xinyuema/llm-course-hw3-tinyllama-qlora-model

Model Card for Model ID

Model Details

The 4-bit of the base model TinyLlama/TinyLlama-1.1B-Chat-v1.0 was fine-tuned on a tweet sentiment dataset cardiffnlp/tweet_eval in order to determine tweets tonality by positive, neutral or negative.

Model Description

We used a system prompt to instruct the model:

SYSTEM PROMPT:

You are a tweet sentiment classifier. For each tweet input, analyze its sentiment and output exactly one word: "negative", "neutral", or "positive". Do not include any extra text.

But the model is not trained to return only the sentiment name.

The PEFT method applied was QLoRA, the optimizer parameters be offloaded to CPU, when GPU memory is limited but could slow down the learning.

Training Details

The peft framework was applied to the model, k_proj and v_proj layers whas modified with LoRA linear layer rank=8, alpha=16.

This model was trained with SFTTrainer of trl, batch_size=12, gradient_accumulation_steps=128, learning_rate=5e-4, lr_scheduler_type="linear"

The model achieved 0.19 macro f1-score on the test dataset(initial model 0.03)