issues
search
abuzarmahmood
/
pytau
Library to streamline batch-fitting of Bayesian changepoint models to neural population spiking data
https://abuzarmahmood.github.io/pytau/
MIT License
1
stars
0
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Error when saving spike arrays from model that was ran twice
#36
cmazzio
closed
2 weeks ago
12
Split `fit_with_fit_handler` script into separate model fitting and plotting scripts
#35
abuzarmahmood
opened
3 weeks ago
1
Changing number of states does not change model fitting
#34
cmazzio
closed
3 weeks ago
4
export inferred changepoints as numpy
#33
cmazzio
opened
1 month ago
0
Same plots for two different models
#32
cmazzio
opened
1 month ago
1
Can't open spyder in pytau environment
#31
cmazzio
opened
1 month ago
1
Add Unit testing and pipeline testing
#30
abuzarmahmood
opened
6 months ago
0
instead of sorted beta variables, use dirichlet distribution to model changepoints
#29
abuzarmahmood
opened
6 months ago
0
Diagnostics: plots of raw data points for each state
#28
abuzarmahmood
opened
7 months ago
0
Upgrade to pymc v5 and use numpyro backend for mcmc sampling
#27
abuzarmahmood
opened
8 months ago
0
when deleting pkl files, also remove info files
#26
cmazzio
opened
9 months ago
0
Create pytau wiki/docs
#25
abuzarmahmood
opened
1 year ago
0
Notebook showing Dirichlet process prior vs elbo based state count determination
#24
abuzarmahmood
opened
1 year ago
0
Summed counts (binomial) likelihood
#23
abuzarmahmood
opened
1 year ago
0
Implement dirichlet process over trial-switch components
#22
abuzarmahmood
opened
1 year ago
0
Useful to have composable models
#21
abuzarmahmood
opened
1 year ago
0
Move directory path handling to top of notebooks
#20
abuzarmahmood
closed
1 year ago
0
Allow parallel model fitting using separate compiledir_s for theano
#19
abuzarmahmood
closed
1 year ago
0
Add function to loop over model number
#18
abuzarmahmood
closed
1 year ago
1
Allow figure sizes to be adjusted from function arguments in plotting
#17
abuzarmahmood
opened
1 year ago
0
Parallelize using separate compile dirs
#16
abuzarmahmood
closed
1 year ago
1
Mean offset model can't do vi or HMC because of min max operations
#15
abuzarmahmood
opened
1 year ago
0
Add <loop over trial-switch changepoint components> as single model
#14
abuzarmahmood
opened
1 year ago
1
in inference_output, also print what region the data came from
#13
thomasrgray
opened
1 year ago
0
in walkthroughs, MODEL_SAVE_DIR is defined after it is called
#12
thomasrgray
closed
1 year ago
0
changepoint_analysis.state_firing only calculates spike_counts, not actual firing rates
#11
abuzarmahmood
opened
1 year ago
0
Instead of dealing with hdf5, change api to deal with spike trains similarly to sklearn flow
#10
abuzarmahmood
opened
1 year ago
1
Allow manually adding spike train
#9
abuzarmahmood
closed
1 year ago
1
Central model repository with versioning
#8
abuzarmahmood
opened
2 years ago
0
Save "lambda' and 'tau' in model output as histograms or quantiles
#7
abuzarmahmood
opened
2 years ago
0
Python-based theano compile-lock aware loop for parallel runs
#6
abuzarmahmood
opened
2 years ago
0
Separate inference samples from model
#5
abuzarmahmood
opened
2 years ago
0
Add theano and python version to metadata
#4
abuzarmahmood
closed
2 years ago
0
Add plotting capability
#3
abuzarmahmood
closed
1 year ago
3
Add integration with Kubernetes to allow multinode parallelization
#2
abuzarmahmood
opened
2 years ago
0
Add theano compile_lock awareness in model fit to allow joblib parallelization
#1
abuzarmahmood
closed
2 years ago
1