Mohit
@mohitmayank.com
The AI Guy | Helping Startups 10x AI Game 🤖 | Author “Lazy Data Science Guide” | Creator of Jaal
Chori karne ke liye bhi akal chahiye
November 13, 2025 at 8:38 AM
Chori karne ke liye bhi akal chahiye
2025 is about to end, stop using pip and use uv, it's 10-100x faster! Here is a cheatsheet to get you started 👇
November 13, 2025 at 6:46 AM
2025 is about to end, stop using pip and use uv, it's 10-100x faster! Here is a cheatsheet to get you started 👇
#1 The way your data is represented can make or break your results - Single-vector vs Multi-vector Embeddings
Single-vector embeddings boil down an entire document or query into one compact vector - simple, fast, and effective for broad searches.
Single-vector embeddings boil down an entire document or query into one compact vector - simple, fast, and effective for broad searches.
November 12, 2025 at 11:43 AM
#1 The way your data is represented can make or break your results - Single-vector vs Multi-vector Embeddings
Single-vector embeddings boil down an entire document or query into one compact vector - simple, fast, and effective for broad searches.
Single-vector embeddings boil down an entire document or query into one compact vector - simple, fast, and effective for broad searches.
Watching the Diffusion LLM generate words in real-time 👀
Unlike AR models, which generate text sequentially (left-to-right), Diffusion models like Dream 7B, support any-order generation as visible below
Unlike AR models, which generate text sequentially (left-to-right), Diffusion models like Dream 7B, support any-order generation as visible below
November 12, 2025 at 8:30 AM
Watching the Diffusion LLM generate words in real-time 👀
Unlike AR models, which generate text sequentially (left-to-right), Diffusion models like Dream 7B, support any-order generation as visible below
Unlike AR models, which generate text sequentially (left-to-right), Diffusion models like Dream 7B, support any-order generation as visible below
Fine-tuning TTS or ASR models? Try Unsloth, it speeds up TTS fine-tuning by 1.5 times and uses 50% less memory!
docs.unsloth.ai/basics/text...
docs.unsloth.ai/basics/text...
November 12, 2025 at 6:13 AM
Fine-tuning TTS or ASR models? Try Unsloth, it speeds up TTS fine-tuning by 1.5 times and uses 50% less memory!
docs.unsloth.ai/basics/text...
docs.unsloth.ai/basics/text...
#1 4 must-know precision formats in AI
- FP32 is the full-precision standard typically used in training for numerical stability.
- FP16, or half-precision, is popular for mixed-precision setups, offering a great tradeoff between speed and accuracy.
- FP32 is the full-precision standard typically used in training for numerical stability.
- FP16, or half-precision, is popular for mixed-precision setups, offering a great tradeoff between speed and accuracy.
November 10, 2025 at 6:16 AM
#1 4 must-know precision formats in AI
- FP32 is the full-precision standard typically used in training for numerical stability.
- FP16, or half-precision, is popular for mixed-precision setups, offering a great tradeoff between speed and accuracy.
- FP32 is the full-precision standard typically used in training for numerical stability.
- FP16, or half-precision, is popular for mixed-precision setups, offering a great tradeoff between speed and accuracy.
Building RAG on a budget? Check out Google's File Search Tool - only pay for document embedding, with free storage and query embedding!
November 9, 2025 at 6:11 AM
Building RAG on a budget? Check out Google's File Search Tool - only pay for document embedding, with free storage and query embedding!
I'm publicly training a TTS model and sharing my journey on a blog. It's a detailed guide covering both intuition and code. Read it, star it, follow it, and let me know your thoughts!
November 7, 2025 at 11:30 AM
I'm publicly training a TTS model and sharing my journey on a blog. It's a detailed guide covering both intuition and code. Read it, star it, follow it, and let me know your thoughts!
One blog, ~5000 words, and one of the cleanest introductions to LLM Training!
November 7, 2025 at 6:18 AM
One blog, ~5000 words, and one of the cleanest introductions to LLM Training!
🔥 Kimo K2 Thinking > GPT-5 or Sonnet 4.5
November 6, 2025 at 4:21 PM
🔥 Kimo K2 Thinking > GPT-5 or Sonnet 4.5
🧙♂️ Imagine rewriting your story just by moving characters and events visually!
Why: Visual Story-Writing aims to make story crafting intuitive and dynamic by bridging visual interaction with AI-driven text generation.
Why: Visual Story-Writing aims to make story crafting intuitive and dynamic by bridging visual interaction with AI-driven text generation.
November 6, 2025 at 8:30 AM
🧙♂️ Imagine rewriting your story just by moving characters and events visually!
Why: Visual Story-Writing aims to make story crafting intuitive and dynamic by bridging visual interaction with AI-driven text generation.
Why: Visual Story-Writing aims to make story crafting intuitive and dynamic by bridging visual interaction with AI-driven text generation.
Building a Lightning-Fast LLM from Scratch? Here’s How It’s Really Done!
If you want to deeply understand the difference between “just running a model” and “squeezing every ounce of performance with your bare hands,” Andrew Chan’s deep-dive blog is your blueprint.
1/4
If you want to deeply understand the difference between “just running a model” and “squeezing every ounce of performance with your bare hands,” Andrew Chan’s deep-dive blog is your blueprint.
1/4
November 6, 2025 at 6:16 AM
Building a Lightning-Fast LLM from Scratch? Here’s How It’s Really Done!
If you want to deeply understand the difference between “just running a model” and “squeezing every ounce of performance with your bare hands,” Andrew Chan’s deep-dive blog is your blueprint.
1/4
If you want to deeply understand the difference between “just running a model” and “squeezing every ounce of performance with your bare hands,” Andrew Chan’s deep-dive blog is your blueprint.
1/4
Maya Research, founded by two 23-year-old Indians, open-sourced Maya1 TTS model that is #20 on the global leaderboard! It was created by finetuned LLaMa model + SNAC audio codec. In case you want to know what SNAC is, here is an article mohitmayank.com/a_lazy_data...
November 5, 2025 at 8:31 AM
Maya Research, founded by two 23-year-old Indians, open-sourced Maya1 TTS model that is #20 on the global leaderboard! It was created by finetuned LLaMa model + SNAC audio codec. In case you want to know what SNAC is, here is an article mohitmayank.com/a_lazy_data...
WindSurf's Codemaps release shows that VibeCoding SaaS recognizes users often struggle due to a lack of understanding of their own codebase. "It's not us, it's you" 🫵
November 5, 2025 at 6:18 AM
WindSurf's Codemaps release shows that VibeCoding SaaS recognizes users often struggle due to a lack of understanding of their own codebase. "It's not us, it's you" 🫵
Did you get your free ChatGPT subscriptions? 🇮🇳
here is how to get it - Go to chatgpt.com/ > Click on "Try Go" button > Click on "Upgrade to Go" > Add Card (you will not be charged) > Click "Purchase". Done!
here is how to get it - Go to chatgpt.com/ > Click on "Try Go" button > Click on "Upgrade to Go" > Add Card (you will not be charged) > Click "Purchase". Done!
November 4, 2025 at 9:47 AM
Did you get your free ChatGPT subscriptions? 🇮🇳
here is how to get it - Go to chatgpt.com/ > Click on "Try Go" button > Click on "Upgrade to Go" > Add Card (you will not be charged) > Click "Purchase". Done!
here is how to get it - Go to chatgpt.com/ > Click on "Try Go" button > Click on "Upgrade to Go" > Add Card (you will not be charged) > Click "Purchase". Done!
Trainable vs. Frozen Parameters 👀
November 4, 2025 at 6:15 AM
Trainable vs. Frozen Parameters 👀
I’m surprised how many people STILL don’t know that with LoRA, you don’t have to fine-tune the entire embedding matrix - just the new tokens!
November 3, 2025 at 11:27 AM
I’m surprised how many people STILL don’t know that with LoRA, you don’t have to fine-tune the entire embedding matrix - just the new tokens!
#1 Did you know you can learn the basics of Agentic AI for free on DeepLearning.AI!?
November 3, 2025 at 8:26 AM
#1 Did you know you can learn the basics of Agentic AI for free on DeepLearning.AI!?
Should you train your own model or not?
November 3, 2025 at 6:18 AM
Should you train your own model or not?
ditch the netflix binge this weekend and dive into this instead
November 1, 2025 at 8:30 AM
ditch the netflix binge this weekend and dive into this instead
I'm surprised how many people don't use Python profilers when it is so easy
November 1, 2025 at 6:15 AM
I'm surprised how many people don't use Python profilers when it is so easy
Just use FP16 and be happy.
October 31, 2025 at 4:36 PM
Just use FP16 and be happy.
When you cancel your @netflix subscription, they show a banner on mobile app, where you can’t click on “No, thanks” 🤦
October 31, 2025 at 8:30 AM
When you cancel your @netflix subscription, they show a banner on mobile app, where you can’t click on “No, thanks” 🤦
🔍 Exploring the diverse world of text similarity algorithms reveals there’s no one-size-fits-all solution; it all depends on your specific use case.
October 31, 2025 at 6:15 AM
🔍 Exploring the diverse world of text similarity algorithms reveals there’s no one-size-fits-all solution; it all depends on your specific use case.