How is the coding performance?
#1
by rombodawg - opened
Does the coding performance compare to gpt-3.5 as in chatgpt? Is it far bellow that? Also can you share what all coding data you trained the model on?
It is lower than gpt-3.5 in terms of encoding. Llama 2 was likely trained on very little encoded data. Also, all data can be found at https://huggingface.co/datasets/openchat/openchat_sharegpt_v3
It is lower than gpt-3.5 in terms of encoding. Llama 2 was likely trained on very little encoded data. Also, all data can be found at https://huggingface.co/datasets/openchat/openchat3_orca_dataset
The link is not working
Sorry, typo, fixed now