Thomas Lefebvre
@tlbvr.com
Product/ML Eng. I'm building unsplash.com
I write things on tlbvr.com
Montreal 🇨🇦
I write things on tlbvr.com
Montreal 🇨🇦
I'm setting up K-fold cross validation producing K models, each trained on a subset of the training set and predicting on a validation set. Appending those predictions together, we ensure that the whole dataset has been seen but equally split amongst the models.
July 28, 2025 at 11:34 PM
I'm setting up K-fold cross validation producing K models, each trained on a subset of the training set and predicting on a validation set. Appending those predictions together, we ensure that the whole dataset has been seen but equally split amongst the models.
Friday night. Will this job pass before I pass out?
July 25, 2025 at 9:56 PM
Friday night. Will this job pass before I pass out?
I spawned two machines. Someone stops me.
July 24, 2025 at 12:12 AM
I spawned two machines. Someone stops me.
I live in the MLFlow dashboard these days. Let's see if `likeable-horse-161` performs better.
July 23, 2025 at 3:31 PM
I live in the MLFlow dashboard these days. Let's see if `likeable-horse-161` performs better.
I'm preparing a presentation for my team @unsplash.com to walk them through our recent progress in categorizing illustrations
If you missed it, I wrote a blog post about it bsky.app/profile/tlbv...
If you missed it, I wrote a blog post about it bsky.app/profile/tlbv...
July 9, 2025 at 9:15 PM
I'm preparing a presentation for my team @unsplash.com to walk them through our recent progress in categorizing illustrations
If you missed it, I wrote a blog post about it bsky.app/profile/tlbv...
If you missed it, I wrote a blog post about it bsky.app/profile/tlbv...
Making good progress. Here's a peek — Probably gonna take me another week or so.
June 23, 2025 at 8:51 PM
Making good progress. Here's a peek — Probably gonna take me another week or so.
When your day is going well, very much know what needs to be done.
June 5, 2025 at 2:59 PM
When your day is going well, very much know what needs to be done.
I dont know how folks can even read news on their phone nowadays
April 30, 2025 at 4:23 PM
I dont know how folks can even read news on their phone nowadays
Saturday morning maintenance?
February 22, 2025 at 1:37 PM
Saturday morning maintenance?
Nice! I’m in the middle of this one. Will queue it up.
January 30, 2025 at 4:08 PM
Nice! I’m in the middle of this one. Will queue it up.
Looks like I found a winner. nomic-embed-text yields a higher F score even when prioritizing precision over recall.
Gonna clean up the notebook and write up a short post about this experiment. Learned a lot!
Gonna clean up the notebook and write up a short post about this experiment. Learned a lot!
December 29, 2024 at 5:03 PM
Looks like I found a winner. nomic-embed-text yields a higher F score even when prioritizing precision over recall.
Gonna clean up the notebook and write up a short post about this experiment. Learned a lot!
Gonna clean up the notebook and write up a short post about this experiment. Learned a lot!
Post side project wind down.
December 28, 2024 at 7:24 PM
Post side project wind down.
Then asked GPT to come up with the test cases given a sample of my data. The algorithm then picks an embedding model given the best F1 score across all test queries and models and recommends which one I should use and at which cosine similarity threshold.
December 27, 2024 at 10:29 PM
Then asked GPT to come up with the test cases given a sample of my data. The algorithm then picks an embedding model given the best F1 score across all test queries and models and recommends which one I should use and at which cosine similarity threshold.
Got myself a little framework to compare embedding models for retrieving similar restaurants given a search query. I'll make the github repo public soon.
December 27, 2024 at 9:12 PM
Got myself a little framework to compare embedding models for retrieving similar restaurants given a search query. I'll make the github repo public soon.
I'm running tests against different embedding models and cosine similarity thresholds to find the most balanced combo for finding similar restaurants given their description and a search query.
December 26, 2024 at 10:36 PM
I'm running tests against different embedding models and cosine similarity thresholds to find the most balanced combo for finding similar restaurants given their description and a search query.
My Christmas projects are in jeopardy.
December 26, 2024 at 8:35 PM
My Christmas projects are in jeopardy.
FWIW the competition isn't that great either on that front. It's only free for repos on individual accounts, not organizations.
December 10, 2024 at 4:16 PM
FWIW the competition isn't that great either on that front. It's only free for repos on individual accounts, not organizations.
Last future flag is in — next up: upgrading to v7.
December 8, 2024 at 7:42 PM
Last future flag is in — next up: upgrading to v7.
Preparing the move to React router v7. Fixes automatically our Date to string serialization issue.
December 7, 2024 at 8:55 PM
Preparing the move to React router v7. Fixes automatically our Date to string serialization issue.
