Today our lab is excited to share a new study in Cell: “Non-invasive control of gene editing in vivo by photoswitchable Cas12f and focused ultrasound.”
We engineered a compact CRISPR system that can be activated at specific locations in the body using focused ultrasound (🧵)
Reasoning from latent representations is interesting. The causal encoder-decoder architecture of DeepSeek V4.1 Flash is more favorable of this manipulation.
starting a new week of sharing notable papers, this series topic is latent communication 🤓
batch 1: the pre-history, where models first started passing each other something richer than text
[1] LLM Augmented LLMs: Expanding Capabilities through Composition (CALM) by Bansal et al, 2024
https://t.co/0tI0jRvQXy composes two frozen models with cross-attention on intermediate representations, the direct ancestor of everything below
[2] Let Models Speak Ciphers (CIPHER) by Pham et al, 2023
https://t.co/zL5ZExVb4x multi-agent debate where messages are embedding-space signals derived from output distributions, the first clean argument that tokenizing a message throws away information the other agent could use
[3] Training Large Language Models to Reason in a Continuous Latent Space (Coconut) by Hao et al, 2024
https://t.co/KRSK9iZNoO feeds the last hidden state back in as the next input embedding, so a model reasons without decoding (one model talking to itself)
Let me tell you a secret: with a sunlight collection device, a fiber-optic cable can channel sunlight into your nasal cavity, and the effect will be much better. Of course, it also works if you point your nasal cavity directly at the sun. Going a step further, if you know a secret of the Maya, you will master the most valuable secret.
@aimalysheva The principle behind is that depth unfolds the underlying manifold. Latent manipulation can be performed in the intrinsic space of the manifold. Please read:
Rank Diminishing in Deep Neural Networks
I spent the last six months trying to deconstruct Taalas's patents. We think it can be better: 100x fewer memfetches than their bitROM, better software by better quantization than their hardware team could.
So we wrote a compiler that:
> takes a huggingface checkpoint,
> quantizes the model
> descends the weights down metal layers to RTL and GDS, do your DRC, Yosys, PEX, make the electrical waveform execute a mat-vec from your huggingface checkpoint (thanks Cambricon tech papers)
in ~7000 lines of human-readable code.
we're looking for someone with contacts with a foundry/access to 7nm PDKs or contacts at a cheap EuroPTW shuttle? I'm broke and unemployed.
@itsclivetime this is what i think is the future of perplexity per picojoule. @zerohedge@zephyr_z9 you guys wanna see 40,000 tokens/sec?