Skip to content
This repository has been archived by the owner on Apr 8, 2019. It is now read-only.

[doc] Report #20

Closed
7 of 15 tasks
at15 opened this issue Dec 12, 2017 · 1 comment
Closed
7 of 15 tasks

[doc] Report #20

at15 opened this issue Dec 12, 2017 · 1 comment
Assignees
Milestone

Comments

@at15
Copy link
Owner

at15 commented Dec 12, 2017

Project presentations during finals week: 12-3pm, Th 12-14

For final report, follow the NIPS paper limits if you can, ie <= 8 pages excluding refs

TODO

  • figure out other "authors" email address .... 5ea6e07
  • abstract
  • background bb09464
    • chatbot
    • seq2seq (more in the model part)
  • Model 82f97e5
    • LSTM seq2seq, seq2one (explain previous assignment)
    • seq2seq can be used on many things
  • dataset 352c7ed
    • abstraction of loading different dataset
    • caveat of cornell dialog corpus
  • frameworks
    • machine learning frameworks
    • seq2seq frameworks
  • evaluation

Reference

Students reports

@at15 at15 added this to the 0.0.1 milestone Dec 12, 2017
@at15 at15 self-assigned this Dec 12, 2017
@at15
Copy link
Owner Author

at15 commented Dec 12, 2017

btw: found other students course project https://github.com/ZaydH/trump_char_rnn (didn't get the repo name, but came across author's name on piazza)

at15 added a commit that referenced this issue Dec 12, 2017
- replace `natbib` w/ biber (which is used by all my previous reports
(based on VLDB's template ...)
- most content in the template are not removed, yet
at15 added a commit that referenced this issue Dec 12, 2017
- built an open domain generative chat bot, most difficult in theory,
easiest in practice (garbage in, garbage out, just speak non sense)
- the data is not good enough (although it is widely used ...)
- we didn't built our own based on low level tf API #23 (be honest)
at15 added a commit that referenced this issue Dec 13, 2017
- list variants of seq2seq
- a graph for seq2seq in QA
at15 added a commit that referenced this issue Dec 13, 2017
- #20
- older version keeps idk in text, newer version already removed it,
but the model is saved so we don't need to train it for 1h ...
at15 added a commit that referenced this issue Dec 13, 2017
@at15 at15 closed this as completed in 58fd3e0 Dec 15, 2017
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant