the stream to receive the response.
See below for examples of streaming with the Python and JavaScript SDKs for each endpoint that supports streaming.
Copy
from writerai import Writer# Initialize the client. If you don't pass the `api_key` parameter,# the client looks for the `WRITER_API_KEY` environment variable.client = Writer()text_generation = client.completions.create( model="palmyra-x5", prompt="Respond to a customer chat request about a delayed shipment with a message that apologizes for the delay, offers a tracking number, and provides a new estimated delivery date.", stream=True)for chunk in text_generation: print(chunk.value, end="", flush=True)
Copy
from writerai import Writer# Initialize the client. If you don't pass the `api_key` parameter,# the client looks for the `WRITER_API_KEY` environment variable.client = Writer()text_generation = client.completions.create( model="palmyra-x5", prompt="Respond to a customer chat request about a delayed shipment with a message that apologizes for the delay, offers a tracking number, and provides a new estimated delivery date.", stream=True)for chunk in text_generation: print(chunk.value, end="", flush=True)
Copy
from writerai import Writer# Initialize the client. If you don't pass the `api_key` parameter,# the client looks for the `WRITER_API_KEY` environment variable.client = Writer()stream = client.chat.chat( model="palmyra-x5", messages=[{"role": "user", "content": "Respond to a customer chat request about a delayed shipment with a message that apologizes for the delay, offers a tracking number, and provides a new estimated delivery date."}], stream=True)for chunk in stream: if chunk.choices[0].delta.content: print(chunk.choices[0].delta.content, end="", flush=True)
Copy
from writerai import Writer# Initialize the client. If you don't pass the `api_key` parameter,# the client looks for the `WRITER_API_KEY` environment variable.client = Writer()stream = client.applications.generate_content( application_id="<application-id>", inputs=[ { "id": "query", "value": ["Provide a list of three hotels in San Francisco near Union Square within the price range of $100 to $200 per night"] } ], stream=True)for chunk in stream: if chunk.delta.content: print(chunk.delta.content, end="", flush=True) else if chunk.delta.stages: print(chunk.delta.stages[0].content)
Copy
from writerai import Writer# Initialize the client. If you don't pass the `api_key` parameter,# the client looks for the `WRITER_API_KEY` environment variable.client = Writer()stream = client.graphs.question( graph_ids=["<graph-id>"], question="What is the generic name for the drug Bavencio?", stream=True)for chunk in stream: print(chunk.answer, end="", flush=True)
The Python and Node SDKs include streaming helpers for chat completions. These helpers provide more granular details about the streaming events and accumulate the response.
To use the streaming helpers, call client.chat.stream. Then, include all the same parameters as you would for a non-streaming chat completion request, except omit the stream parameter.
Copy
from writerai import Writer# Initialize the client. If you don't pass the `api_key` parameter,# the client looks for the `WRITER_API_KEY` environment variable.client = Writer()with client.chat.stream( model="palmyra-x5", messages=[{"role": "user", "content": "Respond to a customer chat request about a delayed shipment with a message that apologizes for the delay, offers a tracking number, and provides a new estimated delivery date."}]) as stream: for event in stream: if event.type == "content.delta": print(event.value, end="", flush=True)# print the final responsecompletion = stream.get_final_completion()print(completion.choices[0].message.content)
For more information about the streaming helpers for chat completions, see the Python and Node SDKs.
