Our blueprint to cut LLM costs by 88%: https://
decodingml.substack.com/p/our-blueprin
t-to-cut-llm-costs-by
… The advanced RAG course (use code paul_15 for 15% off):
@whats_ai
-
Blueprint to Cut LLM Costs by 88 Percent
By
–
-
Fine-tuned GPT-4o Mini cuts costs by 88 percent monthly
By
–
This fine-tuned GPT-4o Mini runs at just ~$180/month—saving us $1,320 monthly- and has comparable results to the original model. Want to dive deeper? Rerad the full article and check out the “From Beginner to Advanced LLM Developer Course” with 50+ hours of content—and grab a
-
Cost-Effective Fine-Tuning of GPT-4o Mini Models
By
–
2. Fine-Tuning a Cost-Effective Model: Instead of deploying a full-scale GPT-4o (costing ~$1,500/month handling 10k queries monthly), we fine-tuned a smaller, more efficient GPT-4o Mini. By using OpenAI’s managed fine-tuning on a curated dataset of 100 Q&A pairs (with 30 for
-
Fine-tuned GPT-4o Mini cuts costs by 88%
By
–
This fine-tuned GPT-4o Mini runs at just ~$180/month—saving us $1,320 monthly- and has comparable results to the original model. Want to dive deeper? Rerad the full article and check out the “From Beginner to Advanced LLM Developer Course” with 50+ hours of content—and grab a
-
RAG Fine-tuning Hybrid Reduces AI Tutor Costs by 88%
By
–
Excited to share insights from our recent (first ever) guest post on Decoding ML by @iusztinpaul ! We wrote about how a proper RAG + fine-tuning (hybrid) setup can reduce expenses by 88% on our own AI tutor :
-
How LLMs Handle Language: Errors, Power, and Practical Applications
By
–
Let's talk about how LLMs handle language, and specifically: Why they sometimes get things wrong, and why they’re still incredibly powerful tools for building next-generation applications (when used right and with the right additions). https://
open.substack.com/pub/louisbouch
ard/p/how-chatgpt-actually-works-no-math?r=25qlky&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
… -
How LLMs Work: Training Pipeline and Limitations Explained
By
–
AI Explained For Complete Beginners – No Math! A new video where we tried to cover how LLMs work, simply, without math or code, focusing on the whole generation and training pipeline to understand better their limitations (why they happen and how we can mitigate them). The
-
Jensen’s Quantum Panel: 14 Industry Leaders Discuss Quantum Computing
By
–
Here’s everything you need to know about the “coolest” event of GTC: Jensen’s Quantum panel with 14 leaders in the field…
— Louis-François Bouchard 🎥🤖 (@Whats_AI) 21 mars 2025
Pun intended with “coolest” (quantum computers literally need to be cooled near absolute zero!).
If, like me, you find quantum computing a bit abstract and… pic.twitter.com/71FZH0FKEbHere’s everything you need to know about the “coolest” event of GTC: Jensen’s Quantum panel with 14 leaders in the field… Pun intended with “coolest” (quantum computers literally need to be cooled near absolute zero!). If, like me, you find quantum computing a bit abstract and
-
GTC 2025 Conference Highlights with Leading AI Experts
By
–
GTC 2024 was my highlight of the year. Well, even if the “feeling of new” wasn’t there, GTC 2025 didn’t disappoint either!
— Louis-François Bouchard 🎥🤖 (@Whats_AI) 20 mars 2025
All of my AI stars were there, starting with @twominutepapers , @chipro , @eugeneyalt , @FanaHOVA , @swyx , Yann LeCun, obviously Jensen, and tons of… pic.twitter.com/qXy71AzpMBGTC 2024 was my highlight of the year. Well, even if the “feeling of new” wasn’t there, GTC 2025 didn’t disappoint either! All of my AI stars were there, starting with @twominutepapers , @chipro , @eugeneyalt , @FanaHOVA , @swyx , Yann LeCun, obviously Jensen, and tons of
-
FlashMLA: Cutting KV Cache Memory in LLMs
By
–
This week, we explore what FlashMLA is, why it matters, and how it builds on key innovations like Mixture of Experts, infinite context attention, and context caching! https://
louisbouchard.substack.com/p/how-flashmla
-cuts-kv-cache-memory
…