capetorch.bsky.social
Chilean 🇨🇱 living in France. I build DL models and pipelines. ML Engineer at W&B
cargobike ♥🚴
https://tcapelle.github.io/
181 posts
651 followers
482 following
Regular Contributor
Active Commenter
comment in response to
post
happy to take a look on a call =)
comment in response to
post
Can you share a workspace?
comment in response to
post
This was a team effort from @morgymcg.bsky.social , Soumik, @parambharat.bsky.social , Agata Mlynarczyk, @ayshthkr.bsky.social and many others!
comment in response to
post
I'm excited to see how the community uses these tools, and I'm looking forward to more innovations in safe and reproducible AI!
Check the scorers and Weave here:
👉 wandb.me/weave_scorers
📚 A colab: wandb.me/scorers_colab
comment in response to
post
A personal highlight was working on the Fluency Scorer powered by AnswerDotAI ModernBERT-base; we hope to move all DeBerta-powered scorers to ModernBert in the next release so we can benefit from the longer context length and training speed!
comment in response to
post
As part of this initiative, we also created comprehensive evaluation datasets, drawing on invaluable contributions from the open-source community. Being a reproducibility-first company, we’ve made the full recipe public, including the scorers, model weights, and the training and evaluation datasets
comment in response to
post
We designed these non-LLM powered scorers to leverage state-of-the-art open source models – from the PleIAI/Celadon toxicity detector to the Vectara hallucination scorer – ensuring that our AI systems are evaluated across multiple dimensions.
comment in response to
post
comment in response to
post
It could have been called Gulf of North America
comment in response to
post
This is my favorite kind of Yoga
comment in response to
post
Échalotes, tomates sèches et moules.
comment in response to
post
We raised this internally! thanks for the info.
comment in response to
post
This budget forcing is really smart. We could do that we prefill on API models no?
comment in response to
post
I think the M3 pro has lower me bandwidth
comment in response to
post
I mostly use Claude these days, and it works very well In cursor integration. I grab o1 for more complex stuff and when I have a detailed plan and output. How is the API speed and reliability?
comment in response to
post
comment in response to
post
It is not there for me (on the paid cursos sub)
comment in response to
post
I want it in composer...
comment in response to
post
Red agentes al estilo Swarm
comment in response to
post
Wow this is a cool resource! Way better than injecting random wikipédia titles.
comment in response to
post
This is the way
comment in response to
post
yeah, I think I will need to do some seeding somehow
comment in response to
post
This is the way, hi hi.
comment in response to
post
Flying back to Santiago from Puerto Montt no.