[BUG] Synapse GPT-4, OpenAIChatCompletion, API documentation: mandatory "name" field not mentioned in documentation for "messages"

[EPMSFT]

SynapseML version

0.11.2-spark3.3

System information

• Language version: python 3.10
• Spark Version: 3.3
• Spark Platform: Synapse

Describe the problem

Referring to:

SynapseML/cognitive/src/main/scala/com/microsoft/azure/synapse/ml/cognitive/openai/OpenAIChatCompletion.scala

Line 31 in 0836e40

" in the chat format. This column should have type Array(Struct(role: String, content: String)).")

Documentation says messages is of type: Array(Struct(role: String, content: String))

However, if I don't add a "name" field and populate it with the deployment name, completion.transform(df) crashes because it can't find the "name" field.

Code to reproduce issue

import synapse.ml.cognitive.openai.OpenAIChatCompletion as OpenAIChatCompletion
from synapse.ml.core.platform import find_secret
import pyspark.sql.types as T

key = find_secret(...)
deployment_name = "openai-xxx-gpt-4-xxx"
service_name = "openai-xxx"
user = "xxx"

Documentation says: Array(Struct(role: String, content: String)),

but if I do that it fails with cryptic error; "name" (deployment) is needed

msg_works = [
{"role": "system", "content": "You can only answer with 'yes' or 'no'.", "name": deployment_name},
{"role": "assistant", "content": "Can you read this?", "name": deployment_name}
]
msg_fails = [
{"role": "system", "content": "You can only answer with 'yes' or 'no'."},
{"role": "assistant", "content": "Can you read this?"}
]

df = sc.parallelize([[msg_fails]], 1).toDF(
T.StructType([
T.StructField("messages", T.ArrayType(
T.StructType([
T.StructField("role", T.StringType(), False),
T.StructField("content", T.StringType(), False),
# T.StructField("name", T.StringType(), False) # add to make it work
]), False
), False)
])
)

completion = (
OpenAIChatCompletion()
.setSubscriptionKey(key)
.setDeploymentName(deployment_name)
.setUrl(f"https://{service_name}.openai.azure.com/")
.setMessagesCol("messages")
.setMaxTokens(100)
.setTemperature(0.25)
.setFrequencyPenalty(0.0)
.setPresencePenalty(0.0)
.setUser(user)
.setErrorCol("error")
.setOutputCol("response")
)

response = completion.transform(df)
display(response)

Other info / logs

No response

What component(s) does this bug affect?

• area/cognitive: Cognitive project
• area/core: Core project
• area/deep-learning: DeepLearning project
• area/lightgbm: Lightgbm project
• area/opencv: Opencv project
• area/vw: VW project
• area/website: Website
• area/build: Project build system
• area/notebooks: Samples under notebooks folder
• area/docker: Docker usage
• area/models: models related issue

What language(s) does this bug affect?

• language/scala: Scala source code
• language/python: Pyspark APIs
• language/r: R APIs
• language/csharp: .NET APIs
• language/new: Proposals for new client languages

What integration(s) does this bug affect?

• integrations/synapse: Azure Synapse integrations
• integrations/azureml: Azure ML integrations
• integrations/databricks: Databricks integrations

[github-actions]

Hey @EPMSFT 👋!
Thank you so much for reporting the issue/feature request 🚨.
Someone from SynapseML Team will be looking to triage this issue soon.
We appreciate your patience.