[Update] modified and add files related to fintuning with internlm2_7b_base

[chg0901]

1. add internlm2_7b_base_qlora_e3.py and modify requirements.txt
2. update data.json [ delete 4 empty lines in data.json (425 483 742 1120)]
3. add dataset processing codes
   • update process.py for multi_turn_dataset(1 and 2) and data.json, data_pro.json to add system prompt
   • add datasets\processed\process_single_turn_conversation_construction.py for single-turn dataset (1 and 2) to modify keys (input and ouput) and add system prompt in each conversation
   • add datasets\processed\process_merge.py for merging these 6 updated dataset in datasets\processed\ as a combined dataset combined_data.json

[aJupyter]

下次请pr到dev分支