add readme

README.md

@@ -1,3 +1,46 @@
 ---
 license: apache-2.0
 ---
+
+# CodeT5-small-Go_generation
+This model is finetuned based on the pre-trained [CodeT5-small model](https://github.com/salesforce/CodeT5#fine-tuning). 
+
+> 5.3 upload the initial version.
+
+The model genarates the missing function body according to the input which privides the necessary class environment and an empty function.
+
+See example below for formatting.
+ 
+# How to use
+Here is how to use this model:
+
+from transformers import T5ForConditionalGeneration, RobertaTokenizer
+
+# 加载模型和tokenizer
+model_path = "intm/codet5-small-go_generation"
+tokenizer = RobertaTokenizer.from_pretrained('Salesforce/codet5-base')
+model = T5ForConditionalGeneration.from_pretrained(model_path)
+
+# 使用模型进行推理
+input_text = "package names\n\nimport \"knative.dev/pkg/kmeta\"\n\n\nfunc Deployment(rev kmeta.Accessor) string {\n\treturn kmeta.ChildName(rev.GetName(), \"-deployment\")\n}\n\n\nfunc ImageCache(rev kmeta.Accessor) string {\n\treturn kmeta.ChildName(rev.GetName(), \"-cache\")\n}\n\n\n\n\nfunc PA(rev kmeta.Accessor) string"
+input_ids = tokenizer.encode(input_text, return_tensors="pt")
+output = model.generate(input_ids=input_ids, max_new_tokens=256)  #最大长度按照数据集的max_trg_len设置
+
+# 将生成的结果转换为字符串
+output_text = tokenizer.decode(output[0], skip_special_tokens=True)
+print(output_text)
+
+
+# this prints "return kmeta.ChildName(rev.GetName(), "-pa")"
+
+# Training data
+YinShicheng
+
+# Training process
+GuQiuhan
+
+# Advisor
+Prof.WangYu
+
+# Evaluation results
+TODO

example_usage.py

@@ -0,0 +1,20 @@
+
+
+from transformers import T5ForConditionalGeneration, RobertaTokenizer
+
+# 加载模型和tokenizer
+model_path = "intm/codet5-small-go_generation"
+tokenizer = RobertaTokenizer.from_pretrained('Salesforce/codet5-base')
+model = T5ForConditionalGeneration.from_pretrained(model_path)
+
+# 使用模型进行推理
+input_text = "package names\n\nimport \"knative.dev/pkg/kmeta\"\n\n\nfunc Deployment(rev kmeta.Accessor) string {\n\treturn kmeta.ChildName(rev.GetName(), \"-deployment\")\n}\n\n\nfunc ImageCache(rev kmeta.Accessor) string {\n\treturn kmeta.ChildName(rev.GetName(), \"-cache\")\n}\n\n\n\n\nfunc PA(rev kmeta.Accessor) string"
+input_ids = tokenizer.encode(input_text, return_tensors="pt")
+output = model.generate(input_ids=input_ids, max_new_tokens=256)  #最大长度按照数据集的max_trg_len设置
+
+# 将生成的结果转换为字符串
+output_text = tokenizer.decode(output[0], skip_special_tokens=True)
+print(output_text)
+
+
+# 应当可以输出：return rev.GetName()