Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

v1.0.0 alpha #154

Draft
wants to merge 242 commits into
base: master
Choose a base branch
from
Draft

v1.0.0 alpha #154

wants to merge 242 commits into from

Conversation

conorheins
Copy link
Collaborator

@conorheins conorheins commented Sep 24, 2024 •
edited
Loading

Draft PR to see how hard/easy merging in the v1.0.0 alpha changes would be at this stage

dimarkov and others added 30 commits June 10, 2024 11:47
@dimarkov
added learning for online variational filtering
12bdc1d
@dimarkov
optimized learning routines to minimize calls to tree_map
754a336
@dimarkov
fix tests and passing of the results of the learning
eda7e01
@dimarkov
fix update state transition
147a37c
move task.py to jax/envs/env.py
bdf1e14
@dimarkov
fix names
71a3cb2
@dimarkov
notebook for showcasing inference and learning using different methods
65b33c8
@conorheins
Merge pull request #133 from infer-actively/pymdp-move-PymdpEnv
a82d7a1 
move task.py to jax/envs/env.py
@OzanCatalVerses
add a pyproject toml
19172db
@conorheins
move task.py to jax/envs/env.py
948409a
@OzanCatalVerses
get a version that passes all tests
a1bc4b0
@conorheins
Merge pull request #134 from infer-actively/feature/pyproject-toml
be0a8f5 
Feature/pyproject toml
@conorheins
added an example of A tensors and B tensors where smoothing should di…
13ae015 
…sambiguate hidden states in the past
@conorheins
did first step of inference in the ovf example
ae528d9
@OzanCatalVerses @conorheins
add a pyproject toml
11a2f91
@OzanCatalVerses @conorheins
get a version that passes all tests
f9f9bab
@OzanCatalVerses @conorheins
add a pyproject toml
99f8253
@OzanCatalVerses @conorheins
get a version that passes all tests
a975e7d
initial graph worlds version
9db1556
@alec-tschantz
named distribution implementation
ad9c098 
First past of implementation for named distribution for easy manipulation and inspection of discrete conditional probability distributions.
@dimarkov
updated example
6d83206
@dimarkov
Merge branch 'full-ovf-with-example' into full-ovf-implementation
b1e50d3
@conorheins
removed oudated markdown cell at the end of the smoothing example not…
97b3716 
…ebook
update graph worlds env
980d86e
@OzanCatalVerses
add a first version of model compilation
0e0e1f5
@OzanCatalVerses
add missing s, add some basic tests
9f17816
@OzanCatalVerses
add unittest
630c59e
@OzanCatalVerses
rename compile to compile_model
3091da7
@alec-tschantz
added example notebook
465bfac
@OzanCatalVerses
add some documentation
4c07332
@Arun-Niranjan
Copy link
Contributor

@conorheins fantastic to see a potential v1 of pymdp!

at 97 files and 21k lines touched, I'd suggest breaking this up into sub PRs that get merged into main. Then you can pull main into this branch and then this PR will slowly get smaller over time.

My general process for this is:

  • sort out the tests to at least get a green build (to maximise signal-to-noise of tests failing)
  • anything that gets deleted because its not used right now (and then break into logical sub units, e.g. deleting a file that isn't used)
  • changes by folder/file, starting with easy ones that are largely self contained.
  • or changes by logical domain (e.g. jax scaffolding)

I'm not sure what verses have as PR guidelines, but I tend to find sub 200 lines (including tests) get the best quality reviews, but if that results in 100 PRs then relaxing this might be better 😅

@Arun-Niranjan
Copy link
Contributor

Also I'm happy to help with reviewing if you upgrade my contributor status to allow me to do so 😄

ran-wei-verses and others added 24 commits November 18, 2024 16:05
@ran-wei-verses
flatten pB action dims for parameter learning with multi actions
6a0c2b0
@conorheins
Merge pull request #161 from ran-wei-verses/v1.0.0_alpha
fa5213a 
flatten pB action dims for parameter learning with multi actions
@nikolamilovic1
Set minimum Python version to 3.10 for improved compatibility and fea…
b49587d 
…ture support
@nikolamilovic1
Add setup.cfg
66ae9f3
@nikolamilovic1
Modify setup.cfg to be equivalent to pyproject.toml.
af28a0f
@conorheins
Merge pull request #162 from nikolamilovic-ft/v1.0.0_alpha
3f06266 
Loosen Python Version Requirement to 3.10 & Add setup.cfg for Jetson Compatibility
@riddhipits
Add venv to .gitignore
38d1002
@riddhipits
getting gen_tmaze and tmaze working with refactored jax and added som…
0eb9b32 
…e comments
@riddhipits
getting tmaze running. need to add gif and agent gen model
ed6ffc9
@riddhipits
added an agent gen model but the agent does not behave correctly. nee…
390ae9d 
…d to fix.
@riddhipits
agent still not working
719dafd
@riddhipits
single agent, deterministic, no learning - works!
eedad26
@riddhipits
keeping rollout and adding comments. works.
8ccf136
@riddhipits
testing if A is being learned. not working yet.
3e1fc28
@riddhipits
A and B learning works separately and jointly
05c619b
@riddhipits
playing around with A and B structure learning in playground.ipynb. n…
3751559 
…ot working.
@riddhipits
clean up
3da6326
@riddhipits
removing unnecessary imports
b7a96a7
@conorheins
remove tmaze gif
d961c1d
@conorheins
tiny changes to generalized_tmaze_demo.ipynb
5ba3ef3
@conorheins
removed jax import from top of agent.py
f734553
@conorheins
undo whitespace additions to remove any file changes to agent.py
9ad97a7
@conorheins
Merge pull request #163 from riddhipits/tmaze_a_b_learning
0043c93 
Tmaze a b learning
@conorheins
ported changes from commit 69dee71 to v1.0.0_alpha branch
d79e697
@conorheins
Copy link
Collaborator Author

@conorheins fantastic to see a potential v1 of pymdp!

at 97 files and 21k lines touched, I'd suggest breaking this up into sub PRs that get merged into main. Then you can pull main into this branch and then this PR will slowly get smaller over time.

My general process for this is:

  • sort out the tests to at least get a green build (to maximise signal-to-noise of tests failing)
  • anything that gets deleted because its not used right now (and then break into logical sub units, e.g. deleting a file that isn't used)
  • changes by folder/file, starting with easy ones that are largely self contained.
  • or changes by logical domain (e.g. jax scaffolding)

I'm not sure what verses have as PR guidelines, but I tend to find sub 200 lines (including tests) get the best quality reviews, but if that results in 100 PRs then relaxing this might be better 😅

Hi @Arun-Niranjan -- sorry for the delayed reply to this. Totally agree, we should / will split this into smaller PRs with smaller bits of code, before we merge. We just used this draft PR as a hacky way to see if the CI tests would pass.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@tverbele tverbele Awaiting requested review from tverbele

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.