关于云脑任务中统一路径访问方式的公告>>>
OpenIOSSG
/
OpenI_LLM_Finetune_Example
Not watched
14
1
Fork 3
Code
Releases 0 Wiki Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
Branch: master
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'master'
${ noResults }
OpenI_LLM_Finetune_Example/.openi-aitemplate

542 lines
11 KiB
Raw Permalink Blame History 

  1. [

  2.   {

  3.     "Name": "llama finetune npu",

  4.     "Description": "",

  5.     "JobType": "TRAIN",

  6.     "Cluster": "C2Net",

  7.     "ComputeSource": "NPU",

  8.     "HasInternet": 2,

  9.     "AccCardsNum": 1,

  10.     "AccCardType": "ASCEND910",

  11.     "CpuCores": 24,

  12.     "MemGiB": 96,

  13.     "GPUMemGiB": 32,

  14.     "ShareMemGiB": 0,

  15.     "Image": {

  16.       "ImageID": "0b79fed7504c4etrain0llamafactory",

  17.       "ImageName": "torch-npu-cann8-train",

  18.       "ImageUrl": ""

  19.     },

  20.     "PretrainModelList": [

  21.       {

  22.         "ID": "94f3dd64-2a66-4612-bb7d-6ca5cac7b2f2",

  23.         "ModelName": "Qwen2-0.5B"

  24.       }

  25.     ],

  26.     "DatasetList": [

  27.       {

  28.         "ID": "a924818a-b82c-4c48-bf78-dd517077244f",

  29.         "DatasetName": "huanhuan.zip"

  30.       },

  31.       {

  32.         "ID": "937a8ab2-3c01-4be7-a4a0-c59be18809d1",

  33.         "DatasetName": "glaive_toolcall_zh_demo.zip"

  34.       }

  35.     ],

  36.     "BranchName": "master",

  37.     "BootFile": "finetune.py",

  38.     "Parameters": [

  39.       {

  40.         "Label": "stage",

  41.         "Value": "sft"

  42.       },

  43.       {

  44.         "Label": "preprocessing_numworkers",

  45.         "Value": "16"

  46.       },

  47.       {

  48.         "Label": "finetuning_type",

  49.         "Value": "lora"

  50.       },

  51.       {

  52.         "Label": "template",

  53.         "Value": "qwen"

  54.       },

  55.       {

  56.         "Label": "dataset",

  57.         "Value": "huanhuan,glaive_toolcall_zh_demo"

  58.       },

  59.       {

  60.         "Label": "model_name",

  61.         "Value": "Qwen2-0.5B"

  62.       },

  63.       {

  64.         "Label": "cutoff_len",

  65.         "Value": "1024"

  66.       },

  67.       {

  68.         "Label": "learning_rate",

  69.         "Value": "1.0e-4"

  70.       },

  71.       {

  72.         "Label": "num_train_epochs",

  73.         "Value": "3"

  74.       },

  75.       {

  76.         "Label": "max_samples",

  77.         "Value": "1000"

  78.       },

  79.       {

  80.         "Label": "per_device_train_batch_size",

  81.         "Value": "1"

  82.       },

  83.       {

  84.         "Label": "gradient_accumulation_steps",

  85.         "Value": "2"

  86.       },

  87.       {

  88.         "Label": "lr_scheduler_type",

  89.         "Value": "cosine"

  90.       },

  91.       {

  92.         "Label": "logging_steps",

  93.         "Value": "10"

  94.       },

  95.       {

  96.         "Label": "save_steps",

  97.         "Value": "500"

  98.       },

  99.       {

  100.         "Label": "warmup_ratio",

  101.         "Value": "0.1"

  102.       },

  103.       {

  104.         "Label": "val_size",

  105.         "Value": "0.1"

  106.       },

  107.       {

  108.         "Label": "per_device_eval_batch_size",

  109.         "Value": "1"

  110.       },

  111.       {

  112.         "Label": "eval_strategy",

  113.         "Value": "steps"

  114.       },

  115.       {

  116.         "Label": "eval_steps",

  117.         "Value": "500"

  118.       },

  119.       {

  120.         "Label": "plot_loss",

  121.         "Value": "true"

  122.       },

  123.       {

  124.         "Label": "overwrite_output_dir",

  125.         "Value": "true"

  126.       },

  127.       {

  128.         "Label": "fp16",

  129.         "Value": "true"

  130.       },

  131.       {

  132.         "Label": "overwrite_cache",

  133.         "Value": "true"

  134.       }

  135.     ]

  136.   },

  137.   {

  138.     "Name": "llama finetune  gpu",

  139.     "Description": "",

  140.     "JobType": "TRAIN",

  141.     "Cluster": "C2Net",

  142.     "ComputeSource": "GPU",

  143.     "HasInternet": 2,

  144.     "AccCardsNum": 1,

  145.     "AccCardType": "V100",

  146.     "CpuCores": 8,

  147.     "MemGiB": 50,

  148.     "GPUMemGiB": 32,

  149.     "ShareMemGiB": 0,

  150.     "Image": {

  151.       "ImageID": "",

  152.       "ImageName": "",

  153.       "ImageUrl": "192.168.242.22:443/default-workspace/2a72307689ae49758c80c896fffda0a1/image:llama-factory-gpu-v2"

  154.     },

  155.     "PretrainModelList": [

  156.       {

  157.         "ID": "94f3dd64-2a66-4612-bb7d-6ca5cac7b2f2",

  158.         "ModelName": "Qwen2-0.5B"

  159.       }

  160.     ],

  161.     "DatasetList": [

  162.       {

  163.         "ID": "a924818a-b82c-4c48-bf78-dd517077244f",

  164.         "DatasetName": "huanhuan.zip"

  165.       },

  166.       {

  167.         "ID": "30e51e73-28b5-4040-a3ea-e63ca8332dba",

  168.         "DatasetName": "alpaca_zh_demo.zip"

  169.       }

  170.     ],

  171.     "BranchName": "master",

  172.     "BootFile": "finetune.py",

  173.     "Parameters": [

  174.       {

  175.         "Label": "stage",

  176.         "Value": "sft"

  177.       },

  178.       {

  179.         "Label": "preprocessing_numworkers",

  180.         "Value": "16"

  181.       },

  182.       {

  183.         "Label": "finetuning_type",

  184.         "Value": "lora"

  185.       },

  186.       {

  187.         "Label": "template",

  188.         "Value": "qwen"

  189.       },

  190.       {

  191.         "Label": "dataset",

  192.         "Value": "huanhuan,alpaca_zh_demo"

  193.       },

  194.       {

  195.         "Label": "cutoff_len",

  196.         "Value": "1024"

  197.       },

  198.       {

  199.         "Label": "learning_rate",

  200.         "Value": "1.0e-4"

  201.       },

  202.       {

  203.         "Label": "num_train_epochs",

  204.         "Value": "3"

  205.       },

  206.       {

  207.         "Label": "max_samples",

  208.         "Value": "1000"

  209.       },

  210.       {

  211.         "Label": "per_device_train_batch_size",

  212.         "Value": "1"

  213.       },

  214.       {

  215.         "Label": "gradient_accumulation_steps",

  216.         "Value": "2"

  217.       },

  218.       {

  219.         "Label": "lr_scheduler_type",

  220.         "Value": "cosine"

  221.       },

  222.       {

  223.         "Label": "logging_steps",

  224.         "Value": "10"

  225.       },

  226.       {

  227.         "Label": "save_steps",

  228.         "Value": "500"

  229.       },

  230.       {

  231.         "Label": "warmup_ratio",

  232.         "Value": "0.1"

  233.       },

  234.       {

  235.         "Label": "val_size",

  236.         "Value": "0.1"

  237.       },

  238.       {

  239.         "Label": "per_device_eval_batch_size",

  240.         "Value": "1"

  241.       },

  242.       {

  243.         "Label": "eval_strategy",

  244.         "Value": "steps"

  245.       },

  246.       {

  247.         "Label": "eval_steps",

  248.         "Value": "500"

  249.       },

  250.       {

  251.         "Label": "model_name",

  252.         "Value": "Qwen2-0.5B"

  253.       },

  254.       {

  255.         "Label": "plot_loss",

  256.         "Value": "true"

  257.       },

  258.       {

  259.         "Label": "overwrite_output_dir",

  260.         "Value": "true"

  261.       },

  262.       {

  263.         "Label": "fp16",

  264.         "Value": "true"

  265.       },

  266.       {

  267.         "Label": "overwrite_cache",

  268.         "Value": "true"

  269.       }

  270.     ]

  271.   },

  272.   {

  273.     "Name": "llama finetune gpgpu",

  274.     "Description": "",

  275.     "JobType": "TRAIN",

  276.     "Cluster": "C2Net",

  277.     "ComputeSource": "ILUVATAR-GPGPU",

  278.     "HasInternet": 2,

  279.     "AccCardsNum": 1,

  280.     "AccCardType": "BI-V100",

  281.     "CpuCores": 30,

  282.     "MemGiB": 64,

  283.     "GPUMemGiB": 32,

  284.     "ShareMemGiB": 0,

  285.     "Image": {

  286.       "ImageID": "",

  287.       "ImageName": "",

  288.       "ImageUrl": "10.208.20.54:1443/default-workspace/1e53e45fca0f4b44b15ca8d00025f2ba/image:llama-factory-gpgpu-train-v2"

  289.     },

  290.     "PretrainModelList": [

  291.       {

  292.         "ID": "94f3dd64-2a66-4612-bb7d-6ca5cac7b2f2",

  293.         "ModelName": "Qwen2-0.5B"

  294.       }

  295.     ],

  296.     "DatasetList": [

  297.       {

  298.         "ID": "30e51e73-28b5-4040-a3ea-e63ca8332dba",

  299.         "DatasetName": "alpaca_zh_demo.zip"

  300.       },

  301.       {

  302.         "ID": "a924818a-b82c-4c48-bf78-dd517077244f",

  303.         "DatasetName": "huanhuan.zip"

  304.       }

  305.     ],

  306.     "BranchName": "master",

  307.     "BootFile": "finetune_gpgpu.py",

  308.     "Parameters": [

  309.       {

  310.         "Label": "stage",

  311.         "Value": "sft"

  312.       },

  313.       {

  314.         "Label": "preprocessing_numworkers",

  315.         "Value": "16"

  316.       },

  317.       {

  318.         "Label": "finetuning_type",

  319.         "Value": "lora"

  320.       },

  321.       {

  322.         "Label": "template",

  323.         "Value": "qwen"

  324.       },

  325.       {

  326.         "Label": "dataset",

  327.         "Value": "alpaca_zh_demo,huanhuan"

  328.       },

  329.       {

  330.         "Label": "cutoff_len",

  331.         "Value": "1024"

  332.       },

  333.       {

  334.         "Label": "learning_rate",

  335.         "Value": "1.0e-4"

  336.       },

  337.       {

  338.         "Label": "num_train_epochs",

  339.         "Value": "3"

  340.       },

  341.       {

  342.         "Label": "max_samples",

  343.         "Value": "1000"

  344.       },

  345.       {

  346.         "Label": "per_device_train_batch_size",

  347.         "Value": "1"

  348.       },

  349.       {

  350.         "Label": "gradient_accumulation_steps",

  351.         "Value": "2"

  352.       },

  353.       {

  354.         "Label": "lr_scheduler_type",

  355.         "Value": "cosine"

  356.       },

  357.       {

  358.         "Label": "logging_steps",

  359.         "Value": "10"

  360.       },

  361.       {

  362.         "Label": "save_steps",

  363.         "Value": "500"

  364.       },

  365.       {

  366.         "Label": "warmup_ratio",

  367.         "Value": "0.1"

  368.       },

  369.       {

  370.         "Label": "val_size",

  371.         "Value": "0.1"

  372.       },

  373.       {

  374.         "Label": "per_device_eval_batch_size",

  375.         "Value": "1"

  376.       },

  377.       {

  378.         "Label": "eval_strategy",

  379.         "Value": "steps"

  380.       },

  381.       {

  382.         "Label": "eval_steps",

  383.         "Value": "500"

  384.       },

  385.       {

  386.         "Label": "model_name",

  387.         "Value": "Qwen2-0.5B"

  388.       },

  389.       {

  390.         "Label": "plot_loss",

  391.         "Value": "true"

  392.       },

  393.       {

  394.         "Label": "overwrite_output_dir",

  395.         "Value": "true"

  396.       },

  397.       {

  398.         "Label": "fp16",

  399.         "Value": "true"

  400.       },

  401.       {

  402.         "Label": "overwrite_cache",

  403.         "Value": "true"

  404.       }

  405.     ]

  406.   },

  407.   {

  408.     "Name": "Baichuan2-7b-base  G",

  409.     "Description": "",

  410.     "JobType": "TRAIN",

  411.     "Cluster": "C2Net",

  412.     "ComputeSource": "GPU",

  413.     "HasInternet": 2,

  414.     "AccCardsNum": 1,

  415.     "AccCardType": "V100",

  416.     "CpuCores": 8,

  417.     "MemGiB": 50,

  418.     "GPUMemGiB": 32,

  419.     "ShareMemGiB": 0,

  420.     "Image": {

  421.       "ImageID": "",

  422.       "ImageName": "",

  423.       "ImageUrl": "192.168.242.22:443/default-workspace/fccb038c23234b9e80105d4ccd152117/image:ChatGLM3-6B"

  424.     },

  425.     "PretrainModelList": [

  426.       {

  427.         "ID": "b5ec0a05-624c-44ea-aeda-0ab8a3bef65a",

  428.         "ModelName": "Baichuan2-7b-base"

  429.       }

  430.     ],

  431.     "DatasetList": [

  432.       {

  433.         "ID": "a924818a-b82c-4c48-bf78-dd517077244f",

  434.         "DatasetName": "huanhuan.zip"

  435.       },

  436.       {

  437.         "ID": "30e51e73-28b5-4040-a3ea-e63ca8332dba",

  438.         "DatasetName": "alpaca_zh_demo.zip"

  439.       }

  440.     ],

  441.     "BranchName": "master",

  442.     "BootFile": "finetune.py",

  443.     "Parameters": [

  444.       {

  445.         "Label": "stage",

  446.         "Value": "sft"

  447.       },

  448.       {

  449.         "Label": "preprocessing_numworkers",

  450.         "Value": "16"

  451.       },

  452.       {

  453.         "Label": "finetuning_type",

  454.         "Value": "lora"

  455.       },

  456.       {

  457.         "Label": "template",

  458.         "Value": "baichuan2"

  459.       },

  460.       {

  461.         "Label": "dataset",

  462.         "Value": "huanhuan,alpaca_zh_demo"

  463.       },

  464.       {

  465.         "Label": "cutoff_len",

  466.         "Value": "1024"

  467.       },

  468.       {

  469.         "Label": "learning_rate",

  470.         "Value": "1.0e-4"

  471.       },

  472.       {

  473.         "Label": "num_train_epochs",

  474.         "Value": "1"

  475.       },

  476.       {

  477.         "Label": "max_samples",

  478.         "Value": "1000"

  479.       },

  480.       {

  481.         "Label": "per_device_train_batch_size",

  482.         "Value": "1"

  483.       },

  484.       {

  485.         "Label": "gradient_accumulation_steps",

  486.         "Value": "2"

  487.       },

  488.       {

  489.         "Label": "lr_scheduler_type",

  490.         "Value": "cosine"

  491.       },

  492.       {

  493.         "Label": "logging_steps",

  494.         "Value": "10"

  495.       },

  496.       {

  497.         "Label": "save_steps",

  498.         "Value": "500"

  499.       },

  500.       {

  501.         "Label": "warmup_ratio",

  502.         "Value": "0.1"

  503.       },

  504.       {

  505.         "Label": "val_size",

  506.         "Value": "0.1"

  507.       },

  508.       {

  509.         "Label": "per_device_eval_batch_size",

  510.         "Value": "1"

  511.       },

  512.       {

  513.         "Label": "eval_strategy",

  514.         "Value": "steps"

  515.       },

  516.       {

  517.         "Label": "eval_steps",

  518.         "Value": "500"

  519.       },

  520.       {

  521.         "Label": "model_name",

  522.         "Value": "Baichuan2-7b-base"

  523.       },

  524.       {

  525.         "Label": "plot_loss",

  526.         "Value": "true"

  527.       },

  528.       {

  529.         "Label": "overwrite_output_dir",

  530.         "Value": "true"

  531.       },

  532.       {

  533.         "Label": "fp16",

  534.         "Value": "true"

  535.       },

  536.       {

  537.         "Label": "overwrite_cache",

  538.         "Value": "true"

  539.       }

  540.     ]

  541.   }

  542. ]
Please read the following content carefully:

Dear OpenI User

Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.

For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》

Agree and continue
Disagree, exit