Loading activity…
@yacinemtb
RL, robots, embedded, simulators. prev eng @ x, stripe. 6'3 (height) tensorpunk, Father 2u2 subscribe to read my blog!
AI
DeepSeek's README offers offline inference for DeepSeek-V4.1-Flash but lists additional systems needed for a complete service.
AI
A user credits DeepSeek directly for technological acceleration and predicts people will generally underestimate its contribution.
AI
A user's quote-post frames coding without AI as the setup for a joke.
AI
A user says they work in alignment, then pivots to “aligning the model into being more capable.”
AI
A user recommends Tim Dettmers’ blog as useful advice for undergrads and grad students.
AI
A user mused that 'everything is beginning to compress' toward the singularity, then replied, 'deepseek flash.'
AI
One user claims Opus explanations let them learn 20 times faster; another says they still need human guidance.
AI
A user claims Claude now readily decompiles requests it lectured users about six months ago, suggesting competition with DeepSeek.
AI
A user calls pretraining scaling 'laws' pretentious and praises reinforcement learning for trying crude ideas that could improve gradients.
AI
SemiAnalysis says an upstream vLLM optimization made the 2.8-trillion-parameter model 1.68 times faster on the same hardware.
AI
A user jokes that it “hits different” when every ablation experiment costs $1 million.
AI
SemiAnalysis shares a quoted comparison calling GLM 5.3's cyber capabilities a “night and day” change from 5.2.
AI
A user calls the university's perceived irrelevance surprising, given their view that reinforcement learning dominates modern AI.
AI
A post argues that AI agents make research and testing easier, enabling more projects that eventually hit human bottlenecks.
AI
DeepSeek says the model is live on its website, app and API, and optimized for DeepSeekHarness and Tencent CodeBuddy.
AI
A user pairs the datacenter quip with a photo, without adding context in the text.
AI
A user suggests even an ardent AGI enthusiast might not fully believe in the power of neural networks.
AI
In October 2026, the same user advocated enormous computers and weeks-to-months training runs in a hyperbolic turn.
AI
A user says the tests they point to in a quoted post provide no value.
AI
A user says living in San Francisco isn't necessary to work on frontier AI.
