Update PR template and more error message

Author: zhaoqizqwang

.github/pull_request_template.md

@@ -1,35 +1,22 @@
-## What's changing and why?
+### What's changing and why?
 <!-- Describe what you're changing and the motivation behind it -->
 
 
-## Before/After UX
+### Before/After UX
 <!-- Show the user experience before and after your changes -->
 **Before:**
 
 
 **After:**
 
 
-## How was this change tested?
+### How was this change tested?
 <!-- Describe your testing approach -->
 
+## Do we need integration tests and why?
 
-## Unit test coverage
-<!-- Check unit test coverage for your changes -->
-- [ ] All new/modified code has unit tests
-- [ ] Coverage verified for changed code
-- [ ] N/A - no testable code changes
+### For Reviewers
+#### No PR should be approved if PR check fails. Author must provide reason in comment when failed PR check
 
-## Do we need integration tests?
-<!-- Consider if integration tests are needed -->
-- [ ] Yes - integration tests added
-- [ ] No - unit tests sufficient
-- [ ] No - infrastructure/config change only
-- [ ] Unsure - please advise
-
----
-
-## Checklist
-- [ ] PR title clearly describes the change
-- [ ] No sensitive information exposed and security is maintained
-- [ ] Ready for review
+- [ ] All PR checks have passed, or the author has provided reason for bypassing.
+- [ ] Unit tests and necessary integration tests have been added.

src/sagemaker/hyperpod/training/quota_allocation_util.py

@@ -247,9 +247,9 @@ def _is_valid(vcpu: Optional[float], memory_in_gib: Optional[float], accelerator
         return False, f"Invalid instance-type {instance_type}. Please re-check the instance type and contact AWS for support."
 
     if instance_type is not None:
-        #both resources and node count specified
+        # both resources and node count specified
         if (has_gpu_quota_allocation and node_specified):
-            return False, f"Missing input: either node-count or a combination of accelerators, vcpu, memory-in-gib must be specified for instance-type {instance_type}"
+            return False, f"Invalid input: For instance-type {instance_type}, specify either node-count OR resource allocation (accelerators, vcpu, memory-in-gib), but not both."
     return True, ""
 
 

test/integration_tests/training/cli/test_gpu_quota_allocation.py

@@ -236,8 +236,7 @@ def test_invalid_node_count_accelerators_parameter(self, test_job_name):
                     text=True
                 )
         assert result.returncode != 0
-        assert "Missing input: either node-count or a combination of accelerators, vcpu, " in result.stdout
-        assert "memory-in-gib must be specified for instance-type ml.g5.8xlarge" in result.stdout
+        assert "Invalid input: For instance-type ml.g5.xlarge, specify either node-count OR resource allocation (accelerators, vcpu, memory-in-gib), but not both." in result.stdout
 
     def test_invalid_no_node_count_or_quota_parameter(self, test_job_name):
         """Test that case where both node-count and any of the quota parameters are provided"""

test/unit_tests/cli/test_quota_allocation_util.py

@@ -204,7 +204,7 @@ def test_is_valid_invalid_instance_type(self):
     def test_is_valid_both_node_count_and_resources(self):
         valid, message = _is_valid(4.0, None, None, 2, "ml.g5.xlarge")
         assert not valid
-        assert message == "Missing input: either node-count or a combination of accelerators, vcpu, memory-in-gib must be specified for instance-type ml.g5.xlarge"
+        assert message == "Invalid input: For instance-type ml.g5.xlarge, specify either node-count OR resource allocation (accelerators, vcpu, memory-in-gib), but not both."
 
     def test_is_valid_both_node_count_and_limits(self):
         valid, message = _is_valid(None, None, None, 2, "ml.g5.xlarge")