cardiffnlp/tweet_eval
Viewer • Updated • 201k • 39.3k • 144
How to use mcnckc/llm-course-hw3-tinyllama-qlora with Transformers:
# Use a pipeline as a high-level helper
from transformers import pipeline
pipe = pipeline("text-classification", model="mcnckc/llm-course-hw3-tinyllama-qlora") # Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("mcnckc/llm-course-hw3-tinyllama-qlora", dtype="auto")Модель TinyLlama/TinyLlama-1.1B-Chat-v1.0 дообученная на датасете cardiffnlp/tweet_eval, задача классификации сентимента твита, вывести одно из трех слов -
negative, neutral, positive.
Модель дообучалась при помощи QLoRA.
8load_in_4bit=Truebnb_4bit_quant_type="nf4"bnb_4bit_compute_dtype=torch.float16lora_alpha=4lora_dropout=0.0bias="none"task_type="CAUSAL_LM"num_train_epochs=1per_device_train_batch_size=32gradient_accumulation_steps=1warmup_steps=100group_by_length=Truelr_scheduler_type="cosine"fp16=Truelearning_rate=2e-4weight_decay=1e-3F1=0.18
Base model
TinyLlama/TinyLlama-1.1B-Chat-v1.0