Theo Rashid
@theorashid.bsky.social
modelling in London
theorashid.github.io
theorashid.github.io
@ericmjl.bsky.social just enjoying some of your modal posts.
The only part missing is how modal would work in an MLOps setting. I wonder if there's anything that can get the most out of modal in a DAG with orchestration support – i.e. flyte/metaflow but modal instead of some Kubernetes cluster.
The only part missing is how modal would work in an MLOps setting. I wonder if there's anything that can get the most out of modal in a DAG with orchestration support – i.e. flyte/metaflow but modal instead of some Kubernetes cluster.
April 30, 2025 at 12:10 AM
@ericmjl.bsky.social just enjoying some of your modal posts.
The only part missing is how modal would work in an MLOps setting. I wonder if there's anything that can get the most out of modal in a DAG with orchestration support – i.e. flyte/metaflow but modal instead of some Kubernetes cluster.
The only part missing is how modal would work in an MLOps setting. I wonder if there's anything that can get the most out of modal in a DAG with orchestration support – i.e. flyte/metaflow but modal instead of some Kubernetes cluster.
A while ago, I used to research atmospheric physics. Specifically, tropical cyclone data.
The original code I used was some low-quality pandas stuff. Given the original dataset is an .nc file, I redid the analysis with xarray. github.com/theorashid/x...
The original code I used was some low-quality pandas stuff. Given the original dataset is an .nc file, I redid the analysis with xarray. github.com/theorashid/x...
GitHub - theorashid/xarray-tropical-cyclones: Rewriting some tropical cyclone analysis.
Rewriting some tropical cyclone analysis. Contribute to theorashid/xarray-tropical-cyclones development by creating an account on GitHub.
github.com
December 28, 2024 at 10:10 AM
A while ago, I used to research atmospheric physics. Specifically, tropical cyclone data.
The original code I used was some low-quality pandas stuff. Given the original dataset is an .nc file, I redid the analysis with xarray. github.com/theorashid/x...
The original code I used was some low-quality pandas stuff. Given the original dataset is an .nc file, I redid the analysis with xarray. github.com/theorashid/x...
Rewriting some simulation code from my undergrad now that I'm better than a python while loop. github.com/theorashid/n...
GitHub - theorashid/numba-oslo: Oslo model with help from numba.
Oslo model with help from numba. Contribute to theorashid/numba-oslo development by creating an account on GitHub.
github.com
December 28, 2024 at 10:09 AM
Rewriting some simulation code from my undergrad now that I'm better than a python while loop. github.com/theorashid/n...
Using bayeux to massively reduce blackjax mcmc boilerplate inference code for a dynamax (jax) state space model:
jax-ml.github.io/bayeux/examp...
jax-ml.github.io/bayeux/examp...
Using with Dynamax - bayeux
The documentation for the bayeux software library.
jax-ml.github.io
December 28, 2024 at 10:09 AM
Using bayeux to massively reduce blackjax mcmc boilerplate inference code for a dynamax (jax) state space model:
jax-ml.github.io/bayeux/examp...
jax-ml.github.io/bayeux/examp...
I had my viva yesterday. This paper came out last night. What a time to be alive.
I had to write these tweets, so read them in my voice. The rest of this thread is to keep track of some of the press.
(12/12/2023)
x.com/Pathways2Equ...
I had to write these tweets, so read them in my voice. The rest of this thread is to keep track of some of the press.
(12/12/2023)
x.com/Pathways2Equ...
x.com
x.com
December 28, 2024 at 10:07 AM
I had my viva yesterday. This paper came out last night. What a time to be alive.
I had to write these tweets, so read them in my voice. The rest of this thread is to keep track of some of the press.
(12/12/2023)
x.com/Pathways2Equ...
I had to write these tweets, so read them in my voice. The rest of this thread is to keep track of some of the press.
(12/12/2023)
x.com/Pathways2Equ...
I was trying to find out where Cauchy distributions were used, was reminded about Gull's lighthouse problem, and found this great article and demonstration:
mjoldfield.com/atelier/2017...
mjoldfield.com/atelier/2017...
Martin’s Atelier: Gull’s Lighthouse
mjoldfield.com
December 28, 2024 at 10:01 AM
I was trying to find out where Cauchy distributions were used, was reminded about Gull's lighthouse problem, and found this great article and demonstration:
mjoldfield.com/atelier/2017...
mjoldfield.com/atelier/2017...
The latest paper from the dream team of Bennett and Rashid (2023). www.sciencedirect.com/science/arti...
December 28, 2024 at 10:00 AM
The latest paper from the dream team of Bennett and Rashid (2023). www.sciencedirect.com/science/arti...
You can now fit aggregated spatial level data in numpyro 0.10.0 with a CAR distribution. So spatial modellers have an alternative to R.
(Thanks a bit to me but a lot to fehiepsi for dealing with me during numpyro's longest ever PR – "3x more comments than the runner up".) github.com/pyro-ppl/num...
(Thanks a bit to me but a lot to fehiepsi for dealing with me during numpyro's longest ever PR – "3x more comments than the runner up".) github.com/pyro-ppl/num...
Add CAR distribution by theorashid · Pull Request #1394 · pyro-ppl/numpyro
Hi @fehiepsi,
This is for #1390. There's a lot going on in the numpyro code base, and I'm fairly new to writing good code, so if it's okay I'll open this draft PR and we can discuss...
github.com
December 28, 2024 at 9:58 AM
You can now fit aggregated spatial level data in numpyro 0.10.0 with a CAR distribution. So spatial modellers have an alternative to R.
(Thanks a bit to me but a lot to fehiepsi for dealing with me during numpyro's longest ever PR – "3x more comments than the runner up".) github.com/pyro-ppl/num...
(Thanks a bit to me but a lot to fehiepsi for dealing with me during numpyro's longest ever PR – "3x more comments than the runner up".) github.com/pyro-ppl/num...
I've ported my multilevel mortality model to numpyro so I can let their JAX samplers loose on my large dataset.
num.pyro.ai/en/latest/ex...
num.pyro.ai/en/latest/ex...
Example: Modelling mortality over space and time — NumPyro documentation
num.pyro.ai
December 28, 2024 at 9:57 AM
I've ported my multilevel mortality model to numpyro so I can let their JAX samplers loose on my large dataset.
num.pyro.ai/en/latest/ex...
num.pyro.ai/en/latest/ex...
December 28, 2024 at 9:56 AM
"I wrote these tweets, so read them in my voice. I'll use this thread to keep track of some of the press."
x.com/Pathways2Equ...
x.com/Pathways2Equ...
x.com
x.com
December 28, 2024 at 9:53 AM
"I wrote these tweets, so read them in my voice. I'll use this thread to keep track of some of the press."
x.com/Pathways2Equ...
x.com/Pathways2Equ...
December 28, 2024 at 9:43 AM
December 28, 2024 at 9:38 AM
x.com
x.com
December 28, 2024 at 9:36 AM
"One Day" - the b-side to "Seasons (Waiting on You)": youtube.com/watch?v=bvkE...
YouTube
Share your videos with friends, family, and the world
youtube.com
December 28, 2024 at 9:35 AM
"One Day" - the b-side to "Seasons (Waiting on You)": youtube.com/watch?v=bvkE...
December 28, 2024 at 9:34 AM
December 28, 2024 at 9:34 AM
"“To put it bluntly, the discipline of economics has yet to get over its childish passion for mathematics and purely theoretical and often highly ideological speculation, at the expense of historical research and collaboration with the other social sciences."
—Thomas Piketty"
-- tweet
—Thomas Piketty"
-- tweet
December 28, 2024 at 9:25 AM
"“To put it bluntly, the discipline of economics has yet to get over its childish passion for mathematics and purely theoretical and often highly ideological speculation, at the expense of historical research and collaboration with the other social sciences."
—Thomas Piketty"
-- tweet
—Thomas Piketty"
-- tweet
going to backfill a few from twitter
December 28, 2024 at 9:22 AM
going to backfill a few from twitter