initlayers
@initlayers
21 UG • LLMs + MLOps + Anything Tech • I don't shitpost, I shit-reply • Trying to build a small, focused circle here on X - Feel free to connect
I am working on an open-source RAG system focused on efficient hybrid retrieval and evaluation. It's been 10 days since I started the project. With mid-terms coming up next week, I've decided to finish it tonight, no matter what. I will be working for the next 10 hours, taking…
Embedding layer vs Linear layer - what's the actual difference? Both can technically be used to create embeddings, but they don't behave the same. An Embedding layer is basically a lookup table of shape v×d (v = vocabulary size, d = embedding dimension). You give it an index,…
19 November 2025 Token embeddings, how are they created for LLMs, coding embedding layers, intuition behind embedding layers and linear neural networks.
18 November 2025 Studied token embeddings: the problem with random initialization, one hot encoding and vectors, word2vec. I wanted to drop the project today, but because of the cloudfare outage I missed my peak reach window. So I'll upload it tomorrow instead.
I've delayed the project launch again, this time it's because of the front-end. The entire repository is ready, but I used Bolt for the front-end and the output looks way more polished than my original version. Now I'm stuck in a dilemma: should I launch the boring current…
If you want to learn cryptography, this is the best book I've found. I used it myself, and every concept and algorithm is explained in great detail. Definitely worth checking out. You don't have to be hardcore cybersecurity person either, you can open it to any algorithm and…
I've finally completed v1. Right now, it doesn't take user uploaded documents. I've included a few sample documents that you can search through for now. I'll be starting another project next, and once that's done, I'll return to this and add the document ingestion feature as…
What do people really mean when they say LLMs understand semantics? Let's break it down without the buzzwords. Most beginners confuse two things: Syntax -> how the sentence is written and Semantics -> what the sentence means. For humans, this separation feels natural. For…
Back to building. Mid-terms are done, and it's project season again. I've officially named my first project VectorFlow-RAG. There's still some documentation left, and I need to vibecode the the front-end. Once that's done, I'll share the repo link here, so that anyone who wants…
Finally, the mid-terms are over. They didn't go well at all, but my first mid-term was strong, so it balances out. For the second half of this month, I'll mostly be applying for internships like mad and working on my second MLOps project. I'll also try to fit some LLM learning…
13 November 2025 NLP Mid Prep: Shallow Parser, Sentiment Analysis, Dialogue and Discount Analysis CC Mid Prep: Disjoint Set, Union Find by Rank, Path Compression, Binomial Heap, Red Black Tree
12 November 2025 CNS Mid Prep: RSA Encryption, Diffie-Hellman Key Agreement, SHA-512 Algorithm, Kerberos Password Management, SSL Architecture, SignCryption NLP Mid Prep: Discourse Analysis, Dialogue Analysis, QA System, Rule Based and Probabilistic Parsing, NER
11 November 2025 CNS Mid Prep: Public Key Infrastructure, Digital Certificate, Digital Signature Working, SignCryption, Kerberos Password Management.
9 November 2025 PRML Mid Prep: Kernel Function, Linear, Fisher, Polynomial, Gaussian Radial Basis Function, Cosine, Two-class problem, Hard constraint SVM, Soft Margin SVM. Linear models for classification, Discriminant functions, fisher's discriminant, Probabilistic…
Our second SIH submission (made on 15th October) wasn't successful. We worked hard for it, but couldn't quite pull it off. Now we're waiting for the batch 2 results. Let's see if our first submission can work some magic.
8 Nov 2025 IDA Mid Prep: Clustering Analysis, K-means, K-mediods, Agglomerative, Divisive, Distance Metrics for Clustering Algorithms, DB Scan, Evaluation of Clustering PRML Mid Prep: Perceptron Algorithm, Support Vector Machine, Hard and Soft Margin SVM.
It's my 🎈 day - just turned 21. My first birthday on X. I've met many awesome people here since I joined in June. Hopefully by my next birthday, you'll look at me as someone who adds real value here.
The most dangerous thing about AI isn't that it lies - it's that it lies confidently. You ask a model something simple. It replies instantly, flawlessly, persuasively, and yet, it's wrong. That's called a hallucination. Large Language Models don't know things. They don't store…
United States Trends
- 1. Knicks 13.4K posts
- 2. Shamet 2,925 posts
- 3. #AEWDynamite 21.8K posts
- 4. #Survivor49 3,949 posts
- 5. #CMAawards 5,536 posts
- 6. Brandon Williams 1,051 posts
- 7. Derik Queen 4,380 posts
- 8. FEMA 68.3K posts
- 9. Vooch 1,162 posts
- 10. Vucevic 6,289 posts
- 11. Labaron Philon 1,118 posts
- 12. Blazers 4,218 posts
- 13. NO CAP 15K posts
- 14. #AEWCollision 8,776 posts
- 15. #SeeRed N/A
- 16. Coby White 1,133 posts
- 17. Nany 2,067 posts
- 18. Naji Marshall N/A
- 19. Chisa 32.1K posts
- 20. FREE HAT 1,659 posts
Something went wrong.
Something went wrong.