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

Updated Model Training #42

Open
wants to merge 42 commits into
base: main
Choose a base branch
from
Open

Updated Model Training #42

wants to merge 42 commits into from

Commits on Jan 31, 2024

  1. updated train.py to reflect changes in cli.py train & trainlaunch

    @wendywangwwt
    wendywangwwt committed Jan 31, 2024
    Configuration menu
    Copy the full SHA
    59d4140 View commit details
    Browse the repository at this point in the history

  2. updated trainlaunch for single node multi gpu

    @wendywangwwt
    wendywangwwt committed Jan 31, 2024
    Configuration menu
    Copy the full SHA
    2f808f7 View commit details
    Browse the repository at this point in the history

Commits on Feb 27, 2024

  1. added auto-determination of tile size for data augementation

    @wendywangwwt
    wendywangwwt committed Feb 27, 2024
    Configuration menu
    Copy the full SHA
    04db7ce View commit details
    Browse the repository at this point in the history

  2. allowed change of subfolder & batchsize when loading data (the former… 

    … for val during training and the latter for batchsize change during val/test)
    @wendywangwwt
    wendywangwwt committed Feb 27, 2024
    Configuration menu
    Copy the full SHA
    d0033f4 View commit details
    Browse the repository at this point in the history

  3. updated segmentation metrics calculation for deepliifext

    @wendywangwwt
    wendywangwwt committed Feb 27, 2024
    Configuration menu
    Copy the full SHA
    4aa5194 View commit details
    Browse the repository at this point in the history

Commits on Mar 5, 2024

  1. added resizeconv, currently inferred from model name / dir

    @wendywangwwt
    wendywangwwt committed Mar 5, 2024
    Configuration menu
    Copy the full SHA
    bcd4d92 View commit details
    Browse the repository at this point in the history

Commits on Mar 7, 2024

  1. fixed case when SDG has only 1 input modality

    @wendywangwwt
    wendywangwwt committed Mar 7, 2024
    Configuration menu
    Copy the full SHA
    c03245b View commit details
    Browse the repository at this point in the history

  2. added an experimental submodule to facilitate model evaluation and me… 

    …tric calculation
    @wendywangwwt
    wendywangwwt committed Mar 7, 2024
    Configuration menu
    Copy the full SHA
    a497cf4 View commit details
    Browse the repository at this point in the history

Commits on Mar 20, 2024

  1. added tv loss

    @wendywangwwt
    wendywangwwt committed Mar 20, 2024
    Configuration menu
    Copy the full SHA
    20ee9a0 View commit details
    Browse the repository at this point in the history

Commits on Apr 4, 2024

  1. disabled TV loss

    @wendywangwwt
    wendywangwwt committed Apr 4, 2024
    Configuration menu
    Copy the full SHA
    507c70c View commit details
    Browse the repository at this point in the history

  2. added experimental model: attention unet

    @wendywangwwt
    wendywangwwt committed Apr 4, 2024
    Configuration menu
    Copy the full SHA
    08a767c View commit details
    Browse the repository at this point in the history

Commits on May 15, 2024

  1. allowed to change net_gs

    @wendywangwwt
    wendywangwwt committed May 15, 2024
    Configuration menu
    Copy the full SHA
    aca2060 View commit details
    Browse the repository at this point in the history

  2. added argument return_seg_intermediate

    @wendywangwwt
    wendywangwwt committed May 15, 2024
    Configuration menu
    Copy the full SHA
    7a9c30c View commit details
    Browse the repository at this point in the history

Commits on Jun 26, 2024

  1. allowed cli.py inference and test.py to return raw segmentation predi… 

    …ctions
    @wendywangwwt
    wendywangwwt committed Jun 26, 2024
    Configuration menu
    Copy the full SHA
    9aebf56 View commit details
    Browse the repository at this point in the history

Commits on Jul 25, 2024

  1. allowed to specify modality-wise generator arch; added validation los… 

    …s calculation
    @wendywangwwt
    wendywangwwt committed Jul 25, 2024
    Configuration menu
    Copy the full SHA
    4cde804 View commit details
    Browse the repository at this point in the history

  2. added schedule-free adamw optimizer

    @wendywangwwt
    wendywangwwt committed Jul 25, 2024
    Configuration menu
    Copy the full SHA
    6d9ca56 View commit details
    Browse the repository at this point in the history

Commits on Jul 31, 2024

  1. fixed schedule-free optimizer setup (to not use scheduler wrapper); a… 

    …dd --optimizer in cli.py train
    @wendywangwwt
    wendywangwwt committed Jul 31, 2024
    Configuration menu
    Copy the full SHA
    952a660 View commit details
    Browse the repository at this point in the history

Commits on Aug 7, 2024

  1. added enable_batchnorm_tracking_stats to be used for validation loss … 

    …calculation during training (after which the stats needs to be re-enabled); added debug mode for cli.py train; added a flag in cli.py train to enable validation loss calculation; allowed to specify epoch in cli.py serialize
    @wendywangwwt
    wendywangwwt committed Aug 7, 2024
    Configuration menu
    Copy the full SHA
    351f542 View commit details
    Browse the repository at this point in the history

  2. minor fixes for options

    @wendywangwwt
    wendywangwwt committed Aug 7, 2024
    Configuration menu
    Copy the full SHA
    70a0456 View commit details
    Browse the repository at this point in the history

  3. forced cli.py serialize to load models to gpu0 if device=gpu to fix i… 

    …nconsistent inference results
    @wendywangwwt
    wendywangwwt committed Aug 7, 2024
    Configuration menu
    Copy the full SHA
    0051acf View commit details
    Browse the repository at this point in the history

  4. updated batchnorm disable and enable functions (missed in the previou… 

    …s commits); moved functions used only for tiff file to the bottom
    @wendywangwwt
    wendywangwwt committed Aug 7, 2024
    Configuration menu
    Copy the full SHA
    f9a8114 View commit details
    Browse the repository at this point in the history

  5. merged latest commits from main-update-training

    @wendywangwwt
    wendywangwwt committed Aug 7, 2024
    Configuration menu
    Copy the full SHA
    55729a7 View commit details
    Browse the repository at this point in the history

  6. removed changes related to stats calculation

    @wendywangwwt
    wendywangwwt committed Aug 7, 2024
    Configuration menu
    Copy the full SHA
    a2119cd View commit details
    Browse the repository at this point in the history

  7. removed schedule-free optimizers; allowed --optimizer to use any opti… 

    …mizer class supported in torch.optim
    @wendywangwwt
    wendywangwwt committed Aug 7, 2024
    Configuration menu
    Copy the full SHA
    e437283 View commit details
    Browse the repository at this point in the history

  8. resolved merge conflict

    @wendywangwwt
    wendywangwwt committed Aug 7, 2024
    Configuration menu
    Copy the full SHA
    bcf69bb View commit details
    Browse the repository at this point in the history

Commits on Aug 8, 2024

  1. added code lines missed during merge

    @wendywangwwt
    wendywangwwt committed Aug 8, 2024
    Configuration menu
    Copy the full SHA
    da6ebc8 View commit details
    Browse the repository at this point in the history

  2. changed netg and netgs configuration from list to tuple to avoid pars… 

    …ing issue; moved val data loading to with_val condition
    @wendywangwwt
    wendywangwwt committed Aug 8, 2024
    Configuration menu
    Copy the full SHA
    661bbb2 View commit details
    Browse the repository at this point in the history

Commits on Aug 13, 2024

  1. changed to use list in config to standardize

    @wendywangwwt
    wendywangwwt committed Aug 13, 2024
    Configuration menu
    Copy the full SHA
    cbf8c81 View commit details
    Browse the repository at this point in the history

  2. allowed SDG to use the same postprocess steps as DeepLIIFExt

    @wendywangwwt
    wendywangwwt committed Aug 13, 2024
    Configuration menu
    Copy the full SHA
    7f4a3cf View commit details
    Browse the repository at this point in the history

Commits on Aug 14, 2024

  1. completed validation implementation for deepliifext and sdg, with add… 

    …ed calculate_losses method; always use opt.lr in optimizers; set cell count metrics in validation only for deepliif
    @wendywangwwt
    wendywangwwt committed Aug 14, 2024
    Configuration menu
    Copy the full SHA
    fb75f20 View commit details
    Browse the repository at this point in the history

  2. added test cases for training

    @wendywangwwt
    wendywangwwt committed Aug 14, 2024
    Configuration menu
    Copy the full SHA
    8cc668b View commit details
    Browse the repository at this point in the history

  3. added fake validation set (same as training set :))

    @wendywangwwt
    wendywangwwt committed Aug 14, 2024
    Configuration menu
    Copy the full SHA
    5a277ed View commit details
    Browse the repository at this point in the history

  4. added method to calculate the cell count metrics for the ground truth… 

    … data
    @wendywangwwt
    wendywangwwt committed Aug 14, 2024
    Configuration menu
    Copy the full SHA
    edab58d View commit details
    Browse the repository at this point in the history

Commits on Aug 21, 2024

  1. minor fixes: disabled marker image in postprocess in cli.py train val… 

    …idation; removed duplicated function; fixed syntax error
    @wendywangwwt
    wendywangwwt committed Aug 21, 2024
    Configuration menu
    Copy the full SHA
    38a4f9e View commit details
    Browse the repository at this point in the history

  2. allowed attunet to use custom input output channel size for ext and sdg

    @wendywangwwt
    wendywangwwt committed Aug 21, 2024
    Configuration menu
    Copy the full SHA
    0af813d View commit details
    Browse the repository at this point in the history

  3. minor fixes: removed print logs

    @wendywangwwt
    wendywangwwt committed Aug 21, 2024
    Configuration menu
    Copy the full SHA
    b18e138 View commit details
    Browse the repository at this point in the history

  4. updated train.py and trainlaunch; minor fixes: number of netgs for de… 

    …epliif
    @wendywangwwt
    wendywangwwt committed Aug 21, 2024
    Configuration menu
    Copy the full SHA
    7c5cb30 View commit details
    Browse the repository at this point in the history

  5. added tests for trainlaunch; changed how the test commands are formed… 

    … in training tests to highlight test params
    @wendywangwwt
    wendywangwwt committed Aug 21, 2024
    Configuration menu
    Copy the full SHA
    0df24a8 View commit details
    Browse the repository at this point in the history

  6. cleaned up sdg model

    @wendywangwwt
    wendywangwwt committed Aug 21, 2024
    Configuration menu
    Copy the full SHA
    136540e View commit details
    Browse the repository at this point in the history

  7. conditional execution of metrics.json loading

    @wendywangwwt
    wendywangwwt committed Aug 21, 2024
    Configuration menu
    Copy the full SHA
    95698ae View commit details
    Browse the repository at this point in the history

Commits on Aug 22, 2024

  1. added changeds in the previous commit to train.py as well

    @wendywangwwt
    wendywangwwt committed Aug 22, 2024
    Configuration menu
    Copy the full SHA
    60d600e View commit details
    Browse the repository at this point in the history

Commits on Sep 3, 2024

  1. fixed match_suffix function for deepliifext/sdg

    @wendywangwwt
    wendywangwwt committed Sep 3, 2024
    Configuration menu
    Copy the full SHA
    0eba7b3 View commit details
    Browse the repository at this point in the history