Loading video...
Video Failed to Load
🇺🇸 Independence Week Special Celebrate with a special offer from Evo Labs Research : purchase any 2 products and receive 1 additional product FREE. Every batch is backed by a COA and verified at 99%+ purity. Fast shipping from Tampa, FL. Link in bio. For research use only. Not... show more
20,308 views • 1 month ago •via X (Twitter)
0 Comments
No comments available
Comments from the original post will appear here
![I'm starting on a new project #buildinpublic 🤠 ✅ ahrefs sub 📚 initial keyword research 🆕 .com domain name But this is different. Why? I'm starting with SEO & marketing, then fleshing out the product afterwards. It's been a busy morning already thanks to Ahrefs. From the learnings of grandmaster sensei — I will be using my other projects for backlinks & focus on SEO instead of adding as an afterthought. The plan: ------------------------- [hacks explained further down] 1. keyword research 2. more keyword research [HACK #1] 3. content plan 4. choose topics, subtopics & post outline using long-tail keywords found in 1. & 2. [HACK #2] 5. landing page, blog & initial marketing 6. barebone MVP 7. blog with content hubs [HACK #3] 8. initial backlinks [HACK #4] 9. marketing & launch 10. talk to users & iterate while SEO is slowly cooking in the background (hopefully by the time the product matures SEO is booming) ... 🔁 continue to fill in posts according to the content plan. then back to 1 & 2. more long-tail posts & create free tools to improve ranking + increase no. of backlinks. Hack #1 --------- Explore keywords for "result intent" SEO. Figure out guides & tutorials for keywords with DR long tail ones. Create structure for internal links, e.g.: /generic-topic-keyword (links to all posts) ➡️/more-specific-subtopic-keyword (links to child posts) ➡️➡️/very-specific-post-1 ➡️➡️/very-specific-post-2 etc. Hack #3 --------- Create hubs for generic keywords with links to subtopics and long-tail posts. The hubs themselves should be somewhat informative but mostly an overview. => My crazy idea: before I have the content, add external links to authoritative sources for each specific post. I'll slowly write my own content to replace those and move the links inside the post. Hack #4 --------- Use my other projects to write posts on the new product and link to it to get a decent domain ranking fast. Launch on PH with a beta, mostly for the good backlink. Add repos with md files to Github, Gitlab, Bitbucket etc. for some easy backlinks. 🤠 Crazy enough to work, right?* *to note: I've validated the idea and am somewhat sure people will pay for it. But, I'd still like to shorten each step to minimize my risk & ship fast. Thinking: 1 week research, 1 week dev, 1 week marketing, 1 week content. What could go wrong? :D](https://image.24vids.com/tw-1696475630504648989/ext_tw_video_thumb/1696473936395948032/pu/img/FhI4fGVM6r1xzoIv.jpg)
![Batch Normalization by hand ✍️ ~ 7 steps walkthrough below Batch normalization is common practice for improving training and achieving faster convergence. It sounds simple. But it is often misunderstood. 🤔 Does batch normalization involve trainable parameters, tunable hyper-parameters, or both? 🤔 Is batch normalization applied to inputs, features, weights, biases, or outputs? 🤔 How is batch normalization different from layer normalization? So I drew and calculated one entirely by hand. Goal: normalize a mini-batch of 4 examples to mean 0 and variance 1, then let the network scale it back. = 1. Given = A mini-batch of 4 training examples, each with 3 features. = 2. Linear layer = Let us multiply by the weights and add the biases. Batch norm sits after this, which answers the second question: what gets normalized is features, not inputs, weights or biases. = 3. ReLU = We apply the activation, and -2 becomes 0. Negative values are suppressed before any statistic is taken. = 4. Batch statistics = Let us compute the sum, mean, variance and standard deviation, one row at a time. A row is a feature and the four columns are the four examples, so every number here measures one feature against the rest of the batch. That is the "batch" in batch normalization, and it is exactly what layer normalization does not do. The statistics are rounded to whole numbers, which is what keeps the rest of the page doable in pen. = 5. Shift to mean 0 = We subtract the mean, in green. The four values in each feature now average to zero. = 6. Scale to variance 1 = Let us divide by the standard deviation, in orange. Each feature now has variance one, whatever scale it arrived at. = 7. Scale and shift = We multiply by a linear transformation and pass the result on. The diagonal and the last column are trainable, so having just forced every feature to mean 0 and variance 1, we hand the network the means to undo it. The outputs: Mean of each feature = [2, 1, 2] Std dev of each feature = [1, 1, 2] To the next layer = [2, -2, 2, 0], [-3, 3, 6, -3], [2, 0, 1, 2] The answers: 🤔 Both. The scale and shift are trainable, the statistics are not. Epsilon and the momentum on the running statistics are the hyper-parameters, and one mini-batch by hand needs neither. 🤔 Features, after the linear layer, not inputs, weights or biases. 🤔 Batch norm measures across the batch, one feature at a time. Layer norm measures across the features, one example at a time. 💾 Save this post!](https://image.24vids.com/tw-2079570896423911664/ext_tw_video_thumb/2079570876530298880/pu/img/w_TJ6G4bD69KBKuz.jpg)
