Zain
@ZainHasan6
Followers
5K
Following
10K
Media
1K
Statuses
4K
AI builder & teacher | AI/ML https://t.co/PDkARZxKEc | Eng βΞ¨ @UofT | ex-(Vector DBs, Health tech, Lecturer) | decoding AIβs future - follow for insights! π¨π¦π΅π°
token factory
Joined August 2012
I'm hosting our first https://t.co/kWsq4hb8PC event in SF on Monday 09/22 at Shack15. I'm joined by @ZainHasan6 from Together AI! - Autopsy of RAG: What weβre killing & what we're keeping - The Agentic Shift: LLMs are shockingly good at search https://t.co/eviqJjpSLq
luma.com
R.I.P. Naive RAG - Welcome Agentic Retrieval Hosted by: HORNET.dev The era of single-turn, "chunk-and-embed" RAG is over. The future lies in agentic retrieval.β¦
2
12
64
I know what I'm doing this weekend! New paper from OpenAI. "Why LLMs Hallucinate"
1
1
2
We have Claude-Sonnet-4 at home now chat!?
π Kimi K2-0905 just landed on Together AI! Enhanced coding capabilities + 256K context (doubled!). Same open-source power from @Kimi_Moonshot that beats proprietary models. Built for agents. Priced for scale. π
0
0
0
ππ... "guaranteed 100% tool-call accuracy" 1) What?
Kimi K2-0905 update π - Enhanced coding capabilities, esp. front-end & tool-calling - Context length extended to 256k tokens - Improved integration with various agent scaffolds (e.g., Claude Code, Roo Code, etc) π Weights & code: https://t.co/83sQekosr9 π¬ Chat with new Kimi
3
0
2
First year me: βWhen will I ever use matrix multiplication in real life?β Present me:
0
0
1
Just completed the EXCELLENT Coursera course: Retrieval Augmented Generation, led by @ZainHasan6. https://t.co/ipidYFboQm By far, the best survey of the topic to date. The detailed section about what to consider in deploying RAG solutions to Production was invaluable.
coursera.org
Offered by DeepLearning.AI. Retrieval Augmented ... Enroll for free.
1
2
5
chinese doordash just dropped a 560B model π
what did you do with your sunday?
0
1
3
40+ notebooks on everything from fine-tuning to agents and evals! Check it out!
β We just hit 1K stars on the Together AI Cookbook repo! A collection of code examples for building with open-source models: β AI Agents & workflows β Fine-tuning guides (LoRA, DPO) β RAG implementations β LLM evaluations β Multimodal examples All with 1-click Colab
0
0
6
The origin story of the world's most beloved programming language: Python.
1
0
1
πthere's a new king sitting atop the multi-turn function calling benchmark and it's OPEN SOURCE! GLM4.5 and 4.5-Air both beat Sonnet4.1 and Opus4 on multi-turn FC pretty big for coding agent apps like claude code, cursor etc.
0
0
4
skimming through the InternVL3.5 report and this is now the best open VLM available (it's a Qwen3-235B fine-tune + 5B vision encoder) BUT the main takeaway for me = look at how close GLM-4.5V is and at less than HALF the number of total and active params!π€―
1
0
3