Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / yangjianxin1/GPT2-chitchat issues and pull requests
#101 - 大佬好,我用两张3090去使用你所提供的那个100w的数据集训练,可是很快loss稳定在3.0就不降低了
Issue -
State: open - Opened by iniroc about 2 years ago
- 8 comments
#101 - 大佬好,我用两张3090去使用你所提供的那个100w的数据集训练,可是很快loss稳定在3.0就不降低了
Issue -
State: open - Opened by iniroc about 2 years ago
- 8 comments
#100 - 请问有办法集成到web应用里吗?
Issue -
State: open - Opened by drizzt00s over 2 years ago
- 5 comments
#100 - 请问有办法集成到web应用里吗?
Issue -
State: open - Opened by drizzt00s over 2 years ago
- 5 comments
#99 - 为什么我这训练过程中学习率是逐步上升的啊?
Issue -
State: open - Opened by Chenny0808 over 2 years ago
#98 - gpt模型的注意力头数n_head会影响模型在哪些方面的性能?怎样有效提升模型处理多轮对话的准确率?
Issue -
State: open - Opened by oerifjmerefver over 2 years ago
#98 - gpt模型的注意力头数n_head会影响模型在哪些方面的性能?怎样有效提升模型处理多轮对话的准确率?
Issue -
State: open - Opened by oerifjmerefver over 2 years ago
#97 - 请问预训练参数去哪里下载啊?
Issue -
State: open - Opened by 13227721183 over 2 years ago
- 5 comments
#97 - 请问预训练参数去哪里下载啊?
Issue -
State: open - Opened by 13227721183 over 2 years ago
- 5 comments
#96 - 有没有可能通过对通用对话的模型进行fine-tune给机器人赋予一个固定的人设和状态?并且对客观条件做出正确的反应?
Issue -
State: open - Opened by lewiswu1209 over 2 years ago
- 3 comments
#95 - 请问dataset里面没有传入labels,后面dataloader是从哪里得到的labels的呢
Issue -
State: open - Opened by pureblacker almost 3 years ago
- 4 comments
#94 - 更换更大的gpt2模型进行训练 如:gpt2_large
Issue -
State: open - Opened by htthYjh almost 3 years ago
- 1 comment
#93 - 模型加载报错
Issue -
State: open - Opened by UnstoppableCurry almost 3 years ago
- 6 comments
#93 - 模型加载报错
Issue -
State: open - Opened by UnstoppableCurry almost 3 years ago
- 6 comments
#91 - 模型训练不动了
Issue -
State: closed - Opened by G-Jarvey almost 3 years ago
- 5 comments
#90 - vocab
Issue -
State: open - Opened by pureblacker almost 3 years ago
- 2 comments
#89 - 自定义语料 报错ValueError: num_samples should be a positive integer value, but got num_samples=0
Issue -
State: closed - Opened by Dynamicboboo almost 3 years ago
- 2 comments
#89 - 自定义语料 报错ValueError: num_samples should be a positive integer value, but got num_samples=0
Issue -
State: closed - Opened by Dynamicboboo almost 3 years ago
- 2 comments
#88 - 多卡负载均衡
Pull Request -
State: open - Opened by chinoll about 3 years ago
#88 - 多卡负载均衡
Pull Request -
State: open - Opened by chinoll about 3 years ago
#87 - 为什么训练时候没有设定attention_mask这个参数
Issue -
State: closed - Opened by Choitsugun about 3 years ago
- 1 comment
#87 - 为什么训练时候没有设定attention_mask这个参数
Issue -
State: closed - Opened by Choitsugun about 3 years ago
- 1 comment
#86 - Update preprocess.py
Pull Request -
State: open - Opened by Aman-4-Real about 3 years ago
#86 - Update preprocess.py
Pull Request -
State: open - Opened by Aman-4-Real about 3 years ago
#85 - error---->train.py
Issue -
State: open - Opened by KangChou about 3 years ago
- 5 comments
#83 - 在tensorflow中使用50w chichat预模型时生成结果不佳
Issue -
State: open - Opened by qiuxia-alone over 3 years ago
#82 - 用训练好的模型生成数据,偶尔会报下边的错,位置不一样,跟您请教一下,怎么解决呢?网上查的方法都不行
Issue -
State: open - Opened by kavinwow100 over 3 years ago
#81 - 可以加个联系方式吗大佬?
Issue -
State: open - Opened by dahefanzhou over 3 years ago
#80 - 输入ctrl+z程序不结束,继续进行对话
Issue -
State: open - Opened by Chunfeng1994 over 3 years ago
- 2 comments
#80 - 输入ctrl+z程序不结束,继续进行对话
Issue -
State: open - Opened by Chunfeng1994 over 3 years ago
- 2 comments
#79 - 是本身就存在的train.py 里的bug吗? 写入正确的数据path,成功导入数据后仍然会报的valueerror问题 到底是哪一步错了?
Issue -
State: open - Opened by Saraooe over 3 years ago
- 12 comments
#79 - 是本身就存在的train.py 里的bug吗? 写入正确的数据path,成功导入数据后仍然会报的valueerror问题 到底是哪一步错了?
Issue -
State: open - Opened by Saraooe over 3 years ago
- 12 comments
#78 - 问一下有建议引用方式吗?
Issue -
State: open - Opened by Pwang001 over 3 years ago
#78 - 问一下有建议引用方式吗?
Issue -
State: open - Opened by Pwang001 over 3 years ago
#77 - 想问一下初始化的预训练模型的词表是2w+,但是现在的词表vocab.txt 只有1.3w,是根据自己的预料处理过吗?
Issue -
State: open - Opened by Cherryjingyao over 3 years ago
- 3 comments
#76 - 还得再请教下您temperature的用意,有点没太看懂
Issue -
State: open - Opened by kavinwow100 over 3 years ago
- 1 comment
#75 - 拆分训练集和验证集没有再用 sklearn的train_test_split是不是因为20%的验证集数据量太大了,所以用了个常量8000
Issue -
State: open - Opened by kavinwow100 over 3 years ago
- 3 comments
#74 - 获取data_list的时候要一次都读到内存里,如果数据量比较大,内存会爆掉。一般大家用什么办法解决?
Issue -
State: open - Opened by lonelydancer over 3 years ago
#74 - 获取data_list的时候要一次都读到内存里,如果数据量比较大,内存会爆掉。一般大家用什么办法解决?
Issue -
State: open - Opened by lonelydancer over 3 years ago
#73 - 训练一次100w的闲聊要多长时间?我这RX580 8G训练了9个小时了,还在跑,这个跑一次要多久呀?
Issue -
State: open - Opened by davewang over 3 years ago
- 13 comments
#73 - 训练一次100w的闲聊要多长时间?我这RX580 8G训练了9个小时了,还在跑,这个跑一次要多久呀?
Issue -
State: open - Opened by davewang over 3 years ago
- 13 comments
#72 - validate部分被注释掉了,这部分能正常使用不?
Issue -
State: closed - Opened by kavinwow100 over 3 years ago
- 3 comments
#72 - validate部分被注释掉了,这部分能正常使用不?
Issue -
State: closed - Opened by kavinwow100 over 3 years ago
- 3 comments
#71 - 真厉害阿
Issue -
State: open - Opened by jjljkjljk over 3 years ago
- 8 comments
#70 - 更新了更新了
Issue -
State: closed - Opened by yanzhuo77 over 3 years ago
- 1 comment
#69 - 期待更新一版,GPT2-Chinese4月22日刚更新
Issue -
State: closed - Opened by kavinwow100 almost 4 years ago
- 1 comment
#69 - 期待更新一版,GPT2-Chinese4月22日刚更新
Issue -
State: closed - Opened by kavinwow100 almost 4 years ago
- 1 comment
#68 - update transformers
Pull Request -
State: open - Opened by jqqqqqqqqqq almost 4 years ago
#68 - update transformers
Pull Request -
State: open - Opened by jqqqqqqqqqq almost 4 years ago
#67 - 好久没维护了 是不是不维护了
Issue -
State: closed - Opened by brook-w almost 4 years ago
- 1 comment
#67 - 好久没维护了 是不是不维护了
Issue -
State: closed - Opened by brook-w almost 4 years ago
- 1 comment
#66 - 提升对话的速度
Issue -
State: closed - Opened by brook-w almost 4 years ago
- 10 comments
#65 - 训练模型遇到cannot allocate memory
Issue -
State: open - Opened by chiangandy almost 4 years ago
- 4 comments
#65 - 训练模型遇到cannot allocate memory
Issue -
State: open - Opened by chiangandy almost 4 years ago
- 4 comments
#64 - 分享的模型的语料库里混进去了奇怪的东西...
Issue -
State: closed - Opened by zbLiuLiu about 4 years ago
#64 - 分享的模型的语料库里混进去了奇怪的东西...
Issue -
State: closed - Opened by zbLiuLiu about 4 years ago
#63 - speed up decoding
Pull Request -
State: closed - Opened by TianHongZXY about 4 years ago
#63 - speed up decoding
Pull Request -
State: closed - Opened by TianHongZXY about 4 years ago
#62 - ran out of memory
Issue -
State: closed - Opened by XixuHu about 4 years ago
- 2 comments
#62 - ran out of memory
Issue -
State: closed - Opened by XixuHu about 4 years ago
- 2 comments
#61 - BUG: no reply when run interact_mmi.py
Issue -
State: open - Opened by GaloisGroGauss about 4 years ago
- 3 comments
#61 - BUG: no reply when run interact_mmi.py
Issue -
State: open - Opened by GaloisGroGauss about 4 years ago
- 3 comments
#60 - train data的格式
Issue -
State: open - Opened by sharon880701 about 4 years ago
- 4 comments
#59 - 为什么训练dialog 模型的时候,self-attention没有进行上三角的attention mask呢?
Issue -
State: open - Opened by ChengchengDu about 4 years ago
- 4 comments
#59 - 为什么训练dialog 模型的时候,self-attention没有进行上三角的attention mask呢?
Issue -
State: open - Opened by ChengchengDu about 4 years ago
- 3 comments
#58 - gpt2中文预训练模型
Issue -
State: closed - Opened by ChengchengDu about 4 years ago
- 1 comment
#58 - gpt2中文预训练模型
Issue -
State: closed - Opened by ChengchengDu about 4 years ago
- 1 comment
#57 - python interact_mmi.py报错
Issue -
State: open - Opened by rorsarach about 4 years ago
- 4 comments
#57 - python interact_mmi.py报错
Issue -
State: open - Opened by rorsarach about 4 years ago
- 4 comments
#56 - 关于模型参数量
Issue -
State: closed - Opened by luofuli over 4 years ago
- 1 comment
#56 - 关于模型参数量
Issue -
State: closed - Opened by luofuli over 4 years ago
- 1 comment
#55 - 作者您好,请问这个模型可以把他训练成英文的模型吗?
Issue -
State: closed - Opened by Ake021 over 4 years ago
#55 - 作者您好,请问这个模型可以把他训练成英文的模型吗?
Issue -
State: closed - Opened by Ake021 over 4 years ago
#54 - tokenizer输入问题
Issue -
State: closed - Opened by xCerisier over 4 years ago
- 1 comment
#54 - tokenizer输入问题
Issue -
State: closed - Opened by xCerisier over 4 years ago
- 1 comment
#53 - 关于MMI输入语料进行逆序拼接
Issue -
State: open - Opened by WenTingTseng over 4 years ago
#52 - 读取train.txt的时候报错,说utf8无法解码
Issue -
State: open - Opened by MozarTuring over 4 years ago
- 4 comments
#51 - 能单独拎MMI的代码出来做response选择吗?为其它的模型结果做最佳的选择
Issue -
State: open - Opened by huangdacheng over 4 years ago
- 2 comments
#51 - 能单独拎MMI的代码出来做response选择吗?为其它的模型结果做最佳的选择
Issue -
State: open - Opened by huangdacheng over 4 years ago
- 2 comments
#50 - 根目录下缺少data文件夹,请问在哪里下载?
Issue -
State: closed - Opened by mali19064 over 4 years ago
- 3 comments
#50 - 根目录下缺少data文件夹,请问在哪里下载?
Issue -
State: closed - Opened by mali19064 over 4 years ago
- 3 comments
#49 - 如何提升Tokenize的速度
Issue -
State: closed - Opened by MrSeven77 over 4 years ago
- 1 comment
#49 - 如何提升Tokenize的速度
Issue -
State: closed - Opened by MrSeven77 over 4 years ago
- 1 comment
#48 - train.py中有bug
Issue -
State: open - Opened by shawroad over 4 years ago
- 6 comments
#48 - train.py中有bug
Issue -
State: open - Opened by shawroad over 4 years ago
- 6 comments
#47 - 模型训练所用时长及服务器配置
Issue -
State: open - Opened by jmfu95 over 4 years ago
- 1 comment
#47 - 模型训练所用时长及服务器配置
Issue -
State: open - Opened by jmfu95 over 4 years ago
- 1 comment
#46 - 请问,我想用博主训练好的模型生成更长的对话内容,是不是只需要把config里的n_ctx, n_positions改大,即可?
Issue -
State: closed - Opened by wulaoshi over 4 years ago
- 1 comment
#46 - 请问,我想用博主训练好的模型生成更长的对话内容,是不是只需要把config里的n_ctx, n_positions改大,即可?
Issue -
State: closed - Opened by wulaoshi over 4 years ago
- 1 comment
#45 - 老哥,训练的时候报错了
Issue -
State: open - Opened by 1615070057 over 4 years ago
- 3 comments
#44 - 老哥,BertTokenizer如何迁移到GPT2Tokenizer啊?
Issue -
State: open - Opened by aRookieMan over 4 years ago
#43 - 加载GPT2LMHeadModel编码错误啊
Issue -
State: closed - Opened by aRookieMan over 4 years ago
- 8 comments
#42 - 请问您提供的预训练model在50W对话上训练前,用其他数据集预训练过吗?
Issue -
State: closed - Opened by Zyh716 almost 5 years ago
- 1 comment
#42 - 请问您提供的预训练model在50W对话上训练前,用其他数据集预训练过吗?
Issue -
State: closed - Opened by Zyh716 almost 5 years ago
- 1 comment
#41 - 为啥我改了n_ctx这个参数没效果
Issue -
State: closed - Opened by kFoodie almost 5 years ago
- 1 comment
#40 - 请问gradient_accumulation这个参数是啥意思?
Issue -
State: closed - Opened by Years-Enron almost 5 years ago
- 4 comments
#39 - Update train.py
Pull Request -
State: open - Opened by Years-Enron almost 5 years ago
#38 - 关于attention mask
Issue -
State: closed - Opened by gmftbyGMFTBY almost 5 years ago
- 1 comment