codefuse-ai/MFTCoder

Issues

• loss计算那里 RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn

  #22 opened a year ago by hhy150
  3

• 实验 MFTCoder 的效果总是不尽人意

  #64 opened 5 months ago by Chaochao2020
  2

• model type

  #65 opened 4 months ago by XiaoMaGe-hero
  0

• mftcoder 新版 Permission denied: '/home/admin' BUG

  #60 opened 5 months ago by Chaochao2020
  5

• 数据集loss 下降不均衡如何处理

  #55 opened 5 months ago by huangmenglong
  1

• RuntimeError: CUDA error: invalid device ordinal

  #57 opened 6 months ago by lwh8915
  1

• mftcoder使用humaneval评估

  #59 opened 6 months ago by lwh8915
  0

• 请问多机训练需要怎么修改？

  #43 opened 6 months ago by jy00161yang
  1

• 在codellama上微调的性能没有提升

  #18 opened a year ago by HPRCEST
  2

• 请教4int的gptq模型能不能进行lora微调

  #34 opened 7 months ago by wengyuan722
  4

• 任务的类型也是用gpt来生成的吗？

  #48 opened 7 months ago by shatealaboxiaowang
  1

• MFTCoder论文中训练数据集

  #51 opened 7 months ago by superqing001
  2

• convergence curves

  #52 opened 7 months ago by twelveand0
  0

• How can i do continue pretraining?

  #47 opened 8 months ago by hwaking
  0

• qlora微调合并权重时出错

  #42 opened 9 months ago by fangzexian
  4

• 模型微调完，合并时报错 NotImplementedError: Cannot copy out of meta tensor; no data!

  #45 opened 10 months ago by xxyp
  2

• 代码中对于3.5 Multitask Fine-Tuning with Balanced Losses的具体实现的位置（只找到了第一个loss的实现）

  #25 opened 10 months ago by YanqiDai
  2

• 请问要支持chatglm3-6b-base的话需要哪些更改

  #35 opened 10 months ago by kevindany
  2

• 请问下是否支持Wandb或者Tensorboard

  #36 opened 10 months ago by pydaxing
  1

• no 7B model size?

  #30 opened 10 months ago by yiyepiaoling0715
  2

• Inquiry about weighted_loss_mode

  #15 opened 10 months ago by tszdanger
  1

• 请问FSDP的训练API啥时候会开源出来

  #14 opened 10 months ago by peiji1981
  1

• ValueError: Asking to pad but the tokenizer does not have a padding token. Please select a token to use as `pad_token` `(tokenizer.pad_token = tokenizer.eos_token e.g.)` or add a new pad token via `tokenizer.add_special_tokens({'pad_token': '[PAD]'})`.

  #32 opened a year ago by sxsxsx
  9

• NotImplementedError: Cannot copy out of meta tensor; no data!

  #24 opened a year ago by zzb2019053515
  1

• nccl 报错了

  #31 opened a year ago by belle9217
  3

• 如何构建codefuse-llamacode的提问和终止符

  #23 opened a year ago by wengyuan722
  29

• Something wrong when run 'bash run_bash.sh'

  #27 opened a year ago by MaoYouSi
  1

• 请问，对模型进行多任务微调该怎么设计jsonl数据集？

  #16 opened a year ago by a793181018
  5

• 麻烦我想问下一个可行性问题，对CodeFuse-CodeGeeX2-6B进行微调时是否可以使用peft的方式中chatglm2 config进行微调？万分感谢🙏

  #13 opened a year ago by whyPeanutbutter
  1

• 模型是否支持商用

  #33 opened a year ago by zhangyukun230
  5

• safetensors_rust.SafetensorError: Error while deserializing header: HeaderTooLarge

  #20 opened a year ago by zzb2019053515
  3

• little bug fix meet

  #21 opened a year ago by elcky
  2

• 单卡v1000，微调报错

  #28 opened a year ago by sxsxsx
  2

• json.decoder.JSONDecodeError: Expecting value: line 1 column 2 (char 1)

  #29 opened a year ago by sxsxsx
  5

• 数据问题ValueError: data format not supported, please use prompt/answer, or chatML or pretrain text

  #26 opened a year ago by mst272
  0

• 模型训练没有进度条

  #19 opened a year ago by liujingqiao
  0

• data.helper 无法加载？

  #11 opened a year ago by liudonglei
  4

• MFTCoder微调codefuse34b模型后，发现模型代码补全这块的回复能力就没了，求解决方案

  #7 opened a year ago by yangyubin1
  3

• readme.txt指出，训练数据为jsonl格式，参考项目中的xxx.jsonl文件。未搜到对应的参考jsonl文件，能否麻烦给出一个示例？谢谢🙏

  #12 opened a year ago by whyPeanutbutter
  2

• 基于chatgpt生成的高质量python练习题数据是如何获取呀

  #6 opened a year ago by 18liumin
  1

• 使用lora + zero3微调CodeFuse-CodeLlama-34B后，合并模型失败

  #8 opened a year ago by 3m123
  3

• HumanEval测试的Pass@1不高

  #3 opened a year ago by wangzhao88
  2

• about focal loss mentioned in the paper

  #10 opened a year ago by iDonal
  1

• 训练数据包含中文数据吗

  #1 opened a year ago by smashfan
  1

• 能否写一个完整的微调例子？

  #9 opened a year ago by liudonglei
  1

• 国内下载方式

  #2 opened a year ago by wuyihz
  1