Álvaro Tejeda

- Finance, Data Science, AI, Recommender systems, Gaming, iGaming..

[Álvaro Tejeda @ATL_85]

@yuriymatso How is your performance during the period Oct-25 to dec-25? Is where I find most of my models taking the bigger DD..

[Álvaro Tejeda @ATL_85]

Free course: Forecasting for Data Scientists by @CBergmeir.
Strong foundations and very practical. Worth checking out.
https://t.co/3jUSVCevqj

[Álvaro Tejeda @ATL_85]

@MapleStax Trade

[Álvaro Tejeda @ATL_85]

@SRxTrades checklist

[vLLM @vllm_project]

🚀 DeepSeek-OCR — the new frontier of OCR from @deepseek_ai , exploring optical context compression for LLMs, is running blazingly fast on vLLM ⚡ (~2500 tokens/s on A100-40G) — powered by vllm==0.8.5 for day-0 model support.

🧠 Compresses visual contexts up to 20× while keeping 97% OCR accuracy at <10×.
📄 Outperforms GOT-OCR2.0 & MinerU2.0 on OmniDocBench using fewer vision tokens.

🤝 The vLLM team is working with DeepSeek to bring official DeepSeek-OCR support into the next vLLM release — making multimodal inference even faster and easier to scale.

🔗 https://t.co/rnBG9VUuMy
#vLLM #DeepSeek #OCR #LLM #VisionAI #DeepLearning

[Álvaro Tejeda @ATL_85]

Lo ridículamente mal organizado que está el @coolturalfest de Almería es épico, cola de dos horas para entrar, hasta nunca! #nuncamas

[Sumanth @Sumanth_077]

Stanford CS336: Large Language Models from Scratch!

This is a comprehensive course on LLMs, covers the full process of building one from scratch, including data collection, pretraining, transformer architecture, training, evaluation, and deployment.

[Dorsa @dorsa_rohani]

New fastest shortest-path algorithm in 41 years!
Tsinghua researchers broke Dijkstra’s 1984 “sorting barrier,” achieving O(m log^(2/3) n) time. This means faster route planning, less traffic, cheaper deliveries, and more efficient networks - and a CS curriculum revamp =)

[Joseph Suarez 🐡 @jsuarez]

https://t.co/DEMnbqPmtw

[Mayank Pratap Singh @Mayank_022]

I trained a 100 million parameter DeepSeek V3 LLM from scratch
Here's what you need to know.

Previously I trained traditional GPT-2 architecture which has become obsolete with recent LLM advancements. Most recent models like Llama, Mistral, DeepSeek, and GPT-4 use latest architectures.

✦  Model Configuration of my SLM

DeepSeek V3
- Parameters: 109,032,032
- Embedding Dimension: 512
- Layers: 8
- Heads: 8
- Experts (MoE): 8
- Experts per token: 2

✦ DeepSeek brings major architectural changes:
- Multi Head Latent Attention
- Mixture of Experts
- RMS Norm
- Multi Token Prediction

✦ Dataset Challenge
- TinyStories is great for learning SLMs. I trained GPT-2 on it previously with good results.
- But I needed a more challenging dataset.
- If I use TinyStories again on DeepSeek, how would I know MHLA, MoE or MTP works better than old architecture?
- The old architecture can handle it, so new DeepSeek would too without utilizing latest advancements.

That's why I moved to FineWeb-Edu dataset
Thanks @YuvrajS9886 for the suggestion for this dataset

✦ Training Journey
- Rented A100 PCIe GPU and trained the model.
- Did test runs. During final run, model was 65% trained but stopped due to glitch after 4 hours.
- Fixed all edge cases and ran training again with increased config parameters.
- Final training: 7 hours, 20,000 epochs

𝐓𝐨𝐭𝐚𝐥 𝐆𝐏𝐔 𝐜𝐨𝐬𝐭: $17
- $9.53 for main 7-hour run
- $7.42 for experiments and demos

✦ Reflection
Amazing long project that taught me latest architectural advancements.
I'll reimplement and revisit after a few weeks because there's too much complexity, mostly in Multi Head Latent Attention part. Need to make concepts stronger.

Code
https://t.co/9HHdTUhJT0

Final trained Model
https://t.co/FediD7hDWE

Dataset
https://t.co/XkbxsCoe6F

Resources
Huge shoutout to @raj_dandekar again for creating one of the most detailed video series about DeepSeek
- this was my primary resource for the implementation.

Playlist
https://t.co/89VKpyhgUe

Blogs by @MaartenGr
These are excellent visual blogs to understand MoE in detail. Thanks Maarten for your amazing contributions to the community through your books and blogs
https://t.co/kxKj4zrU5g

Blogs on MoE
https://t.co/mWT5tYkhZB

Implemention of MoE from scratch by @aviTwit3

https://t.co/Bd5VzCnjXZ

One of the most detailed blogs on implementing Mixture of Experts.
Thanks Avinash for this blog

- it helped me understand Mixture of Experts much better.

If you're someone in the 𝐌𝐋 & 𝐋𝐋𝐌 space, would love to 𝐜𝐨𝐧𝐧𝐞𝐜𝐭 and discuss this field in general, so give a follow up for that.

[𝗿𝗮𝗺𝗮𝗸𝗿𝘂𝘀𝗵𝗻𝗮— 𝗲/𝗮𝗰𝗰 @techwith_ram]

Top 50 LLM Interview Questions :

[Deedy @deedydas]

DeepSeek just dropped the single best end-to-end paper on large model training.

It covers
— Software (MLA, training in FP8, DeepEP, LogFMT)
— Hardware (Multi-Rail Fat Tree, Ethernet RoCE switches)
— Mix (IBGDA, 3FS filesystem)

DeepSeek's engineering depth is insane. Must read.

[elvis @omarsar0]

AI Agents vs. Agentic AI

Interesting paper summarizing distinctions between AI Agents and Agentic AI.

It also talks about the key ideas, solutions, and the future.

Here are my notes:

[Tom Yeh @ProfTomYeh]

Autoencoder by hand✍️Excel~ I designed this exercise to show how an Encoder-Decoder network convert input to code and reconstruct input from code. It is annotated with equations, PyTorch, and graphs. 👇Join the 'AI Math' community. Download xlsx.

[Álvaro Tejeda @ATL_85]

Awesome event by @AiBirras ! The AI future in Granada is brighter than ever! Listening to experts such as @draxus is always a pleasure!

[Álvaro Tejeda @ATL_85]

A must-see: Stanford Machine Learning Course: CS229

https://t.co/SXmRFcU1w6

#ai #stanford #llm #genai

[Máster DATCOM UGR @DatcoMugr]

Me complace presentar a mi buen compañero y amigo @ATL_85 para impartir una ponencia sobre la carrera profesional en Ciencia e Ingeniera del Dato, consejos para que nuestros estudiantes progresen en el mundo del business intelligence

[Markus J. Buehler @ProfBuehlerMIT]

We trained a graph-native AI, then let it reason for days, forming a dynamic relational world model on its own - no pre-programming. Emergent hubs, small-world properties, modularity, & scale-free structures arose naturally. The model then exploited compositional reasoning & uncovered uncoded properties from deep synthesis: Materials with memory, microbial repair, self-evolving systems. Video shows it unfolding, made with @grok @xai.

[John Rush @johnrushx]

I've tried all (24) AI coding agents &amp; IDEs 😵‍💫

[Cursor, Softgen, Windsurf, Wrapifai, Copilot, Lovable, Bolt, v0, Replit, MarsX, Claude, AmazonQ, Pear, Devin, Github Spark, IDX, Webdraw, Tempo, Cline, Continue, Databutton, Base44, Qodo, Aider]

The Vibe Coding giga-thread:

[Operador Nuclear @OperadorNuclear]

Salvemos la central nuclear de Almaraz. Manifestación el sábado 18 de enero a las 10:00 que saldrá desde el ayuntamiento de Almaraz. @SiAlmaraz