GaoxyingM
commented
3 years ago Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks.
Closed gourango01 closed 3 years ago
GaoxyingM
commented
3 years ago Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks.
haitian-sun
commented
3 years ago Hi,
Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model.
Thanks, Haitian
On Apr 5, 2021, at 8:11 PM, GaoxyingM @.***> wrote:
Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA.
GaoxyingM
commented
3 years ago Thank you for your answer. Could you share the key code to reproduce this paper?Thanks.
------------------ 原始邮件 ------------------ 发件人: "OceanskySun/GraftNet" @.>; 发送时间: 2021年4月6日(星期二) 上午8:14 @.>; @.**@.>; 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
Hi,
Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model.
Thanks, Haitian
> On Apr 5, 2021, at 8:11 PM, GaoxyingM @.***> wrote: > > > Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA>. >
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
haitian-sun
commented
3 years ago The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function.
On Apr 5, 2021, at 8:18 PM, GaoxyingM @.***> wrote:
Thank you for your answer. Could you share the key code to reproduce this paper?Thanks.
------------------ 原始邮件 ------------------ 发件人: "OceanskySun/GraftNet" @.>; 发送时间: 2021年4月6日(星期二) 上午8:14 @.>; @.**@.>; 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
Hi,
Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model.
Thanks, Haitian
> On Apr 5, 2021, at 8:11 PM, GaoxyingM @.***> wrote: > > > Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA>. >
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA.
GaoxyingM
commented
3 years ago Thank you for your answer. I will try my best to reproduce it. If I encounter other problems, I may trouble you again. I hope you can help to answer when you have time, I will be very grateful. Could you share on the specific Google internal tools you used? When reproducing the code, what other tools can be used to replace these internal tools?
------------------ 原始邮件 ------------------ 发件人: "OceanskySun/GraftNet" @.>; 发送时间: 2021年4月6日(星期二) 上午8:20 @.>; @.**@.>; 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function.
> On Apr 5, 2021, at 8:18 PM, GaoxyingM @.> wrote: > > > Thank you for your answer. Could you share the key code to reproduce this paper?Thanks. > > > > > ------------------ 原始邮件 ------------------ > 发件人: "OceanskySun/GraftNet" @.>; > 发送时间: 2021年4月6日(星期二) 上午8:14 > @.>; > @*.**@*.>; > 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > Hi, > > Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model. > > Thanks, > Haitian > > > > On Apr 5, 2021, at 8:11 PM, GaoxyingM **@.***> wrote: > > > > > > Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315&gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA&gt;. > > > > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA>. >
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
haitian-sun
commented
3 years ago Nothing specific. It’s some internal tool for RPC calls and such. You don’t need it if you plan to implement it locally.
On Apr 5, 2021, at 8:32 PM, GaoxyingM @.***> wrote:
Thank you for your answer. I will try my best to reproduce it. If I encounter other problems, I may trouble you again. I hope you can help to answer when you have time, I will be very grateful. Could you share on the specific Google internal tools you used? When reproducing the code, what other tools can be used to replace these internal tools?
------------------ 原始邮件 ------------------ 发件人: "OceanskySun/GraftNet" @.>; 发送时间: 2021年4月6日(星期二) 上午8:20 @.>; @.**@.>; 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function.
> On Apr 5, 2021, at 8:18 PM, GaoxyingM @.> wrote: > > > Thank you for your answer. Could you share the key code to reproduce this paper?Thanks. > > > > > ------------------ 原始邮件 ------------------ > 发件人: "OceanskySun/GraftNet" @.>; > 发送时间: 2021年4月6日(星期二) 上午8:14 > @.>; > @*.**@*.>; > 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > Hi, > > Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model. > > Thanks, > Haitian > > > > On Apr 5, 2021, at 8:11 PM, GaoxyingM **@.***> wrote: > > > > > > Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315&gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA&gt;. > > > > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA>. >
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813740196, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADE5XLYLIVKI7FX2VQYC25LTHJJC7ANCNFSM4TOY62JA.
GaoxyingM
commented
3 years ago Thanks.Can you share other recurring code that removes calls to RPC internal functions? This is very helpful for me to reproduce the paper. Hope you can help me, thank you very much!
------------------ 原始邮件 ------------------ 发件人: "OceanskySun/GraftNet" @.>; 发送时间: 2021年4月6日(星期二) 上午8:36 @.>; @.**@.>; 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
Nothing specific. It’s some internal tool for RPC calls and such. You don’t need it if you plan to implement it locally.
> On Apr 5, 2021, at 8:32 PM, GaoxyingM @.> wrote: > > > Thank you for your answer. > I will try my best to reproduce it. If I encounter other problems, I may trouble you again. I hope you can help to answer when you have time, I will be very grateful. Could you share on the specific Google internal tools you used? When reproducing the code, what other tools can be used to replace these internal tools? > > > ------------------ 原始邮件 ------------------ > 发件人: "OceanskySun/GraftNet" @.>; > 发送时间: 2021年4月6日(星期二) 上午8:20 > @.>; > @*.**@*.>; > 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function. > > > On Apr 5, 2021, at 8:18 PM, GaoxyingM **@.> wrote: > > > > > > Thank you for your answer.&nbsp;Could you share the key code to reproduce this paper?Thanks. > > > > > > > > > > ------------------&nbsp;原始邮件&nbsp;------------------ > > 发件人: "OceanskySun/GraftNet" @.&gt;; > > 发送时间:&nbsp;2021年4月6日(星期二) 上午8:14 > > @.&gt;; > > @*.**@*.&gt;; > > 主题:&nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > > > > > > > Hi, > > > > Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model. > > > > Thanks, > > Haitian > > > > > > &gt; On Apr 5, 2021, at 8:11 PM, GaoxyingM **@.***&gt; wrote: > > &gt; > > &gt; > > &gt; Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > &gt; > > &gt; — > > &gt; You are receiving this because you were mentioned. > > &gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315&amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA&amp;gt;. > > &gt; > > > > > > — > > You are receiving this because you commented. > > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589&gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA&gt;. > > > > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813740196>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYLIVKI7FX2VQYC25LTHJJC7ANCNFSM4TOY62JA>. >
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
GaoxyingM
commented
3 years ago Hi！Could you share the original experimental data used in PullNet and the initialization code for the data?I would appreciate it if you could provide！Thanks.
GaoxyingM
commented
3 years ago @OceanskySun I liked your both (GRAFT-Net: https://www.aclweb.org/anthology/D18-1455.pdf, PullNet: https://www.aclweb.org/anthology/D19-1242.pdf ) papers on Question Answering over KB+Text task. My question is related to the baseline. and according to table 3 in PullNet paper, you have used KV-Mem as a baseline, and for this instead of using the original implementation KV-Mem, you have reimplemented it. If possible would you please share the implementation of KV-Mem*, which you have used in your paper?
Have you implemented the algorithm of PullNet?If possible could you share it?
GaoxyingM
commented
3 years ago Hi！If possible,could you share share the key code of the pullnet iteration process?Thank you very much.
------------------ Original ------------------ From: @.>; Send time: Tuesday, Apr 6, 2021 8:36 AM @.>; @.>; @.>; Subject: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
Nothing specific. It’s some internal tool for RPC calls and such. You don’t need it if you plan to implement it locally.
> On Apr 5, 2021, at 8:32 PM, GaoxyingM @.> wrote: > > > Thank you for your answer. > I will try my best to reproduce it. If I encounter other problems, I may trouble you again. I hope you can help to answer when you have time, I will be very grateful. Could you share on the specific Google internal tools you used? When reproducing the code, what other tools can be used to replace these internal tools? > > > ------------------ 原始邮件 ------------------ > 发件人: "OceanskySun/GraftNet" @.>; > 发送时间: 2021年4月6日(星期二) 上午8:20 > @.>; > @*.**@*.>; > 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function. > > > On Apr 5, 2021, at 8:18 PM, GaoxyingM **@.> wrote: > > > > > > Thank you for your answer.&nbsp;Could you share the key code to reproduce this paper?Thanks. > > > > > > > > > > ------------------&nbsp;原始邮件&nbsp;------------------ > > 发件人: "OceanskySun/GraftNet" @.&gt;; > > 发送时间:&nbsp;2021年4月6日(星期二) 上午8:14 > > @.&gt;; > > @*.**@*.&gt;; > > 主题:&nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > > > > > > > Hi, > > > > Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model. > > > > Thanks, > > Haitian > > > > > > &gt; On Apr 5, 2021, at 8:11 PM, GaoxyingM **@.***&gt; wrote: > > &gt; > > &gt; > > &gt; Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > &gt; > > &gt; — > > &gt; You are receiving this because you were mentioned. > > &gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315&amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA&amp;gt;. > > &gt; > > > > > > — > > You are receiving this because you commented. > > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589&gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA&gt;. > > > > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813740196>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYLIVKI7FX2VQYC25LTHJJC7ANCNFSM4TOY62JA>. >
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
haitian-sun
commented
3 years ago Thanks for your interest. It’s hard to open source the code because it’s built on some Google internal infrastructure. We apologize about that. Please let me know if you have any questions about paper. I am happy to answer the questions.
On Apr 7, 2021, at 7:56 AM, GaoxyingM @.***> wrote:
Hi！If possible,could you share share the key code of the pullnet iteration process?Thank you very much.
------------------ Original ------------------ From: @.>; Send time: Tuesday, Apr 6, 2021 8:36 AM @.>; @.>; @.>; Subject: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
Nothing specific. It’s some internal tool for RPC calls and such. You don’t need it if you plan to implement it locally.
> On Apr 5, 2021, at 8:32 PM, GaoxyingM @.> wrote: > > > Thank you for your answer. > I will try my best to reproduce it. If I encounter other problems, I may trouble you again. I hope you can help to answer when you have time, I will be very grateful. Could you share on the specific Google internal tools you used? When reproducing the code, what other tools can be used to replace these internal tools? > > > ------------------ 原始邮件 ------------------ > 发件人: "OceanskySun/GraftNet" @.>; > 发送时间: 2021年4月6日(星期二) 上午8:20 > @.>; > @*.**@*.>; > 主题: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function. > > > On Apr 5, 2021, at 8:18 PM, GaoxyingM **@.> wrote: > > > > > > Thank you for your answer.&nbsp;Could you share the key code to reproduce this paper?Thanks. > > > > > > > > > > ------------------&nbsp;原始邮件&nbsp;------------------ > > 发件人: "OceanskySun/GraftNet" @.&gt;; > > 发送时间:&nbsp;2021年4月6日(星期二) 上午8:14 > > @.&gt;; > > @*.**@*.&gt;; > > 主题:&nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > > > > > > > Hi, > > > > Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model. > > > > Thanks, > > Haitian > > > > > > &gt; On Apr 5, 2021, at 8:11 PM, GaoxyingM **@.***&gt; wrote: > > &gt; > > &gt; > > &gt; Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > &gt; > > &gt; — > > &gt; You are receiving this because you were mentioned. > > &gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315&amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA&amp;gt;. > > &gt; > > > > > > — > > You are receiving this because you commented. > > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589&gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA&gt;. > > > > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813740196>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYLIVKI7FX2VQYC25LTHJJC7ANCNFSM4TOY62JA>. >
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
GaoxyingM
commented
3 years ago Thank you for your answer.Could you provide the datasets used in PullNet？Such as MetaQA, WebQuestionsSP, Complex WebQuestions, and the corresponding knowledge graph and corpus.Excuse me, section 3.2.4 of the paper said in The Update Operation: other new edges may connect newly-added nodes with nodes that already exist in the previous subgraph. Could you tell us more about its implementation process?Thank you very much.
------------------ Original ------------------ From: @.>; Send time: Thursday, Apr 8, 2021 5:09 AM @.>; @.>; @.>; Subject: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
Thanks for your interest. It’s hard to open source the code because it’s built on some Google internal infrastructure. We apologize about that. Please let me know if you have any questions about paper. I am happy to answer the questions.
> On Apr 7, 2021, at 7:56 AM, GaoxyingM @.> wrote: > > > Hi！If possible,could you share share the key code of the pullnet iteration process?Thank you very much. > > > > > ------------------ Original ------------------ > From: @.>; > Send time: Tuesday, Apr 6, 2021 8:36 AM > @.>; > @.>; @.>; > Subject: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > Nothing specific. It’s some internal tool for RPC calls and such. You don’t need it if you plan to implement it locally. > > > On Apr 5, 2021, at 8:32 PM, GaoxyingM **@.> wrote: > > > > > > Thank you for your answer. > > I will try my best to reproduce it. If I encounter other problems, I may trouble you again. I hope you can help to answer when you have time, I will be very grateful. Could you share on the specific Google internal tools you used? When reproducing the code, what other tools can be used to replace these internal tools? > > > > > > ------------------&nbsp;原始邮件&nbsp;------------------ > > 发件人: "OceanskySun/GraftNet" @.&gt;; > > 发送时间:&nbsp;2021年4月6日(星期二) 上午8:20 > > @.&gt;; > > @*.**@*.&gt;; > > 主题:&nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > > > > > > > The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function. > > > > &gt; On Apr 5, 2021, at 8:18 PM, GaoxyingM **@.&gt; wrote: > > &gt; > > &gt; > > &gt; Thank you for your answer.&amp;nbsp;Could you share the key code to reproduce this paper?Thanks. > > &gt; > > &gt; > > &gt; > > &gt; > > &gt; ------------------&amp;nbsp;原始邮件&amp;nbsp;------------------ > > &gt; 发件人: "OceanskySun/GraftNet" @.&amp;gt;; > > &gt; 发送时间:&amp;nbsp;2021年4月6日(星期二) 上午8:14 > > &gt; @.&amp;gt;; > > &gt; @*.**@*.&amp;gt;; > > &gt; 主题:&amp;nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > &gt; > > &gt; > > &gt; > > &gt; > > &gt; > > &gt; Hi, > > &gt; > > &gt; Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model. > > &gt; > > &gt; Thanks, > > &gt; Haitian > > &gt; > > &gt; > > &gt; &amp;gt; On Apr 5, 2021, at 8:11 PM, GaoxyingM **@.***&amp;gt; wrote: > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > &gt; &amp;gt; > > &gt; &amp;gt; — > > &gt; &amp;gt; You are receiving this because you were mentioned. > > &gt; &amp;gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315&amp;amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA&amp;amp;gt;. > > &gt; &amp;gt; > > &gt; > > &gt; > > &gt; — > > &gt; You are receiving this because you commented. > > &gt; Reply to this email directly, view it on GitHub, or unsubscribe. > > &gt; — > > &gt; You are receiving this because you were mentioned. > > &gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589&amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA&amp;gt;. > > &gt; > > > > > > — > > You are receiving this because you commented. > > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813740196&gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYLIVKI7FX2VQYC25LTHJJC7ANCNFSM4TOY62JA&gt;. > > > > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub, or unsubscribe.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
haitian-sun
commented
3 years ago You may find this script useful for generating data for MetaQA.
https://github.com/google-research/language/blob/master/language/emql/preprocess/metaqa_preprocess.py https://github.com/google-research/language/blob/master/language/emql/preprocess/metaqa_preprocess.py
For the update operation, you can basically iterate through each node in the newly added nodes, and check if there exists a fact in the knowledge base or text piece in your corpus that can connect the new nodes (entities) and existing ones in the previous subgraph.
On Apr 8, 2021, at 9:49 PM, GaoxyingM @.***> wrote:
Thank you for your answer.Could you provide the datasets used in PullNet？Such as MetaQA, WebQuestionsSP, Complex WebQuestions, and the corresponding knowledge graph and corpus.Excuse me, section 3.2.4 of the paper said in The Update Operation: other new edges may connect newly-added nodes with nodes that already exist in the previous subgraph. Could you tell us more about its implementation process?Thank you very much.
------------------ Original ------------------ From: @.>; Send time: Thursday, Apr 8, 2021 5:09 AM @.>; @.>; @.>; Subject: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
Thanks for your interest. It’s hard to open source the code because it’s built on some Google internal infrastructure. We apologize about that. Please let me know if you have any questions about paper. I am happy to answer the questions.
> On Apr 7, 2021, at 7:56 AM, GaoxyingM @.> wrote: > > > Hi！If possible,could you share share the key code of the pullnet iteration process?Thank you very much. > > > > > ------------------ Original ------------------ > From: @.>; > Send time: Tuesday, Apr 6, 2021 8:36 AM > @.>; > @.>; @.>; > Subject: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > Nothing specific. It’s some internal tool for RPC calls and such. You don’t need it if you plan to implement it locally. > > > On Apr 5, 2021, at 8:32 PM, GaoxyingM **@.> wrote: > > > > > > Thank you for your answer. > > I will try my best to reproduce it. If I encounter other problems, I may trouble you again. I hope you can help to answer when you have time, I will be very grateful. Could you share on the specific Google internal tools you used? When reproducing the code, what other tools can be used to replace these internal tools? > > > > > > ------------------&nbsp;原始邮件&nbsp;------------------ > > 发件人: "OceanskySun/GraftNet" @.&gt;; > > 发送时间:&nbsp;2021年4月6日(星期二) 上午8:20 > > @.&gt;; > > @*.**@*.&gt;; > > 主题:&nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > > > > > > > The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function. > > > > &gt; On Apr 5, 2021, at 8:18 PM, GaoxyingM **@.&gt; wrote: > > &gt; > > &gt; > > &gt; Thank you for your answer.&amp;nbsp;Could you share the key code to reproduce this paper?Thanks. > > &gt; > > &gt; > > &gt; > > &gt; > > &gt; ------------------&amp;nbsp;原始邮件&amp;nbsp;------------------ > > &gt; 发件人: "OceanskySun/GraftNet" @.&amp;gt;; > > &gt; 发送时间:&amp;nbsp;2021年4月6日(星期二) 上午8:14 > > &gt; @.&amp;gt;; > > &gt; @*.**@*.&amp;gt;; > > &gt; 主题:&amp;nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > &gt; > > &gt; > > &gt; > > &gt; > > &gt; > > &gt; Hi, > > &gt; > > &gt; Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model. > > &gt; > > &gt; Thanks, > > &gt; Haitian > > &gt; > > &gt; > > &gt; &amp;gt; On Apr 5, 2021, at 8:11 PM, GaoxyingM **@.***&amp;gt; wrote: > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > &gt; &amp;gt; > > &gt; &amp;gt; — > > &gt; &amp;gt; You are receiving this because you were mentioned. > > &gt; &amp;gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315&amp;amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA&amp;amp;gt;. > > &gt; &amp;gt; > > &gt; > > &gt; > > &gt; — > > &gt; You are receiving this because you commented. > > &gt; Reply to this email directly, view it on GitHub, or unsubscribe. > > &gt; — > > &gt; You are receiving this because you were mentioned. > > &gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589&amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA&amp;gt;. > > &gt; > > > > > > — > > You are receiving this because you commented. > > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813740196&gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYLIVKI7FX2VQYC25LTHJJC7ANCNFSM4TOY62JA&gt;. > > > > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub, or unsubscribe.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-816344255, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADE5XL5GXX5Z6Q43K442TNDTHZMMHANCNFSM4TOY62JA.
GaoxyingM
commented
3 years ago Thank you for your answer, I will try my best to reproduce it. ------------------ Original ------------------ From: @.>; Send time: Saturday, Apr 10, 2021 3:09 AM @.>; @.>; @.>; Subject: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem*) (#17)
You may find this script useful for generating data for MetaQA.
https://github.com/google-research/language/blob/master/language/emql/preprocess/metaqa_preprocess.py <https://github.com/google-research/language/blob/master/language/emql/preprocess/metaqa_preprocess.py>
For the update operation, you can basically iterate through each node in the newly added nodes, and check if there exists a fact in the knowledge base or text piece in your corpus that can connect the new nodes (entities) and existing ones in the previous subgraph.
> On Apr 8, 2021, at 9:49 PM, GaoxyingM @.> wrote: > > > Thank you for your answer.Could you provide the datasets used in PullNet？Such as MetaQA, WebQuestionsSP, Complex WebQuestions, and the corresponding knowledge graph and corpus.Excuse me, section 3.2.4 of the paper said in The Update Operation: other new edges may connect newly-added nodes with nodes that already exist in the previous subgraph. Could you tell us more about its implementation process?Thank you very much. > > > > > ------------------ Original ------------------ > From: @.>; > Send time: Thursday, Apr 8, 2021 5:09 AM > @.>; > @.>; @.>; > Subject: Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > Thanks for your interest. It’s hard to open source the code because it’s built on some Google internal infrastructure. We apologize about that. Please let me know if you have any questions about paper. I am happy to answer the questions. > > > On Apr 7, 2021, at 7:56 AM, GaoxyingM **@.> wrote: > > > > > > Hi！If possible,could you share share the key code of the pullnet iteration process?Thank you very much. > > > > > > > > > > ------------------&nbsp;Original&nbsp;------------------ > > From: @.&gt;; > > Send time:&nbsp;Tuesday, Apr 6, 2021 8:36 AM > > @.&gt;; > > @.&gt;; @.&gt;; > > Subject: &nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > > > > > > > > > > > Nothing specific. It’s some internal tool for RPC calls and such. You don’t need it if you plan to implement it locally. > > > > &gt; On Apr 5, 2021, at 8:32 PM, GaoxyingM **@.&gt; wrote: > > &gt; > > &gt; > > &gt; Thank you for your answer. > > &gt; I will try my best to reproduce it. If I encounter other problems, I may trouble you again. I hope you can help to answer when you have time, I will be very grateful. Could you share on the specific Google internal tools you used? When reproducing the code, what other tools can be used to replace these internal tools? > > &gt; > > &gt; > > &gt; ------------------&amp;nbsp;原始邮件&amp;nbsp;------------------ > > &gt; 发件人: "OceanskySun/GraftNet" @.&amp;gt;; > > &gt; 发送时间:&amp;nbsp;2021年4月6日(星期二) 上午8:20 > > &gt; @.&amp;gt;; > > &gt; @*.**@*.&amp;gt;; > > &gt; 主题:&amp;nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > &gt; > > &gt; > > &gt; > > &gt; > > &gt; > > &gt; The classifier is the exact GraftNet model. The classifier is implemented with pylucene using the default scoring function. > > &gt; > > &gt; &amp;gt; On Apr 5, 2021, at 8:18 PM, GaoxyingM **@.&amp;gt; wrote: > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; Thank you for your answer.&amp;amp;nbsp;Could you share the key code to reproduce this paper?Thanks. > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; ------------------&amp;amp;nbsp;原始邮件&amp;amp;nbsp;------------------ > > &gt; &amp;gt; 发件人: "OceanskySun/GraftNet" @.&amp;amp;gt;; > > &gt; &amp;gt; 发送时间:&amp;amp;nbsp;2021年4月6日(星期二) 上午8:14 > > &gt; &amp;gt; @.&amp;amp;gt;; > > &gt; &amp;gt; @*.**@*.&amp;amp;gt;; > > &gt; &amp;gt; 主题:&amp;amp;nbsp;Re: [OceanskySun/GraftNet] Code for baseline (KV-Mem) (#17) > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; Hi, > > &gt; &amp;gt; > > &gt; &amp;gt; Thanks for your email. Unfortunately the implementation is tied with Google infrastructure and is hard to open-source. Please let me know if you have any questions. I can help you as much as possible to reproduce the model. > > &gt; &amp;gt; > > &gt; &amp;gt; Thanks, > > &gt; &amp;gt; Haitian > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; &amp;amp;gt; On Apr 5, 2021, at 8:11 PM, GaoxyingM **@.***&amp;amp;gt; wrote: > > &gt; &amp;gt; &amp;amp;gt; > > &gt; &amp;gt; &amp;amp;gt; > > &gt; &amp;gt; &amp;amp;gt; Hi，do you have the code to implement PullNet?If possible would you please share it？Thanks. > > &gt; &amp;gt; &amp;amp;gt; > > &gt; &amp;gt; &amp;amp;gt; — > > &gt; &amp;gt; &amp;amp;gt; You are receiving this because you were mentioned. > > &gt; &amp;gt; &amp;amp;gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813733315&amp;amp;amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYG2LXNJ3FSIYWG4OTTHJGR7ANCNFSM4TOY62JA&amp;amp;amp;gt;. > > &gt; &amp;gt; &amp;amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; > > &gt; &amp;gt; — > > &gt; &amp;gt; You are receiving this because you commented. > > &gt; &amp;gt; Reply to this email directly, view it on GitHub, or unsubscribe. > > &gt; &amp;gt; — > > &gt; &amp;gt; You are receiving this because you were mentioned. > > &gt; &amp;gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813735589&amp;amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYITTRNU6I6CDKZQ6TTHJHMFANCNFSM4TOY62JA&amp;amp;gt;. > > &gt; &amp;gt; > > &gt; > > &gt; > > &gt; — > > &gt; You are receiving this because you commented. > > &gt; Reply to this email directly, view it on GitHub, or unsubscribe. > > &gt; — > > &gt; You are receiving this because you were mentioned. > > &gt; Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-813740196&amp;gt;, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XLYLIVKI7FX2VQYC25LTHJJC7ANCNFSM4TOY62JA&amp;gt;. > > &gt; > > > > > > — > > You are receiving this because you commented. > > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub <https://github.com/OceanskySun/GraftNet/issues/17#issuecomment-816344255>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADE5XL5GXX5Z6Q43K442TNDTHZMMHANCNFSM4TOY62JA>. >
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
@OceanskySun I liked your both (GRAFT-Net: https://www.aclweb.org/anthology/D18-1455.pdf, PullNet: https://www.aclweb.org/anthology/D19-1242.pdf ) papers on Question Answering over KB+Text task. My question is related to the baseline. and according to table 3 in PullNet paper, you have used KV-Mem as a baseline, and for this instead of using the original implementation KV-Mem, you have reimplemented it. If possible would you please share the implementation of KV-Mem*, which you have used in your paper?