Abed Khooli retweeted
You heard about LLM inference on WebGPU, but what about... finetuning LLM on WebGPU? 🤯 I put together a super earlier PoC that proves it's possible, backed by llama.cpp / wllama 😁 Working on LoRA next...
15
27
2
190
20,218
Abed Khooli retweeted
Pi 0.84.4 is out🥳 3 updates we’re excited for you to try:    - Compaction now runs between large tool results and the next reply, in the same run  - DeepSeek V4 Flash Vision  - Terminal overrides for hyperlinks, images and true color Run pi update for the latest
33
29
7
774
62,046
Open-source construction ERP with CAD-BIM takeoff, BOQ estimation and 180+ modules OpenConstructionERP v15 github.com/datadrivenconstru…
2
2
4
266
Abed Khooli retweeted
I present to you an Aug 2026 massively revised Machine Learning Engineering open book github.com/stas00/ml-enginee… If you prefer the pdf it's now 498 page long github.com/stas00/ml-enginee… The book went through a massive update to bring all the high end hw specs to being up-to-date, and updated many examples and studies to the recent torch / cuda and other critical in ML components. Many bugs and inaccuracies have been fixed. If you read the book and find any remaining please kindly flag them to me.
10
90
5
551
27,996
Abed Khooli retweeted
We use Inkling-Small to turn paper abstracts into quick, useful summaries. open weights × open science 🤝
85
64
8
409
172,538
Abed Khooli retweeted
This Robot scans books and high speed and no need to destroy the spine, no broken bindings. The Treventus ScanRobot places an open book in a 60-degree V-cradle and uses airflow to turn pages while scanning both sides at once.
18
15
4
112
15,338
Abed Khooli retweeted
JUST IN: Alibaba unveils Qwen3.8-Max — a 2.4 trillion-parameter AI model it claims rivals Anthropic’s Fable on some benchmarks.
97
115
13
2,064
154,606
Abed Khooli retweeted
KimikK3's numbers on this feels unreal. ~1/3 the price of Fable 5 and and still beating it on Frontend Code Arena - kimi k3 — $3 / $15 - claude fable 5 — $10 / $50 - gpt-5.6 sol — $5 / $30
Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5. This is a 17-place jump from Kimi-k2.6 (#18 -> #1). In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics, Consumer Product, Simulations, and Content Creation Tools, landing #2 only in Gaming behind Fable 5. The full model weights will be released by July 27. Congrats to the @Kimi_Moonshot team on this major milestone!
11
11
3
40
8,549
Abed Khooli retweeted
I couldn't resist extracting that Rust Mermaid rendering code out and compiling it to WebAssembly so you can try it out directly in a browser tools.simonwillison.net/grok…
14
16
5
252
23,877
Abed Khooli retweeted
Today, we're launching Wamda's H1 2026 MENA Investment Report. Despite one of the most turbulent periods the region has experienced in recent years, MENA startups raised $1.7 billion across 242 deals during the first six months of 2026. A few takeaways from the country breakdown: 🇦🇪 The UAE dominated the landscape, attracting $1.2 billion across 83 deals, accounting for roughly 70% of all capital deployed in the region. 🇸🇦 Saudi Arabia remained one of the region's most active ecosystems by deal count, closing 80 transactions. 🇪🇬 Egypt maintained third place with 29 deals despite a challenging fundraising environment. 📍 Beyond the three largest markets, countries including Oman, Morocco, Bahrain, Jordan and Qatar continued to generate investment activity, reflecting the growing geographical diversity of MENA's startup ecosystem. 📊 Read the full report on Wamda: wamda.com/en/2026/07/mena-st… #MENA #Startups #VentureCapital #VC #Investment #Fintech #UAE #SaudiArabia #Egypt #Wamda
1
6
4
1,201
Abed Khooli retweeted
Introducing Seraj, an enterprise Al model purpose-built to advance Arabic-language Al with stronger reasoning, greater linguistic accuracy, and deeper cultural understanding. Developed by @Inception_42 and @Microsoft, Seraj is now available through Core42 Compass, enabling government and enterprise organizations to deploy high-performance Arabic Al with the scalability, security, and reliability required for production. Explore Core42 Compass: eu1.hubs.ly/H0wGhGF0 #Core42 #GenAl #Artificiallntelligence #EnterpriseAl #Inception #Microsoft
1
3
8
331
Abed Khooli retweeted
vLLM v0.24.0 is out! 571 commits from 256 contributors (77 new). 🎉 Highlights: MiniMax-M3 support (FP8/MXFP4 + broad AMD tuning), DeepSeek-V4 keeps maturing (FlashInfer sparse index cache, prefill chunk-planning, now on SM120), Model Runner V2 now handles quantized models by default, a new unified Streaming Parser Engine for tool-calls + reasoning, DiffusionGemma, DeepEP v2 for wide expert parallelism, and a maturing Rust frontend. Thread 👇
12
50
9
512
52,709
Abed Khooli retweeted
After 18 months of writing, coding, and experimenting, Build a Reasoning Model (From Scratch) is finally out! My first copies just arrived! 📚 440 full-color pages. Inference scaling, reinforcement learning, and distillation from scratch.
281
553
53
5,937
535,756
Abed Khooli retweeted
hello I'm Merve 👋🏻 I want every single developer to build computer vision applications with agents using open models, but possibilities are endless they get lost here to change it, come to my talk at @aiDotEngineer WF tomorrow to see my new project, here's a sneak peek
16
34
4
407
42,406
MAI-image-2.5 is new at #2 for text-to-image and #3 for image editing on @ArtificialAnlys! We’re second now only to GPT models. Also, MAI-Image-2.5-Flash is world-beating for quality/price. Super proud of the team. Now available through the Foundry API and rolling out across OneDrive + PowerPoint, or check it out in the MAI Playground: playground.microsoft.ai/chat
11
13
1
119
18,685
Abed Khooli retweeted
Fable gone dark. GPT-5.6 delayed. Gemini 3.5 Pro stalled. Is it a coincidence? Is the USG holding back frontier model releases? @deanwball attributed the Fable export control decision to three factors on @CogRev_Podcast: 1. Genuine concern about safety and security 2. Broad lack of context on frontier AI amongst admin leadership 3. A dimension of conflicting political views Is the Trump admin getting cold feet about AI acceleration? Doesn't DJT know how great the "AGI President" label would be for his legacy? I'm not a fan of humanity's most consequential technological achievement happening in the dark.
🚨 SCOOP(s): - GPT-5.6 has been delayed and will no longer release this week. New target is ~mid-July. - DeepMind are not satisfied with the current state of 3.5 Pro and it will no longer launch this month. - Preparations for the launch of Bidi, OpenAI's new voice model, are underway in ChatGPT and we could see it available as soon as this week. - Claude Sonnet 5 is currently available for select enterprise customers under an Early Access Program and is seen as a stop-gap as progress on getting Mythos/Fable 5 back out have stalled. A bit of a disappointing end to the month, but July should prove more fruitful!
13
4
2
82
7,450
Abed Khooli retweeted
Introducing the Fusion API, the smartest compound model in the market. Fusion achieves Fable-level intelligence at half the price. How it works 👇
716
1,754
1,279
14,954
6,438,245
Abed Khooli retweeted
Releasing an answer equivalence reward dataset. ~20K research QA reference + predicted answers with quality scores for each. Contains some tough confounding examples that easily fool F1 and BertScore evals. Produced using Deepseek, weak models, and Qwen3.6-35B-A3B.
1
3
1
22
2,919