--- language: - en - de - fr - it - pt - hi - es - th pipeline_tag: text-generation license: llama3.1 --- # Meta-Llama-3.1-70B-Instruct-quantized.w8a16 ## Model Overview - **Model Architecture:** Meta-Llama-3 - **Input:** Text - **Output:** Text - **Model Optimizations:** - **Weight quantization:** INT8 - **Intended Use Cases:** Intended for commercial and research use multiple languages. Similarly to [Meta-Llama-3.1-70B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-70B-Instruct), this models is intended for assistant-like chat. - **Out-of-scope:** Use in any manner that violates applicable laws or regulations (including trade compliance laws). - **Release Date:** 7/24/2024 - **Version:** 1.0 - **License(s):** Llama3.1 - **Model Developers:** Neural Magic Quantized version of [Meta-Llama-3.1-70B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-70B-Instruct). It achieves scores within 1.5% of the scores of the unquantized model for MMLU, ARC-Challenge, GSM-8k, Hellaswag, Winogrande and TruthfulQA. ### Model Optimizations This model was obtained by quantizing the weights of [Meta-Llama-3.1-70B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-70B-Instruct) to INT8 data type. This optimization reduces the number of bits per parameter from 16 to 8, reducing the disk size and GPU memory requirements by approximately 50%. Only the weights of the linear operators within transformers blocks are quantized. Symmetric per-channel quantization is applied, in which a linear scaling per output dimension maps the INT8 and floating point representations of the quantized weights. The [GPTQ](https://arxiv.org/abs/2210.17323) algorithm is applied for quantization, as implemented in the [llm-compressor](https://github.com/vllm-project/llm-compressor) library. GPTQ used a 10% damping factor and 256 sequences taken from Neural Magic's [LLM compression calibration dataset](https://huggingface.co/datasets/neuralmagic/LLM_compression_calibration). ## Deployment This model can be deployed efficiently using the [vLLM](https://docs.vllm.ai/en/latest/) backend, as shown in the example below. ```python from vllm import LLM, SamplingParams from transformers import AutoTokenizer model_id = "neuralmagic/Meta-Llama-3.1-70B-Instruct-quantized.w8a16" number_gpus = 4 max_model_len = 8192 sampling_params = SamplingParams(temperature=0.6, top_p=0.9, max_tokens=256) tokenizer = AutoTokenizer.from_pretrained(model_id) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Who are you?"}, ] prompts = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False) llm = LLM(model=model_id, tensor_parallel_size=number_gpus, max_model_len=max_model_len) outputs = llm.generate(prompts, sampling_params) generated_text = outputs[0].outputs[0].text print(generated_text) ``` vLLM aslo supports OpenAI-compatible serving. See the [documentation](https://docs.vllm.ai/en/latest/) for more details. ## Creation This model was created by using the [llm-compressor](https://github.com/vllm-project/llm-compressor) library as presented in the code snipet below. ```python from transformers import AutoTokenizer from datasets import Dataset from llmcompressor.transformers import SparseAutoModelForCausalLM, oneshot from llmcompressor.modifiers.quantization import GPTQModifier import random model_id = "meta-llama/Meta-Llama-3.1-70B-Instruct" num_samples = 256 max_seq_len = 8192 tokenizer = AutoTokenizer.from_pretrained(model_id) def preprocess_fn(example): return {"text": tokenizer.apply_chat_template(example["messages"], add_generation_prompt=False, tokenize=False)} ds = load_dataset("neuralmagic/LLM_compression_calibration", split="train") ds = ds.shuffle().select(range(num_samples)) ds = ds.map(preprocess_fn) examples = [tokenizer(example["text"], padding=False, max_length=max_seq_len, truncation=True) for example in ds] recipe = GPTQModifier( targets="Linear", scheme="W8A16", ignore=["lm_head"], dampening_frac=0.1, ) model = SparseAutoModelForCausalLM.from_pretrained( model_id, device_map="auto", trust_remote_code=True, ) oneshot( model=model, dataset=ds, recipe=recipe, max_seq_length=max_seq_len, num_calibration_samples=num_samples, ) model.save_pretrained("Meta-Llama-3.1-70B-Instruct-quantized.w8a16") ``` ## Evaluation The model was evaluated on MMLU, ARC-Challenge, GSM-8K, Hellaswag, Winogrande and TruthfulQA. Evaluation was conducted using the Neural Magic fork of [lm-evaluation-harness](https://github.com/neuralmagic/lm-evaluation-harness/tree/llama_3.1_instruct) (branch llama_3.1_instruct) and the [vLLM](https://docs.vllm.ai/en/stable/) engine. This version of the lm-evaluation-harness includes versions of ARC-Challenge and GSM-8K that match the prompting style of [Meta-Llama-3.1-Instruct-evals](https://huggingface.co/datasets/meta-llama/Meta-Llama-3.1-70B-Instruct-evals). ### Accuracy
Benchmark Meta-Llama-3.1-70B-Instruct Meta-Llama-3.1-70B-Instruct-quantized.w8a16 (this model) Recovery
MMLU (5-shot) 82.21 82.12 99.9%
ARC Challenge (0-shot) 95.05 93.60 98.5%
GSM-8K (CoT, 8-shot, strict-match) 93.10 92.27 99.1%
Hellaswag (10-shot) 86.40 86.11 99.7%
Winogrande (5-shot) 85.00 84.14 99.0%
TruthfulQA (0-shot, mc2) 59.83 58.90 98.5%
Average 83.60 82.66 99.1%
### Reproduction The results were obtained using the following commands: #### MMLU ``` lm_eval \ --model vllm \ --model_args pretrained="neuralmagic/Meta-Llama-3.1-70B-Instruct-quantized.w8a16",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=4 \ --tasks mmlu \ --num_fewshot 5 \ --batch_size auto ``` #### ARC-Challenge ``` lm_eval \ --model vllm \ --model_args pretrained="neuralmagic/Meta-Llama-3.1-70B-Instruct-quantized.w8a16",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=4 \ --tasks arc_challenge_llama_3.1_instruct \ --apply_chat_template \ --num_fewshot 0 \ --batch_size auto ``` #### GSM-8K ``` lm_eval \ --model vllm \ --model_args pretrained="neuralmagic/Meta-Llama-3.1-70B-Instruct-quantized.w8a16",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=4 \ --tasks gsm8k_cot_llama_3.1_instruct \ --fewshot_as_multiturn \ --apply_chat_template \ --num_fewshot 8 \ --batch_size auto ``` #### Hellaswag ``` lm_eval \ --model vllm \ --model_args pretrained="neuralmagic/Meta-Llama-3.1-70B-Instruct-quantized.w8a16",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=4 \ --tasks hellaswag \ --num_fewshot 10 \ --batch_size auto ``` #### Winogrande ``` lm_eval \ --model vllm \ --model_args pretrained="neuralmagic/Meta-Llama-3.1-70B-Instruct-quantized.w8a16",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=4 \ --tasks winogrande \ --num_fewshot 5 \ --batch_size auto ``` #### TruthfulQA ``` lm_eval \ --model vllm \ --model_args pretrained="neuralmagic/Meta-Llama-3.1-70B-Instruct-quantized.w8a16",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=4 \ --tasks truthfulqa \ --num_fewshot 0 \ --batch_size auto ```