-
_source: [Preference Tuning LLMs with Direct Preference Optimization Methods_](https://huggingface.co/blog/pref-tuning)_
-
Hello,
I am studying the RAFT-based WEIS (Level 1) program that allows optimization of floating body specifications. I have three questions while reviewing this optimization program:
1. I unders…
-
## *Repository Creation Request*
Use this to get your experiment repository created or updated on GitHub under Virtual Labs organization.
1. #### Coordinating Institute: IIT Bombay
2. #### Appr…
-
It would be great if it could support BERT, LLaMA and other model training.
-
Is this on purpose that these are plural (values, units)?
_Originally posted by @danielhuppmann in https://github.com/iiasa/ixmp4/pull/101#discussion_r1673363628_
As mentioned abov…
-
_Not guaranteed improvement_
Look into general optimization of queries performance of the builders methods which assemble queries from the client size. Not the query itself that will be sent to th…
-
Following the issues with #238 we could experiment with initialisation of `pathfinder`.
I've had success with pre-heating the pre-heating of `pathfinder` but doing a sequence of tempered MAP optimi…
-
-
Once CUDA implementation finished, we should perform NDArray methods review, and remove/hide certain methods, that are either excessive or redundant or misleading.
Also, we should do another pass …
-
@vsavram I took a look at your optimizer notebook. A few comments:
1. I used the data produce by generate_data and tweaked the architecture a bit and steepest descent looks good now! I think steepe…