36 protein LMs: 3 archs x 4 scales x 3 seeds on UniRef50, compute-optimal. Transformers out-scale Mamba SSMs. Code: github.com/ygzdvr/nanoprot
Yagiz Devre
yagizdevre
AI & ML interests
None yet
Recent Activity
upvoted a paper 4 days ago
Semigroup-JEPA: Latent Dynamics Consistency for Zero-Shot Physics Generalization updated a collection 4 months ago
nanoprot v0.5 — protein LM scaling suite updated a collection 4 months ago
nanoprot v0.5 — protein LM scaling suiteOrganizations
models 15
yagizdevre/nanoprot-mamba-XS
Text Generation • Updated • 6
yagizdevre/nanoprot-mamba-S
Text Generation • Updated • 5
yagizdevre/nanoprot-mamba-M
Text Generation • Updated • 3
yagizdevre/nanoprot-mamba-L
Text Generation • Updated
yagizdevre/nanoprot-gpt2-S
Text Generation • Updated • 7
yagizdevre/nanoprot-gpt2-M
Text Generation • Updated • 5
yagizdevre/nanoprot-gpt2-L
Text Generation • Updated • 5
yagizdevre/nanoprot-esm2-XS
Fill-Mask • Updated • 4
yagizdevre/nanoprot-esm2-S
Fill-Mask • Updated • 4
yagizdevre/nanoprot-esm2-M
Fill-Mask • Updated • 1
datasets 0
None public yet