issues
search
ymcui
/
Chinese-LLaMA-Alpaca-2
中文LLaMA-2 & Alpaca-2大模型二期项目 + 64K超长上下文模型 (Chinese LLaMA-2 & Alpaca-2 LLMs with 64K long context models)
Apache License 2.0
7k
stars
570
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
YaRN support implementation
#505
iMountTai
closed
5 months ago
0
指令精调模型后生成的文件中只有adapter_model.safetensors,没有adapter_model.bin
#504
slamdunk77
closed
5 months ago
4
奖励模型如何进行推理
#503
wuhuanon
closed
5 months ago
3
使用flash attention会报错
#502
Go4miii
closed
5 months ago
2
有没有专门面向编程的版本
#501
qwas982
closed
5 months ago
2
关于chinese-alpaca-2-7b-64k模型在inference_hf.py推理部署中使用vllm报错的问题
#500
hoohooer
closed
5 months ago
6
考虑训练中文版的MOE模型吗?
#499
xujunrt
closed
5 months ago
4
Chinese-Alpaca-2的7B指令精调,用作分类,微调的结果accuracy为什么16bit比8bit只高0.001,另外13B模型不升反降?
#498
feifei05
closed
5 months ago
3
仓库里面包含中文词表扩充相关的代码吗
#497
weiiWill
closed
6 months ago
2
关于`在llama.cpp部署chinese-alpaca-2-7b-64k-hf,运行make LLAMA_CUBLAS=1发生的错误
#496
liaoran321
closed
5 months ago
4
请教一个问题,sft之后的pth如何转bin呢?可以给点建议吗
#495
smallBy
closed
5 months ago
5
ntk上下文扩展
#494
tkone2018
closed
5 months ago
5
是不是不支持在mac M系列芯片上进行预训练
#493
ban-shi-yi-sheng
closed
5 months ago
4
run_clm_sft_with_peft.py微调过程中计算acc,在评估指标计算函数compute_metrics中,要求拿到预测label和原始label,先对模型预测结果predictions的tokenid解码成中文再评估中文准确率acc,为什么predictions(预测结果)的前面系统提示词和instruction与inputs(原始所有输入)的tokenid不同,导致解码混乱,不能用"[/INST]"分割出label,这是什么原因,或者还有什么方法能拿到预测label?
#492
feifei05
closed
6 months ago
5
raylet内存警告,Object creation will fail if spilling is required.请问如何移除警告信息
#491
ByteCaprice
closed
5 months ago
4
Add AWQ models (pre-computed search results)
#490
ymcui
closed
6 months ago
0
在微调过程中,推理训练集和测试集,在trainer初始化传eval_dataset参数为字典,都包含两个数据集,但是key会作为前缀,然后报错,请问咋修改?
#489
feifei05
closed
6 months ago
5
模型能否上传到modelscope
#488
luoqishuai
closed
6 months ago
1
DeepSpeed提示Loss scale already at minimum,然后就自动退出。
#486
leonunix
closed
5 months ago
3
如何开启13B全量预训练?
#484
huwei02
closed
6 months ago
1
Add RLHF models
#483
GeekDream-x
closed
6 months ago
0
为什么同样的模型在命令行执行对话返回很快,但使用text-generation-webui就很慢
#482
skysea82410
closed
5 months ago
3
给模型批量喂数据
#481
dongziyu1016
closed
5 months ago
2
update inference scripts
#479
iMountTai
closed
6 months ago
0
Add 64k-context models
#478
ymcui
closed
6 months ago
0
为啥run_clm_sft_with_peft.py可以run不能debug,debug会报错AttributeError: 'PeftModelForCausalLM' object has no attribute 'save_checkpoint'?
#477
feifei05
closed
6 months ago
8
Model 在 RAG推理中不断重覆生成相同的tokens
#476
williamyu20
closed
6 months ago
3
请教一下 run_pt.sh 和 run_sft.sh 可以训练除了llma2以外的模型的吗 比如我下面贴的
#475
Mr1994
closed
6 months ago
1
位置插值训练数据相关咨询
#474
KyrieXu11
closed
6 months ago
5
如何通过metric_for_best_model 参数还是别的方式保存eval_accuracy最高的checkpoint ,并且能顺利执行完run_clm_sft_with_peft.py?
#473
feifei05
closed
6 months ago
5
为什么在指令微调(run_sft.sh)给定少量数据去训练了,合并模型之后还是没有效果?
#472
Kris-rod
closed
5 months ago
5
预训练数据格式
#469
AI-Study-Han
closed
6 months ago
2
Update requirements
#468
iMountTai
closed
6 months ago
0
Update requirements
#467
iMountTai
closed
6 months ago
0
二次预训练的blocksize和指令微调的blocksize应该一致吗?
#466
AIFFFENG
closed
6 months ago
2
模型可以上传阿里云盘吗?
#465
lyndonliu89757
closed
6 months ago
2
在ALpaca上进行指令精调,运行完成后没有生成lora的权重
#464
XuefangJia
closed
6 months ago
5
DeepSpeed zero3
#462
Abolfazl-kr
closed
7 months ago
0
合并 lora 模型与原版 LLaMA 模型时,合并后的文件夹里面只有部分文件,缺少config文件等
#461
Luka0770
closed
7 months ago
3
请问上下文16k模型的微调训练是否使用了8bit量化
#460
Junpliu
closed
6 months ago
3
在进行训练的时候 是否对txt内容共有格式要求
#458
Mr1994
closed
7 months ago
3
Update merge_lora script
#457
iMountTai
closed
7 months ago
0
NTK-ROPE
#456
IT-five
closed
6 months ago
3
长度外推的三种方式得到的answer竟一模一样?
#455
IT-five
closed
7 months ago
1
1.3B 模型没有包含 pytorch_model.bin.index.json ,如何在调优后进行合并?
#454
reterVision
closed
7 months ago
1
NTK长度外推,
#453
IT-five
closed
6 months ago
3
请问下,中文处理这块,你是先把中文翻译成英文,英文处理完成后,再转成中文的吗
#452
douguohai
closed
6 months ago
2
怎么启动chinese llama
#451
dongziyu1016
closed
6 months ago
4
macos 上训练 eval_loss 和 perplexity出现nan
#450
longkeyy
closed
6 months ago
4
对Alpaca2预训练后,回复内容很短
#449
lry0223
closed
6 months ago
4
Previous
Next