fix: adapt test_config.json to new format

Author: Chamberlain0w0

scripts/test_config.json

@@ -211,6 +211,18 @@
                         "use_distributed_optimizer": true
                     }
                 },
+                {
+                    "id": "3_distopt_zero2",
+                    "args": {
+                        "dtype": "float32",
+                        "nthread_per_process": 8,
+                        "num_iteration": 10,
+                        "batch_size": 10,
+                        "total_batch_size": 5120,
+                        "use_distributed_optimizer": true,
+                        "zero_stage": 2
+                    }
+                },
                 {
                     "id": "3_bfloat16_distopt",
                     "args": {
@@ -222,6 +234,18 @@
                         "use_distributed_optimizer": true
                     }
                 },
+                {
+                    "id": "3_bfloat16_distopt_zero2",
+                    "args": {
+                        "dtype": "bfloat16",
+                        "nthread_per_process": 8,
+                        "num_iteration": 10,
+                        "batch_size": 10,
+                        "total_batch_size": 5120,
+                        "use_distributed_optimizer": true,
+                        "zero_stage": 2
+                    }
+                },
                 {
                     "id": "4_distopt",
                     "args": {
@@ -234,6 +258,19 @@
                         "use_distributed_optimizer": true
                     }
                 },
+                {
+                    "id": "4_distopt_zero2",
+                    "args": {
+                        "dtype": "float32",
+                        "nthread_per_process": 8,
+                        "num_iteration": 10,
+                        "batch_size": 40,
+                        "total_batch_size": 5120,
+                        "tensor_parallel": 4,
+                        "use_distributed_optimizer": true,
+                        "zero_stage": 2
+                    }
+                },
                 {
                     "id": "4_bfloat16_distopt",
                     "args": {
@@ -246,6 +283,19 @@
                         "use_distributed_optimizer": true
                     }
                 },
+                {
+                    "id": "4_bfloat16_distopt_zero2",
+                    "args": {
+                        "dtype": "bfloat16",
+                        "nthread_per_process": 8,
+                        "num_iteration": 10,
+                        "batch_size": 40,
+                        "total_batch_size": 5120,
+                        "tensor_parallel": 4,
+                        "use_distributed_optimizer": true,
+                        "zero_stage": 2
+                    }
+                },
                 {
                     "id": "5_distopt",
                     "args": {
@@ -259,6 +309,20 @@
                         "use_distributed_optimizer": true
                     }
                 },
+                {
+                    "id": "5_distopt_zero2",
+                    "args": {
+                        "dtype": "float32",
+                        "nthread_per_process": 8,
+                        "num_iteration": 10,
+                        "batch_size": 40,
+                        "total_batch_size": 5120,
+                        "tensor_parallel": 4,
+                        "sequence_parallel": true,
+                        "use_distributed_optimizer": true,
+                        "zero_stage": 2
+                    }
+                },
                 {
                     "id": "5_bfloat16_distopt",
                     "args": {
@@ -272,6 +336,20 @@
                         "use_distributed_optimizer": true
                     }
                 },
+                {
+                    "id": "5_bfloat16_distopt_zero2",
+                    "args": {
+                        "dtype": "bfloat16",
+                        "nthread_per_process": 8,
+                        "num_iteration": 10,
+                        "batch_size": 40,
+                        "total_batch_size": 5120,
+                        "tensor_parallel": 4,
+                        "sequence_parallel": true,
+                        "use_distributed_optimizer": true,
+                        "zero_stage": 2
+                    }
+                },
                 {
                     "id": "8_distopt",
                     "args": {