thanosan

cs @caltech | prev at @rbc

[Xander Chin @XanderChin]

i spent the last 3 months building https://t.co/6HZis0g5SC, a minimal TPU that runs both batched inference and training fully on-chip!

here's a breakdown of how it was built 🧵

[Robleh @robjama]

Tyler Cowen on why Mississauga, Brampton, Markham, Scarborough are producing world class talent

Toronto suburbs are our YC and Tyler Cowen is our PG

[Benchling @benchling]

🧑‍🔬 At age 14, @pdhsu cold-emailed 50 professors asking to join a lab. One said yes. Now he’s a #CRISPR pioneer, cofounder of @arcinstitute, and builder of large-scale #AI models for biology.

His story is a reminder: follow the fun, bet on impact, and share what you learn. https://t.co/oxvHCHvAJi #Transcribed

[Jay Yang @Jayyanginspires]

Co-founder and CEO of Stripe's advice to ambitious young people:

[Eric Topol @EricTopol]

Discovery of a biodegradable, fully recyclable plastic
@ScienceMagazine
https://t.co/ib3Kogp6jT

[Eric Topol @EricTopol]

Adding to the remarkable acceleration of digital biology via large language of life models (LLLMs) reviewed in our panel this week, today there's RhoFold+, accurately predicting 3-D RNA structure https://t.co/g4Xo1xOE99
@naturemethods @MITdeptofBE @MIT

[Gabriele Corso @GabriCorso]

Thrilled to announce Boltz-1, the first open-source and commercially available model to achieve AlphaFold3-level accuracy on biomolecular structure prediction! An exciting collaboration with @jeremyWohlwend, @pas_saro and an amazing team at MIT and Genesis Therapeutics. A thread!

[Andrej Karpathy @karpathy]

Remember exercise pages from textbooks? Large-scale collection of these across all realms of knowledge now moves billions of dollars. Textbooks written primarily for LLMs, compressed to weights, emergent solutions served to humans, or (over time) directly enacted for automation.

[thanosan @thanosanp]

@jai_mansukhanii @_buildspace @_nightsweekends @FarzaTV for sure, will check it out :)

[thanosan @thanosanp]

introducing flourish - ai hardware to help gardeners make smarter decisions

it analyzes various metrics from your garden and uses ai to help give insights and recommendations

dm me to get on the waitlist!

cc @_buildspace @_nightsweekends @FarzaTV

[thanosan @thanosanp]

@_buildspace https://t.co/FWBd5HTakZ

[thanosan @thanosanp]

introducing flourish - ai hardware to help gardeners make smarter decisions

it analyzes various metrics from your garden and uses ai to help give insights and recommendations

dm me to get on the waitlist!

cc @_buildspace @_nightsweekends @FarzaTV

[thanosan @thanosanp]

@Sravan1011 @_buildspace @_nightsweekends @FarzaTV the llm runs on the website, not on the esp32.

the esp32 just collects the data and then passes it over to a firebase database. the website then uses the data from the firebase database for llm inference.

[thanosan @thanosanp]

@thisis_arnav @_buildspace @_nightsweekends @FarzaTV yep, it's related to the specific plant

[thanosan @thanosanp]

@thisis_arnav @_buildspace @_nightsweekends @FarzaTV at this stage, it just uses the sensor data to provide insights and tips on how to improve my plants (e.g. to water it more often, to change the type of soil, etc.)

[thanosan @thanosanp]

@Sravan1011 @_buildspace @_nightsweekends @FarzaTV so currently, the toy version of this just periodically takes various metrics through various sensors, and then performs algorithms to preprocess the value, and then finally, passes it through a fine-tuned llm to give insights

[thanosan @thanosanp]

@pablocpz_ai @_buildspace @_nightsweekends @FarzaTV thank you. will definitely check it out :)

[thanosan @thanosanp]

@pablocpz_ai @_buildspace @_nightsweekends @FarzaTV i'm using an esp32.

the speed of inference for the llm is ~2s. over the next few days, my goal is to focus more on the software and to improve the accuracy and speed of llm inference.

[thanosan @thanosanp]

@Prajwxl @_buildspace @_nightsweekends @FarzaTV thanks :)

i use conformal coating to serve as a protective layer.

i also make sure to follow any of the tips that the manufacturers gave me to increase product lifetime (such as only powering on soil moisture sensor when needed to prevent corrosion)

[thanosan @thanosanp]

@davezaxh @_buildspace @_nightsweekends @FarzaTV the cost of the entire product is ~$80. most of the cost is for the ph sensor.

the entire module is powered on by a 9v battery. eventually, i'm planning on adding some sort of casing around the entire module.

[Z Fellows @zfellows]

“A bird sitting on a tree is never afraid of the branch breaking, because its trust is not on the branch, but on its own wings.”