Ayush Kaushal

I would like to work on this issue. I have read and understood the paper and related concepts. I have done a crude implementation of paper for solving elliptic PDEs...

> Yes, let's make sure it converges "all of the way" first. It still looks like it's missing some of the corner behavior. I trained it till converged and seems...

> instead of writing down the discretisation, it should just call DifferentialEquations.jl. I have a doubt here. For PDEs involving time, there are two methods mentioned in the paper. First...

ping @ChrisRackauckas . ^

Sure. I am proceeding with Algorithm 3 then.

I have implemented the Algorithm 3 as in the paper ([link](https://github.com/Ayushk4/WAN_PDE/blob/master/time_dependent_pde.jl)). I have uploaded plots (also separate images) for the [same](https://github.com/Ayushk4/WAN_PDE/blob/master/Time_Dependent_pdes/Plots-TimeDependentPDEs.ipynb). It seems to have trained well, but took somewhat...

> I am interested in investigating and improving the sentence tokenizers part of WordTokenizers.jl. Would that be of interest to you if I work on a PR regarding this? Sure....

I think that the Tokenizer API should also be able to expose the TokenBuffer API and its various lexer functions for building custom tokenizers.

https://twitter.com/NolanoOrg/status/1635409631530057728 Q4_0 mode will be worse with GPTQ for 7B than current Round-to-nearest quantization approach. However, in other cases it's better (only tested upto 13B models). In Q4_1 and 13B...

Also, the above has results on both Q4_0 and Q4_1. GPTQ can be implemented for Q4_0 by hardcoding zero-offset at https://github.com/qwopqwop200/GPTQ-for-LLaMa/blob/9bbfdeda2c80c20d8bc1babf4d4368df574afe30/quant.py#L6