Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / LianjiaTech/BELLE issues and pull requests
#585 - expect 'max_length' in BELLE-2/Belle-whisper-large-v3-zh/config.json when converting to ggml
Issue -
State: open - Opened by XUJiahua 4 months ago
- 1 comment
#548 - synchronize before creating output_dir
Pull Request -
State: closed - Opened by zmzhang2000 11 months ago
#494 - ChatHome中,使用多任务指令结合预训练的方式相关疑问
Issue -
State: closed - Opened by sun1092469590 about 1 year ago
- 7 comments
#402 - 使用ZeRO stage 3比stage 2和stage 1更消耗GPU,很奇怪
Issue -
State: open - Opened by Flywolfs over 1 year ago
- 5 comments
#102 - ERROR: Could not consume arg: --model_config_file
Issue -
State: closed - Opened by bittteerr over 1 year ago
- 6 comments
#101 - 如何提高推理的性能
Issue -
State: closed - Opened by ianzangwill2 over 1 year ago
- 13 comments
#100 - fix add eval data
Pull Request -
State: closed - Opened by isunfloweryg over 1 year ago
#99 - add eval data
Pull Request -
State: closed - Opened by isunfloweryg over 1 year ago
#98 - 请问在两张3090上推理时显存溢出,请问可以使用nn.DataParallel分布式推理吗?具体怎么操作呢?谢谢!
Issue -
State: closed - Opened by lcneyc over 1 year ago
- 2 comments
#97 - 中文的推理能力
Issue -
State: closed - Opened by solojoe over 1 year ago
- 3 comments
#96 - Request for docker enviroment
Issue -
State: closed - Opened by JomeiLiu over 1 year ago
- 2 comments
#95 - 是否会支持embedding的输出?
Issue -
State: closed - Opened by BIGPPWONG over 1 year ago
#94 - 关于模型使用的例子 输出与展示不同
Issue -
State: closed - Opened by zhangzhen-research over 1 year ago
#93 - 1.5M 增加utf-8格式转换,减少在不同平台上生成的文件中出现中文编码,以及因为默认gbk导致的报错中断循环
Pull Request -
State: closed - Opened by BertieJim over 1 year ago
#92 - 请问quant_cuda这个library要怎样正常import呢?
Issue -
State: closed - Opened by another1s over 1 year ago
- 8 comments
#91 - RuntimeError: expected scalar type Half but found Float
Issue -
State: closed - Opened by XY2323819551 over 1 year ago
- 22 comments
#90 - 基于LLama的版本belle的协议还是apache 2.0吗?似乎与LLama的CC-BY-NC 4.0不兼容啊?
Issue -
State: closed - Opened by Trista0823 over 1 year ago
- 1 comment
#89 - 文檔typo
Pull Request -
State: closed - Opened by amy17519 over 1 year ago
#88 - Update README.md
Pull Request -
State: closed - Opened by xianghuisun over 1 year ago
#87 - Update README.md
Pull Request -
State: closed - Opened by xianghuisun over 1 year ago
#86 - Add train
Pull Request -
State: closed - Opened by xianghuisun over 1 year ago
#85 - Add train-v2
Pull Request -
State: closed - Opened by xianghuisun over 1 year ago
#84 - Add train
Pull Request -
State: closed - Opened by xianghuisun over 1 year ago
#83 - 哪位大佬有1M和0.5M的python合并脚本,弄了一晚上json,open实在合并不了
Issue -
State: closed - Opened by WUHU-G over 1 year ago
- 1 comment
#82 - 目前gpt4数学还比较差
Issue -
State: closed - Opened by zhangbo2008 over 1 year ago
- 3 comments
#81 - 为什么调整了训练的数据格式?
Issue -
State: closed - Opened by rayguo01 over 1 year ago
- 3 comments
#80 - 为什么BLOOMZ-7B1-mt模型只有14.1G,BELLE训练出来的模型却有28.3G?
Issue -
State: closed - Opened by liguodongiot over 1 year ago
- 5 comments
#79 - To fine-tune how much gpu is required for the BELLE-7B-2M model, I am now fine-tuning the error memory overflow reported on the a100
Issue -
State: closed - Opened by Amy234543 over 1 year ago
- 7 comments
#78 - 2M的数据是否后续会开源?
Issue -
State: closed - Opened by albertwy over 1 year ago
- 2 comments
#77 - 实验部分测试数据集
Issue -
State: closed - Opened by Awyshw over 1 year ago
- 1 comment
#76 - TypeError: vecquant2matmul(): incompatible function arguments.
Issue -
State: closed - Opened by johnny0213 over 1 year ago
- 1 comment
#75 - LLAMA微调是否对tokenizer词表有操作?
Issue -
State: closed - Opened by scarydemon2 over 1 year ago
- 1 comment
#74 - 请问数据答案是belle生成的还是chatgpt生成的
Issue -
State: closed - Opened by Chenzongchao over 1 year ago
- 1 comment
#73 - CUDA error: invalid device function
Issue -
State: open - Opened by linbang over 1 year ago
- 1 comment
#72 - 论文没有比较base model与微调后的模型的能力比较
Issue -
State: closed - Opened by EnHuiPug over 1 year ago
- 1 comment
#71 - bloom.py
Issue -
State: closed - Opened by LiuChen19960902 over 1 year ago
- 2 comments
#70 - 请问模型量化时的error是正常输出吗
Issue -
State: closed - Opened by liguodongiot over 1 year ago
- 1 comment
#69 - 试了下在1B的MT5-base上的效果
Issue -
State: closed - Opened by vxfla over 1 year ago
- 1 comment
#68 - 新instruct生成
Issue -
State: closed - Opened by hutbery over 1 year ago
- 4 comments
#67 - special token
Issue -
State: closed - Opened by guozhiyao over 1 year ago
- 1 comment
#66 - gpqt下用bloom_inference的报错
Issue -
State: closed - Opened by linbang over 1 year ago
- 1 comment
#65 - 如何批量生成prompt
Issue -
State: closed - Opened by wccccp over 1 year ago
- 1 comment
#64 - 小白求问 为啥量化版本没有可用LlamaForCausalLM直接调用的方式呀
Issue -
State: closed - Opened by akiori over 1 year ago
- 3 comments
#63 - 考虑提供13B甚至65B的中文训练版本么?
Issue -
State: closed - Opened by ldfandian over 1 year ago
- 4 comments
#62 - train和inference时的prompt
Issue -
State: closed - Opened by guozhiyao over 1 year ago
- 2 comments
#61 - 请问是这样运行app.py吗
Issue -
State: closed - Opened by ddzz0210 over 1 year ago
- 3 comments
#60 - 使用colab还是会这样 WARNING:root:OpenAIError: Invalid URL (POST /v1/chat/completions).
Issue -
State: closed - Opened by wccccp over 1 year ago
- 1 comment
#59 - BELLE-LLAMA-7B-2M能不能提供量化版本
Issue -
State: closed - Opened by whynothackme over 1 year ago
- 23 comments
#58 - 执行 python setup_cuda.py install 报错
Issue -
State: closed - Opened by liuyunrui123 over 1 year ago
- 4 comments
#57 - 多轮对话,和对话角色与情境设定可以吗?或有对应的演示案例不?
Issue -
State: closed - Opened by ZenXir over 1 year ago
- 1 comment
#56 - 以BELLE放出来的模型为基础模型,如BELLE-LLAMA-7B-2M为基础模型,用自定义的语料finetune 可以吗?
Issue -
State: closed - Opened by ZenXir over 1 year ago
- 2 comments
#55 - WARNING:root:OpenAIError: Invalid URL (POST /v1/chat/completions).
Issue -
State: closed - Opened by wccccp over 1 year ago
- 2 comments
#54 - context length 2049 我请求的3643token 请问在哪里设置
Issue -
State: closed - Opened by wccccp over 1 year ago
- 2 comments
#53 - bigscience/bloomz-7b1怎么如何下载?
Issue -
State: closed - Opened by alfgo over 1 year ago
- 1 comment
#52 - 代理设置无效 生成不了数据
Issue -
State: closed - Opened by wccccp over 1 year ago
- 1 comment
#51 - BELLE-LLAMA-7B-2M load ERROR
Issue -
State: closed - Opened by jkkl over 1 year ago
- 4 comments
#50 - 关于bloom_inference的使用问题
Issue -
State: closed - Opened by Tian14267 over 1 year ago
- 20 comments
#49 - 请问在本地下载好了模型,还需要openai_api_key吗?
Issue -
State: closed - Opened by wccccp over 1 year ago
- 1 comment
#48 - Exception: expected value at line 1 column 1
Issue -
State: closed - Opened by wccccp over 1 year ago
- 3 comments
#47 - Exception: expected value at line 1 column 1报错
Issue -
State: closed - Opened by wccccp over 1 year ago
- 1 comment
#46 - 请问0.5M和1M数据的prompt来源是什么?0.5M数据是做了什么样的数据质量控制吗?
Issue -
State: closed - Opened by hadoopmore over 1 year ago
- 1 comment
#45 - llama-7b的预训练模型不支持中文,这里是直接使用2M中文信息进行SFT吗
Issue -
State: closed - Opened by dihin11 over 1 year ago
- 5 comments
#44 - TypeError: 'type' object is not subscriptable
Issue -
State: closed - Opened by wccccp over 1 year ago
- 1 comment
#43 - 请问基于BLOOM和基于LLaMA的哪个效果好点?有做过测试吗?
Issue -
State: closed - Opened by ruidongtd over 1 year ago
- 6 comments
#42 - BelleGroup/BELLE-LLAMA-7B-2M模型是否还未发布
Issue -
State: closed - Opened by wqn1 over 1 year ago
- 8 comments
#41 - 执行报错!
Issue -
State: closed - Opened by StarRanger over 1 year ago
- 1 comment
#40 - 请问如何加载呢
Issue -
State: closed - Opened by caopeng000 over 1 year ago
- 2 comments
#39 - export OPENAI_API_KEY=YOUR_API_KEY
Issue -
State: closed - Opened by caopeng000 over 1 year ago
- 1 comment
#38 - Finetuning bloom with stanford_alpaca repo problem
Issue -
State: closed - Opened by raihan0824 over 1 year ago
- 2 comments
#36 - 语料生成相关
Issue -
State: closed - Opened by ZenXir over 1 year ago
- 1 comment
#35 - 想询问一下你们的训练环境是什么样的配置?
Issue -
State: closed - Opened by znsoftm over 1 year ago
- 1 comment
#34 - RM和PPO的部分
Issue -
State: closed - Opened by zhangsanfeng86 over 1 year ago
- 1 comment
#33 - 训练时的max_len
Issue -
State: closed - Opened by yangjianxin1 over 1 year ago
- 3 comments
#32 - bloom 二次训练
Issue -
State: closed - Opened by frankzhao112 over 1 year ago
- 1 comment
#31 - Update README.md
Pull Request -
State: closed - Opened by GoooIce over 1 year ago
- 1 comment
#30 - 请问理论上对于基于bloom的belle使用load_in_8bit会让推理速度变慢吗
Issue -
State: closed - Opened by neutron-1114 over 1 year ago
- 3 comments
#29 - 有对比过llama-7B和Bloom-7B在中文上的finetune后的效果吗
Issue -
State: closed - Opened by Morxrc over 1 year ago
- 3 comments
#28 - 加上多轮对话后belle-7b-2m模型会生成自问自答的内容。
Issue -
State: closed - Opened by jeave over 1 year ago
- 8 comments
#27 - 关于generate_instruction.py:如果prompt里要求输出10,输入例子3的话经常超出max_token,生成效率极低,想问下你们生产模式下是和现有代码相同设置生成数据的吗?
Issue -
State: closed - Opened by LeopoldACC over 1 year ago
- 2 comments
#26 - 如何fintune呀
Issue -
State: closed - Opened by wheniseeyou over 1 year ago
- 6 comments
#25 - Update zh_seed_tasks.json
Pull Request -
State: closed - Opened by eltociear over 1 year ago
- 1 comment
#23 - 多大的gpu 能跑起这个模型,4个12g的gpu能跑起这个模型么?
Issue -
State: closed - Opened by zhuchangjiang over 1 year ago
- 8 comments
#22 - 同级别参数量,模型大小差这么多?
Issue -
State: closed - Opened by FrankWhh over 1 year ago
- 2 comments
#21 - BELLE 7B-2M的安全性评测
Issue -
State: closed - Opened by TissueC over 1 year ago
- 3 comments
#20 - 谁有量化后的版本?
Issue -
State: closed - Opened by pangguoqing over 1 year ago
- 6 comments
#19 - 请问有尝试过bloom其他参数规模的模型进行finetune吗?效果如何?
Issue -
State: closed - Opened by ZhonghaoWang over 1 year ago
- 5 comments
#18 - 能用llama.cpp 4位量化 出来跑跑嘛
Issue -
State: closed - Opened by cgisky1980 over 1 year ago
- 2 comments
#17 - generate_instruction.py生成的数据集与Belle.train.json的格式不一致么
Issue -
State: closed - Opened by lihuicong over 1 year ago
- 3 comments
#16 - 运行generate_instruction.py会报错
Issue -
State: closed - Opened by lihuicong over 1 year ago
- 3 comments
#15 - 请问模型运行时的内存和显存需要多少?
Issue -
State: closed - Opened by MingJiaAn over 1 year ago
- 7 comments
#14 - 测试了一下,感觉模型的常识还不够
Issue -
State: closed - Opened by Degfy over 1 year ago
- 8 comments
#13 - 我看咱们使用的模型是bloomz 7b1-mt 没有使用斯坦福提到的llama 这是纯出于bloom 是多语言模型考虑?还是有做性能测试后的结论?
Issue -
State: closed - Opened by Syno8 over 1 year ago
- 3 comments
#12 - 175个中文种子任务 这数据在哪里?能让我们看下嘛?
Issue -
State: closed - Opened by joostshao over 1 year ago
- 1 comment
#11 - generate_instruction.py报错
Issue -
State: closed - Opened by MingJiaAn over 1 year ago
- 2 comments
#10 - 全局和Lora微调脚本参考
Issue -
State: closed - Opened by feizc over 1 year ago
- 1 comment
#9 - 利用bloomz.cpp转化模型的时候出错
Issue -
State: closed - Opened by dihin11 over 1 year ago
- 5 comments
#8 - 后续有开源全部数据集的计划吗
Issue -
State: closed - Opened by chinoll over 1 year ago
- 1 comment
#6 - FT 7B1 模型需要多少资源啊?
Issue -
State: closed - Opened by TTCoding over 1 year ago
- 1 comment
#5 - 可以用 alpaca.cpp 运行吗
Issue -
State: closed - Opened by linonetwo over 1 year ago
- 5 comments
Labels: enhancement