init commit

configs/instruction_expansion_api.json

@@ -0,0 +1,16 @@
+{
+  "job_type": "instruction_expansion_api",
+  "dataset": {
+    "input_path": "./train.json",
+    "output_path": "./train_extended.json",
+    "num_in_context_samples": 3,
+    "num_output_samples": 10
+  },
+  "inference":{
+    "base_url": "ENDPOINT",
+    "api_key": "TOKEN",
+    "stream": true,
+    "prompt" : "Assume you are a data synthesis expert. Given a few instructions as in-context examples, you should generate a new instruction similar to the examples to support the training of large language models. You should place your answer enclosed within <answer></answer> tags. The examples are as follows:",
+    "max_new_tokens": 512
+  }
+}