NeurIPS / ICLR already require it.
Explain more.
NeurIPS / ICLR already require it.
Explain more.
2026 SOTA
What is this?
Theodoris 2026 Nat Methods (in press) — "Where single-cell FMs go after the linear-baseline reckoning."
Please help me to find this paper
Many AACR 2026 posters that didn't put scGPT/Geneformer in the title still use the embeddings under the hood.
I wonder how they did that
Deep dives
We can have one more table on this session
Geneformer-V2-104M_CLcancer ties scGPT and beats scFoundation on cell-type classification at 1/3 the params; no model has a robust perturbation-prediction win over latent-additive + scGPT-embeddings baseline
Can you explain more on this?
foundation models
I think it is a time for us to have detail description of what is a certain term. Like, what does 'Foundation Model' means? What does 'Virtual Cells' means? Etc.
Quantization
What does those quantization means?
Distillation
Is there a further resource I can read for learning that concept?
The Pipeline
Is that still the standard / new to art training we have?
tokens
what is the definition of token here
multiplexed protein imaging (CODEX),
I need to learn more about what is CODEX
the human DLPFC dataset
Check what is that dataset