charlie gauthier
LightNews LightNews
banner
velythyl.bsky.social
charlie gauthier
@velythyl.bsky.social
270 followers 190 following 8 posts
PhD student | machine learning & robots

i like bouldering, dnd (sometimes), vibing, writing prose, and being silly and goofy

Posts Replies Media Videos
Reposted by charlie gauthier
lily-is.online
My new video, Notes on Vanishing, is out now.

I keep returning to an unthinkable question: is it time for trans people to make ourselves disappear?

Enjoy 🌃
youtu.be/cqhiup5qSY8
Notes on Vanishing
YouTube video by Lily Alexandre
youtu.be
May 1, 2025 at 2:43 AM
velythyl.bsky.social
Been using Molmo a lot, super excited for this Olmo release too!
kylelo.bsky.social Kyle Lo @ NeurIPS 2025 @kylelo.bsky.social · Nov 26
Excited to share OLMo 2!

🐟 7B and 13B weights, trained up to 4-5T tokens, fully open data, code, etc
🐠 better architecture and recipe for training stability
🐡 staged training, with new data mix Dolmino🍕 added during annealing
🦈 state-of-the-art OLMo 2 Instruct models

#nlp #mlsky

links below👇
A scatter plot comparing language models by performance (y-axis, measured in average performance on 10 benchmarks) versus training computational cost (x-axis, in approximate FLOPs). The plot shows OLMo 2 models (marked with stars) achieving Pareto-optimal efficiency among open models, with OLMo-2-13B and OLMo-2-7B sitting at the performance frontier relative to other open models like DCLM, Llama 3.1, StableLM 2, and Qwen 2.5. The x-axis ranges from 4x10^22 to 2x10^24 FLOPs, while the y-axis ranges from 35 to 70 benchmark points.
November 27, 2024 at 5:01 AM
Reposted by charlie gauthier
kylelo.bsky.social
Excited to share OLMo 2!

🐟 7B and 13B weights, trained up to 4-5T tokens, fully open data, code, etc
🐠 better architecture and recipe for training stability
🐡 staged training, with new data mix Dolmino🍕 added during annealing
🦈 state-of-the-art OLMo 2 Instruct models

#nlp #mlsky

links below👇
A scatter plot comparing language models by performance (y-axis, measured in average performance on 10 benchmarks) versus training computational cost (x-axis, in approximate FLOPs). The plot shows OLMo 2 models (marked with stars) achieving Pareto-optimal efficiency among open models, with OLMo-2-13B and OLMo-2-7B sitting at the performance frontier relative to other open models like DCLM, Llama 3.1, StableLM 2, and Qwen 2.5. The x-axis ranges from 4x10^22 to 2x10^24 FLOPs, while the y-axis ranges from 35 to 70 benchmark points.
November 26, 2024 at 8:59 PM
Reposted by charlie gauthier
agart-s.bsky.social
I have been drawing a lot of sh #silenthill #silenthi2remake #jamessunderland
November 20, 2024 at 2:23 PM
velythyl.bsky.social
i love bsky i love bsky fuck fascist bird social media
November 12, 2024 at 12:35 AM
velythyl.bsky.social
We’re thrilled to announce that the Morphology-Aware Policy and Design Learning Workshop (MAPoDeL) will be held at CoRL 2024 @corl_conf
in Munich,Germany!! We are now soliciting submissions of workshop papers to be presented at the workshop!
Info ⬇️🧵 #MLSky #robotics #corl #ml
August 28, 2024 at 6:06 PM
Reposted by charlie gauthier
brenttoderian.bsky.social
IMPORTANT: This ranks a bunch of ways we get around cities, using average carbon emissions by transport type in grams per pkm. It breaks emissions down by operation, manufacture etc.

Feet and bikes win. Lots of ways beat electric cars, and everything beats gas cars.

HT @copenhenken
December 31, 2023 at 4:15 PM