Aurélien-Morgan CLAUDON
aurelienmorgan.bsky.social
Aurélien-Morgan CLAUDON
@aurelienmorgan.bsky.social
Wandering around. Mind if I do.
Nice tip for non-natives, BTW.
March 1, 2025 at 5:44 PM
It has made some programming suggestions to me reminiscent of the early Claude 3.5 days (before Anthropic silently allegedly "pseudo-optimized" it). Yeah, I can say that I've been enjoying GPT4.5 on a few occasions as well.
March 1, 2025 at 5:16 PM
Am building hf.co/retrain-pipe...
Am still small but, pilling-up revisions at some important scales could become a thing (for both model and datasets) so, Xet is something I think of a lot since you guys joined HF.
Am just pinging you here, maybe some day will say a more formal Hi, who knows ?
retrain-pipelines (retrain-pipelines)
`pip install retrain-pipelines`, automatic retraining, continuous learning, auto-doc, traceability, auditability, versioning
hf.co
February 21, 2025 at 12:41 PM
Gotta love him
February 14, 2025 at 6:23 PM
So badly sad that they initially did it that way.
February 12, 2025 at 1:23 PM
Betcha
January 31, 2025 at 4:00 PM
Lots of envy. Feels like you guys live the dream life.
January 31, 2025 at 3:15 PM
Mysteries of memory accesses..
January 31, 2025 at 3:14 PM
So much to learn from this
January 31, 2025 at 3:13 PM
🎩hats off
January 31, 2025 at 3:12 PM
`lighteval`. One of the last big block for me to integrate before the next release of `retrain-pipelines`. huggingface.co/retrain-pipe...
retrain-pipelines (retrain-pipelines)
`pip install retrain-pipelines`, automatic retraining, continuous learning, auto-doc, traceability, auditability, versioning
huggingface.co
January 6, 2025 at 12:27 PM
exhilarating
January 4, 2025 at 2:36 PM
Thanks for sharing.
January 2, 2025 at 5:07 PM
Looking forward to reading that report !
December 8, 2024 at 12:39 PM
¡Muchas felicidades, Omar!
December 2, 2024 at 4:16 PM
I'd, yep
November 28, 2024 at 2:35 PM
Look for starter packs (lists of accounts grouped by theme) and, definitely "pin to home feed" that page :
bsky.app/profile/did:...
November 24, 2024 at 5:42 PM
Garreth may have opened Pandora's box here
Brain firecracker!
Many research topics can emerge from this
Studies of the effect of R2L tokenizers on multilingualism with languages such as Arabic, Hebrew, Sindhi for varying tasks: creativity, reasoning
@CohereForAI @sarahooker.bsky.social @cohere.com
November 24, 2024 at 1:44 PM