| license: mit | |
| this is a model made by me on a 5090 (rented). its trained by scratch | |
| its a 112m prameter model trained on huggingfaceFW on 200k rows (i could do more if i want to) and 15k rows on dolly-15k. | |
| there will be a 400m prameter model trained on 2m rows on huggingfaceFW soon maybe once i get the money. |