@TensorTonic: in machine learning, the RBF kernel measures similarity between points. increasing gamma makes similarity fall faster w…
Summary
A promotional post for TensorTonic, an online platform for learning machine learning through coding practice, starting with an explanation of the RBF kernel.
View Cached Full Text
Cached at: 09/21/26, 11:42 PM
in machine learning, the RBF kernel measures similarity between points. increasing gamma makes similarity fall faster with distance.
explained visually below.
https://t.co/rwSSNnPGfa https://t.co/XuaApvIxm6
TensorTonic | Learn ML through Code
Source: https://www.tensortonic.com/
LearnMachine LearningMachine Learning through code
Implement1000+ algorithmsfrom scratch, from foundational ML toCUDA kernels, with the math, visualizations, and research papers behind them.
gradient_descent.pyRun
Loss surfaceL(w)L(\mathbf{w})
step0|loss5.372
- ∇L(w)\nabla L(\mathbf{w})is correct
- L(wk+1)<L(wk)L(w_{k+1}) < L(w_k)every step
- w→0\mathbf{w} \to 0converges
All tests passed ✓
What learners are saying
0Million+
code submissions
Plans
Free
$0forever
Start building your ML foundations
- Unlimited code submissions
- 300+ coding problems
- 35 CUDA / GPU kernel problems
- Interactive visualizations
- Theory content per problem
- 13 research paper implementations
- ML Math blogs
- Community Discord access
- Structured study plans
- Real-time interview mock assessments
- ML System Design problems
Pro
$12/month
or $100/yr · save 31%
- Everything in Free
- Structured roadmap study plansPandas, NumPy, PyTorch, SQL FoundationsLinear Algebra, Probability, Calculus, OptimizationML, DL, NLP, RL, CV specializations
- 500+ quiz questions across 12 modules
- Real-time interview mock assessmentsData Scientist, ML Engineer, AI EngineerResearch Engineer, GPU and Performance
- Priority support
- ML System Design problems
- Modern LLM architectures like GPT2, DeepSeek
Most popular
Plus
$20/month
or $168/yr · save 30%
- Everything in Pro
- Implement modern LLM architectures like GPT-2, DeepSeek etc in PyTorch
- ML System Design problems
- Interactive design canvas
- FAANG-styled system design scenarios
Questions, answered
Stop watching. Start building.
Don’t just watch ML theory. Implement it yourself, one algorithm at a time, and understand it in depth.
1000+ problemsReal cloud GPUsLive AI sandboxes
Similar Articles
@ickma2311: ∇ Stanford CS336 (Spring 2026) — Lecture 6: Kernels and Triton A shorter tensor expression does not necessarily mean fe…
斯坦福 CS336（2026 春）第 6 讲的笔记，讲解如何通过减少 GPU 内存往返来优化内核：对比 softmax 多次遍历与融合行内核的内存流量，介绍 Triton 的块级编程模型，并强调先用 CUDA event 测量基线、再做内核融合与分块优化。
@raphaelsrty: Computing max similarity (scoring step of colbert, colpali) on gpus can be optimized and this is what @tonywu_71 did. I…
Tony Wu released late-interaction-kernels (LIK): fused Triton kernels for MaxSim, the scoring step behind ColBERT and ColPali, integrated into PyLate and colpali-engine, offering memory efficiency and performance gains.
@TheAhmadOsman: You wanna learn how do these AI kernels work? Start here
Tweet by @TheAhmadOsman pointing to a resource for learning how AI kernels work.
MetaKernelBench: Measuring GPU Kernel Knowledge Transfer Beyond Code
MetaKernelBench introduces a benchmark measuring whether experience distilled by GPU kernel optimization agents in one DSL (CuTeDSL) transfers to another (TIRx) on the same problem, finding mixed results with lift ranging from -19% to +29% across six models.
@liao_lucas: https://x.com/liao_lucas/status/2097149853499588971
This article provides an introduction to GPU kernels in the context of AI inference and performance engineering, explaining their definition, how they are used, and the advantages of custom kernels for optimization.