This model was created using OpenHermes-2.5 as the base, and finetuned with argilla/distilabel-intel-orca-dpo-pairs. Finetuned with these specific parameters: Steps: 200 Learning Rate: 5e5 Beta: 0.1 — Developed by: @decruz — Funded by [optional]: my full-time job — Finetuned from model [optional]: teknium/OpenHermes-2.5-Mistral-7B You can use this for basic inference. You could probably finetune with this if you want to. You can create a space out of this, or use basic python code to call the model directly and make inferences to it. The following was used: trainingargs = TrainingArguments( perdevicetrainbatchsize=4, gradientaccumulationsteps=4, gradientcheckpointing=True, learningrate=5e-5, lrschedulertype=»cosine», maxsteps=200, savestrategy=»no», loggingsteps=1, outputdir=newmodel, optim=»pagedadamw32bit», warmupsteps=100, bf16=True, report_to=»wandb», ) dpotrainer = DPOTrainer( model, refmodel, args=trainingargs, traindataset=dataset, tokenizer=tokenizer, peftconfig=peftconfig, beta=0.1, maxpromptleng
Модальности:
Генерация текста
Области применения:
Диалог / чат
Задача: Генерация текста
Автор: decruz07
Теги: mistral, conversational, text-generation-inference, endpoints_compatible
Лайков: 3 | Загрузок: 31
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.