QLoRA (Efficient Finetuning of Quantized LLMs) for sentiment analysis task

Описание задания

В этой домашке была дообучена языковая модель TinyLlama-1.1B-Chat-v1.0 с помощью QLoRA на датасете cardiffnlp/tweet_eval для задачи анализа тональности текстов

Пример генерации

Вопрос

@user Alciato: Bee will invest 150 million in January, another 200 in the Summer and plans to bring Messi by 2017

Ответ модели

positive

Качество на тестовой выборке

F1 macro: 0.34

image/png

Пример запуска

from transformers import AutoModelForCausalLM, AutoTokenizer

REPO_NAME = "MurDanya/llm-course-hw3-tinyllama-qlora"

model = AutoModelForCausalLM.from_pretrained(REPO_NAME, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(REPO_NAME)
tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "left"
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for MurDanya/llm-course-hw3-tinyllama-qlora

Finetuned
(440)
this model

Dataset used to train MurDanya/llm-course-hw3-tinyllama-qlora

Collection including MurDanya/llm-course-hw3-tinyllama-qlora