umar ibrahim

16.7K posts

umar ibrahim banner
umar ibrahim

umar ibrahim

@imp213x

Perfectly imperfect!!! Toxic as hell| I bite with words!| my DM is open for a reason, don't get my replies twisted, you'd be disappointed.

kaduna Katılım Ağustos 2012
3.1K Takip Edilen6.3K Takipçiler
umar ibrahim
umar ibrahim@imp213x·
Yusuf, I'm awed about people's knowledge as I am about their sincerity in applying it. You touched a core that many a Muslim have been blind to. Even worse, our scholars have deliberately refused to explain properly. You succinctly captured apostasy. Thank you!!!
A.Y.O@YusufAsunmogejo

Hello Prince Adewale @Maya_leeke , I have been watching your tweets and your attacks on our faith lately on this app. Remember you had once tried to engage me by stalking me, and I used to ignore you. I did that deliberately. For the sake of people who may want to be deceived, let me make some explanations as a rebuttal. Prince Adewale is pushing a surface level narrative about Sharia without understanding how classical Islamic practice works. In Islamic law, the ruling on apostasy was never about someone changing their personal beliefs in the privacy of their bedroom. It was the historical equivalent of high treason. In the early days, Islamic societies were under constant military threat. Leaving the faith almost always meant defecting to the enemy army, leaking state secrets, and taking up arms against your own people. The penalty was for political and military betrayal. Even in the ancient world, religion and the state were tied together. If you lived in the Roman Empire or the Persian Empire and defected to their enemies, the penalty was death. In fact, in our Nigerian law, what happens if a Nigerian citizen renounces their allegiance to the country, crosses over to join an armed insurgency, leaks military intelligence, and starts fighting the Nigerian state? Can Nigeria accept that in the name of freedom of association? No. The constitution defines that as treason, and the penalty is death. The United States and every other nation on earth have severe penalties for treason and armed rebellion. In early Islamic history, the Muslim community was a new state fighting for its survival. So leaving the faith was a public defection to the hostile enemy camp. If it was just about changing religion, we would see a very different history. We have authentic records where people left Islam during the time of the Prophet Muhammad (peace be upon him), and they walked away freely. You can read the famous Hadith narrated by Jabir bin Abdullah in Sahih al-Bukhari, found in the Book of Virtues of Madinah. An unnamed Bedouin came to the Prophet (peace be upon him), he openly canceled his pledge of Islam, and walked right out of the city. Nobody touched him. Read that again. You cannot strip away the historical and political context of a law just to push a fearmongering agenda on Twitter. Prince Adewale, Be Warned.

English
0
0
0
91
umar ibrahim
umar ibrahim@imp213x·
This has been my workflow for about a month now. Designing a tokenizer already and so far I'm head-to-head with BPE. I hope to come up with something novel.
Andrej Karpathy@karpathy

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

English
0
0
0
17
umar ibrahim
umar ibrahim@imp213x·
Nothing I detest like religiosity in politics. They both can never mix. It is impossible to mix. It is like a Halocline, one part always comes with dirt and dense saline contaminants. Wallahi ko ibn Taymiyya sai ta gagareshi. Your propesor is only performing 😭
English
0
0
0
40
umar ibrahim
umar ibrahim@imp213x·
@el_bonga @ProfIsaPantami But we all know this is impossible in a political space. There is nowhere, two parties, contesting for the same thing, in this case power and access, can unite. It is a pipe dream. I feel bad for what the current situation portends for Arewa, but Pantami's advice is impossible.
English
0
0
0
2K
Mr. El-Bonga
Mr. El-Bonga@el_bonga·
During El-Rufai’s mother’s funeral, @ProfIsaPantami used a few minutes at the central mosque to call for unity by speaking against enmity & malignant politics creating division. All the politicians involved were there.He stated that what unites us is greater than what divides us.
English
23
270
999
69.6K
umar ibrahim
umar ibrahim@imp213x·
When I saw that announcement, I knew instantly it was targeted at Nigerians, and I don't think it is wrong. You only need to see them in comment sections on issues they know nothing about, and you'd wish X to simply just ban us. We de over do.
Nikita Bier@nikitabier

@aaronp613 Precisely. Fewer Ivanka Trump Fan accounts based in Nigeria—and more Nigerians sharing their thoughts about Nigeria.

English
0
0
0
36
umar ibrahim
umar ibrahim@imp213x·
They are roasting Nigerians under this tweet, and for good reasons! For God's sake we are not the only ones that know about X's monetization or AI use. Even worse are those MAGA ones who've turned to arse lickers for America over the current war in the middle East.
Dan@RikkyRichardso1

@CollinRugg Cannot wait for regional filters. 3 Nigerian AI responses in a row

English
0
0
0
48
umar ibrahim
umar ibrahim@imp213x·
Again, I am totally open to partnership with anyone on these 2 projects. I made this too long. Not my plan to make you read this long. Just wanted to explain a little. Lastly I'll appreciate it if you can create an account and let me Demo the both projects. Thank you! 🙏
English
0
0
0
12
umar ibrahim
umar ibrahim@imp213x·
This is also to ask those who'd be interested in sharing this idea with me to indicate so I can share the paper with them on the AI data & memory framework I am building that has enhanced this native Model build and training on a non-CUDA device.
English
1
0
0
11
umar ibrahim
umar ibrahim@imp213x·
Alhamdulillah ala ni'imatil Islam and Eid Mubarak to everyone! To @Abdulrahmanleme , your place in this world and the hereafter will be lined with the highest of Allah's blessings. And may Allah protect you from the fitnah in the Nigerian political space. .....Amin.
English
0
0
1
43
umar ibrahim
umar ibrahim@imp213x·
The fact that Gemini beats Opus and GPT 5.4 in instruction following is a clear reason to call this benchmark bullshit! Of all of them the worst at following instructions is all of the Gemini models.
X Freeze@XFreeze

The new Grok 4.20 Beta benchmarks are wild 🥇 #1 lowest hallucinating AI (22%) 🥇 #1 at following instructions (83%) 🥈 #2 in agentic tool use (97%) Grok 4.20 ranks #1 in the lowest hallucination rate ever recorded across all AI models tested globally Most models race to sound smart. Grok 4.20 was built to never lie and still dominates on instruction following and agentic tasks This is literally a 500B model performing top-notch in the things that matter most

English
0
0
0
41
umar ibrahim
umar ibrahim@imp213x·
What you should ask is: How did an unlettered man know all of these before ever coming in contact with or knowing to read any of the languages they were written in? That alone is enough for anyone who has brains to think.
Ekumeku@MrBrown_114

@SALHACHIMI The Quran is like reading a sub text that developed from other existing texts as the original text.

English
0
0
0
17