Nick
@nickwal
Followers
165
Following
5K
Media
1
Statuses
170
banger.
Releasing INTELLECT-2: We’re open-sourcing the first 32B parameter model trained via globally distributed reinforcement learning:. • Detailed Technical Report.• INTELLECT-2 model checkpoint.
0
0
4
RT @PrimeIntellect: Releasing INTELLECT-2: We’re open-sourcing the first 32B parameter model trained via globally distributed reinforcement….
primeintellect.ai
We're excited to release INTELLECT-2, the first 32B parameter model trained via globally distributed reinforcement learning. Unlike traditional centralized training efforts, INTELLECT-2 trains a...
0
302
0
whale bros don’t miss.
🚀 Day 5 of #OpenSourceWeek: 3FS, Thruster for All DeepSeek Data Access. Fire-Flyer File System (3FS) - a parallel file system that utilizes the full bandwidth of modern SSDs and RDMA networks. ⚡ 6.6 TiB/s aggregate read throughput in a 180-node cluster.⚡ 3.66 TiB/min.
0
0
3
hopefully 2025 is the year we stop flexing models overfit on slop benchmarks.
0
0
2
the strangest o1-mini hallucination I have ever seen - I would love to see the hidden CoT for this.
0
0
1
RT @WilliamBryk: o1 is the biggest AI update since transformers and it paves a clear path toward AGI. This is what Ilya saw. This is why S….
0
196
0
rl/mcts has seemed like an obvious progression for months and will continue to get exploration but distillation is an unsung hero.
1
0
0
hoping o1 inspires somebody with actual compute budgets to heavily investigate the test-time-compute/parameter count pareto.
1
0
0
thank you @astral_sh for ending my suffering with python environments.
wtf uv made python good now?. @nickwal's Python PR is so crazy, it lazy loads uv. Which in turn installs Python correctly. So I can set the version I want in the pyproject.toml, match it to my Lambda function and. sst dev
2
1
20
There is a hilarious overlap between people who claim that LLMs cannot take us anywhere because they are exclusively stochastic parrots and a subset of the AI moratorium signatories.
0
0
0