Explore tweets tagged as #Avb
VLM training repo is now public! 🚨 Contains the code to: - prepare img-text datasets using img2dataset - create VIT + Q-Former image backbone - training Q-Former from a distillbert - train huggingface text LMs on vision tasks - multimodal inference Repo link below 👇🏼
1
0
1
Made this result visualizer for my VLM video (tomorrow) icydk i trained a VIT + Q-Former image encoder to finetune small LMs into vision LMs. This was trained for 4 hrs with only 50k images. Used to be a 100M param text-only model but now it can "see" objects & themes!
0
1
16
Will be finally catching up on Recursive Language Models. I am getting peer-pressured by the DSPy release. May post about my learnings later.
2
2
49
After months of prep and testing my audience's patience, my VLM video is finally here! Goes deep into building VIT-based image encoder pipelines, training Q-Former models, finetuning LLMs on vision tasks, and more. One of my best vids, I am super proud of it. Give it a watch!
1
1
15
Only a handful of youtubers in AI/ML who make regular evergreen content on deeply technical topics. Massive shoutouts. Pictured as examples: @yacinelearning , Welch Labs, @jbhuang0604 @ykilcher Personally, when I choose to make a video, my primary motivation is to satisfy my
8
60
593
Funny how Tim Cotten has gotten compliments for his contributions to ERC8004 from the ETH lead himself Funny how Tim Cotten has a project - $avb (yes comms are bad but are being talked about) - which has been 8004 and x402 before all the hype Funny how the project has never
2
3
3
I sleep like a baby every night
1
0
7
Je repense à ce PATRON qui tenait la barraque absolument TOUT SEUL face aux mêmes médias. Il avait parfaitement identifié les personnes toxiques et avait posé ses c#### sur leur nez. Grand homme, AVB.
20
124
2K
@officialgtfc I made this a few seasons ago. Its been circulated without credit on the net for ages. Im passing it on to the #gtfc fans. Look after him 😪.
0
2
20
Within seconds Quartr came back with Q3 2025 commentary from management teams across the multi-family REIT space. As shown below, management commentary across $EQR, $CPT, $AVB, $ESS, and $MAA points to high quality assets trading at 4.5-5.0% cap rates in the private market.
1
0
2
On a insulté Garcia , on ne voulait plus de lui . On a insulté AVB , on ne voulait plus de lui . On a insulté Sampaoli , on ne voulait plus de lui . On à insulté Tudor , on ne voulait plus de lui . On a insulté RDZ , on ne veut plus de lui . Conclusion. On a jamais rien gagné
77
100
1K
My recommendation engine suggested me to read this random paper about a multi agent llm-system simulating war and peace 🫨 🫨 I'm lowkey excited to read what's this about.
2
1
7