efromomr/llm-course-hw3-tinyllama-qlora

Model Card for Model ID

Model Details

Model Description

TinyLlama/TinyLlama-1.1B-Chat-v1.0 finetuned on cardiffnlp/tweet_eval for sentiment-analysis task with trl's SFTTrainer.

How to Get Started with the Model

Use the code below to get started with the model.

model = AutoModelForCausalLM.from_pretrained(f"efromomr/llm-course-hw3-tinyllama-qlora", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(f"efromomr/llm-course-hw3-tinyllama-qlora")
tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "left"

input_ids = tokenizer(text, return_tensors="pt").input_ids

output_ids = model.generate(input_ids, max_new_tokens=16)
generated_text = tokenizer.decode(output_ids[0][len(input_ids[0]) :], skip_special_tokens=True)
print(generated_text)
#positive

Training Details

Training Data

cardiffnlp/tweet_eval

Evaluation

Testing Data, Factors & Metrics

Metrics

F1: 0.46 on test set