codet5-small-go_generation_v2

This model is finetuned based on the pre-trained CodeT5-small model.

This model is fine-tuned on dataset: data_71421(44.2MB)

max_src_len = 512, max_trg_len = 256

2023.5.15 update README.md

2023.5.5 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

# load model and tokenizer
model_path = "PPY039/codet5-small-go_generation_v2"

model = T5ForConditionalGeneration.from_pretrained(model_path, cache_dir="D:\huggingface_cache")
tokenizer = RobertaTokenizer.from_pretrained(model_path, cache_dir="D:\huggingface_cache")

# use model
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)  # max_trg_len = 256

output_text = tokenizer.decode(output[0], skip_special_tokens=True)

# this prints "{\n\treturn kmeta.ChildName(rev.GetName(), "-pa")\n}"
print(output_text)

Training data

YinShicheng

Training process

GuQiuhan

Advisor

WangYu

Evaluation results

TODO