This repo contains a
70% sparse Llama 2 7B finetuned for arithmetic reasoning using the
GSM8k dataset.
Below we share some code snippets on how to get quickly started with running the model.
By leveraging a pre-sparsified model's structure, you can efficiently fine-tune on new data, leading to reduced hyperparameter tuning, training times, and computational costs. Learn about this process
here.
This model may be run with the transformers library. For accelerated inference with sparsity, deploy with
nm-vllm or
deepsparse.
1# pip install transformers accelerate
2from transformers import AutoTokenizer, AutoModelForCausalLM
3
4tokenizer = AutoTokenizer.from_pretrained("neuralmagic/Llama-2-7b-gsm8k-pruned_70")
5model = AutoModelForCausalLM.from_pretrained("neuralmagic/Llama-2-7b-gsm8k-pruned_70", device_map="auto")
6
7input_text = "Natalia sold clips to 48 of her friends in April, and then she sold half as many clips in May. How many clips did Natalia sell altogether in April and May?"
8input_ids = tokenizer.apply_chat_template(input_text, add_generation_prompt=True, return_tensors="pt").to("cuda")
9
10outputs = model.generate(**input_ids)
11print(tokenizer.decode(outputs[0]))
Model evaluation metrics and results.
This model was obtained by sparse-tranfer of the sparse foundational model
Llama-2-7b-pruned70-retrained on the
GSM8k dataset.
Sparse-transfer was performed with
SquareHead knowledge distillation with
Llama-2-7b-gsm8k as teacher.
For further support, and discussions on these models and AI in general, join
Neural Magic's Slack Community
