🔄 In a Training Loop

Jorge Munoz Laredo

jorgemunozl

LeRobot-worldwide-hackathon

·

https://jorgemunozl.github.io

AI & ML interests

I like Vision Language Action Models, Machine Learning Inter atomic Potentials AI4Science, Diffusion based architectures and I love physics.

Recent Activity

new activity about 12 hours ago

context-course/unit_1_quiz:Certificat Fundamentals of MCP Ranaivozanany

reacted to ProCreations's post with 👍 1 day ago

who want grug 35b?

reacted to Quazim0t0's post with 👀 1 day ago

I never really posted about my DaisyChain project because it's still work in progress. I decided to post a small bit about it and the demo. DaisyChain Genomics: four small DNA/RNA specialists chained behind a learned router that behave like one big genomics model, at ~7× less active compute. I built a modular genomics model chasing a 500M-parameter foundation model, then caught myself measuring it wrong. Here's the honest version. DaisyChain is a different bet: instead of one monolithic DNA model, it's four ~74M specialists (eukaryote, prokaryote, mRNA, splice) chained behind a learned router, each distilled per-domain from HuggingFaceBio's Carbon-500M. Every specialist reports how surprised it is (bits/base) and the router hands each sequence to the link most at home with it. In lineage it's a cluster Branch-Train-Merge mixture of experts, so you can chain on a new domain without retraining the others. The pitch: ~295M total params (under Carbon-500M), but only one ~74M specialist runs per query, so ~7× cheaper per token, routing at 100% held-out. The mistake: Carbon works in 6-mers, and I'd been scoring likelihood as 6-mer cross-entropy. By that number I was +0.043 bits/base behind, splice even "beating" Carbon. But Carbon scores at the base-pair level, which is harder and more honest. Re-run their way: Real gap: 1.862 vs 1.787 bits/base, +0.089 behind, not +0.043 No domain actually beats Carbon; the "splice win" was an artifact Seq recovery: euk 31.5% vs 38.9%, bacteria 40.9% vs 54.1% DaisyChain is still behind Carbon-500M (itself a draft model, not built to top benchmarks), but by a number I can defend, and the gap closes with every per-domain pass. 🌼 https://huggingface.co/DaisyChainAI https://huggingface.co/spaces/DaisyChainAI/Daisychain-Genomics-Demo https://huggingface.co/DaisyChainAI/daisychain-genomics

View all activity

Organizations

jorgemunozl 's models 17

jorgemunozl/pi05_ki_vlm

Any-to-Any • 4B • Updated 9 days ago • 21 • 1

jorgemunozl/mace_omat_medium

Graph Machine Learning • Updated 14 days ago • 1

jorgemunozl/mace_omat_lora_v1

Graph Machine Learning • Updated 14 days ago • 1

jorgemunozl/pi05_ki_pg2_general_task_index

4B • Updated May 4 • 1 • 1

jorgemunozl/pi05_denoise_flow_matching

Updated Feb 24 • 1

jorgemunozl/pi05_ki_vlm_train_v2

Updated Feb 9 • 1

jorgemunozl/pi05_overfit_test

Robotics • 4B • Updated Feb 6 • 7 • 1

jorgemunozl/pi05_unified

Updated Feb 6 • 1

jorgemunozl/pi05_ki_cropped

Updated Feb 5 • 1

jorgemunozl/pi05_ki_open

Updated Feb 1 • 1

jorgemunozl/pi05_ki_vlm_freq_1

4B • Updated Jan 29 • 5 • 1

jorgemunozl/pi05_ki_vlm_freq_2

4B • Updated Jan 29 • 1

jorgemunozl/pi05_ki_vlm_stable

4B • Updated Jan 29 • 1 • 1

jorgemunozl/pi05_ki_vlm_v2

Updated Jan 29 • 1

jorgemunozl/psiformer_torch

Reinforcement Learning • Updated Jan 8 • 1

jorgemunozl/pi05_base_ex3_slow_best

Updated Nov 26, 2025 • 1

jorgemunozl/flowchart2mermaid

Updated Jul 22, 2025 • 1