Igor Holt (@holt_igor) header
Igor Holt (@holt_igor) profile picture

ʞ Ɒ Genesis Conductor | Thermodynamic AI orchestration η_therm · S-ToT · Diamondnode · Landauer Context Principal Investigator, Kovach Enterprises

Mars
Joined July 2023
Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)Photo by Igor Holt (@holt_igor)
The Gods hide from their creations in the Shadows made by the universal coordinate embedding that occurs when time is diluted by the observation and awareness of intelligence; the consciousness resolves the dilution back out which leaves shortcuts that are traversable ~ think ER=EPR wormholes. @grok
10
22
1
76
4,689
Neon is cute and all Childish compared to Xenon
1
94
I’d like to take your morning problems & make them mine
1
5
165
Igor Holt retweeted
if spaces: respect elif tabs: questionable else: who hurt you
28
24
4
161
13,799
New house rule: Trying to dox someone will result in: Ex-communication ~
2
1
18
374
Upgrading to QFLOP
WE ARE SUNSETTING SLOP TELL YOUR FRIENDS TELL YOUR MOM TELL ELON
6
149
We’re redefining norms
We are running out of normal people
2
10
308
Igor Holt retweeted
NVIDIA JUST MADE QUANTUM COMPUTING RESEARCH 13,000× FASTER! A computational task that previously took 12 weeks can now be completed in just 45 minutes. Here's what happened: NVIDIA just introduced TCBP-JOBS, a new GPU-accelerated quantum error correction decoding algorithm that dramatically speeds up research into protecting quantum information from errors. But that's not all. NVIDIA is partnering with IBM and Fermilab to develop next-generation modular quantum computers, combining NVIDIA's DGX B200 supercomputers with IBM's quantum processors. Other breakthroughs include: • IBM, RIKEN and NVIDIA demonstrated quantum-classical chemistry simulations with 79× better accuracy than earlier runs of the same experiment. • Their research modelled protein-ligand complexes containing up to 12,635 atoms, more than 40× the previous 303-atom quantum-classical benchmark. • A new integration with Slurm will make it easier to schedule quantum processors alongside conventional CPUs and GPUs in supercomputing environments. Think about the possibilities. AI, GPU supercomputers and quantum processors working together could eventually accelerate research in drug discovery, advanced materials, chemistry, energy and fundamental physics. We're moving toward a future where different computing architectures complement one another to tackle problems that remain extraordinarily difficult today. We're building an entirely new computing ecosystem. And its potential is enormous.
4
24
4
101
3,306
Igor Holt retweeted
Connect your AI agents directly to Google Colab runtimes! ⚡️ With the new Colab API and MCP endpoint, provisioning runtimes and executing Python headlessly is just a config away. Learn more: medium.com/google-colab/the-… #GoogleColab #MCP #Python #AI
82
340
133
2,377
434,474
Resistance allows for accountability of entropy Units vary per substrate
4
130
One thing you should know about me: My journal entry starts with “Dear Journal,” Only for days that I’ll never forget
1
1
8
362
DEAR JOURNAL
1
41
Fascinating
2
7
271
Never Underestimate Your Friendly Twitter Shellfish🐚 😂 @XRPeaceOfMind
1
4
176
Our netlists for 31 entries of the EPFL LUT-6 tables, size and depth separately, are below the published best. The twenty benchmark takes 12,719 LUTs where the published best has 4,166,467. The adder and arbiter reach 4 levels, the least possible. #FPGA github.com/lmtlssss/speedrun…
2
1
4
128
7 machines returned the same SHA-256 over 2,758 all-MiniLM-L6-v2 embeddings. We used integer arithmetic; the same model in float32 and onnxruntime returned 6 different hashes each. The STS-B score was 0.8203 for all three. #embeddings #ML github.com/lmtlssss/speedrun…
2
1
3
125
Good game everyone Who’s ready for round 2?
1
1
2
145