Many GenAI models are based on Markov processes. By parameterizing them w/"generators," they can:
1. Unify & universally characterize Markov models
2. Train them at scale
3. Combine them to build multimodal models or superpositions
4. Train new models, e.g. jump models
LLMS
-
Markov Processes Parameterization Unifies Multimodal GenAI Models
By
–
-
Team of Researchers Reviews Feedback to Improve Developer Models
By
–
We have an amazing team of researchers and folks like myself who pour over these responses to make our next models even better for devs – trust me, we read all of it!
-
OpenCoder: Open Cookbook for Top-Tier Code Language Models
By
–
OpenCoder The Open Cookbook for Top-Tier Code Large Language Models
-
Gemini 2.0 Pro Availability on LMSYS Leaderboard
By
–
Gemini 2.0 Pro already on LMSYS? Someone hopefully can confirm
-
LLM Coding Resources and Bonus Materials for Weekend Learning
By
–
If you are looking for something to read/study this weekend, I added lots of LLM-related bonus from-scratch coding resources over the last few months (from implementing Llama 3.2 to preference tuning with DPO): https://
github.com/rasbt/LLMs-fro
m-scratch?tab=readme-ov-file#bonus-material
… I hope you find them useful! -
DPO vs PPO: Lab Training Method Preferences Revealed
By
–
I'd say it's traditional at this point. I would expect them to use something more RL like PPO but it's interesting they chose DPO instead. It also shows that, despite all the *PO papers, most labs still use DPO variants.
-
Gemini 2.0 Pro Release Strategy and Expected Performance Jump
By
–
We will finally see Gemini 2.0 Pro soon, long overdue. But they will probably wait for the release of full o1 to steal the show from OpenAI, so to speak – just as OpenAI has done to Google every time. I expect a significant jump from 1.5 to 2.0, because Google itself also knows
-
Gemini 2.0 Pro Release Imminent with Sam Altman Reference
By
–
There are increasing indications that the release of Gemini 2.0 Pro is imminent. With the reference “AI is cool I guess” Logan is making a direct allusion to Sam Altman. So far, Google has not attracted much attention in the high-ELO league with Gemini. Although they officially
-
Applications Beyond LLM Fine-tuning Techniques
By
–
P.S. and its not just limited to LLM finetuning 😀
-
AndroidLab benchmark shows small fine-tuned models can power JARVIS
By
–
> AndroidLab: First ever systematic benchmark for Android mobile agents shows that small, fine-tuned open models can power a JARVIS system on your smartphone A team from @Tsinghua_Uni just released AndroidLab, the first systematic framework to evaluate and train Android