Meta Llama 3.3 70B Instruct AWQ INT4

🚀 Llama 3.3 70B Instruct AWQ INT4量化模型

本项目提供了Llama 3.3 70B Instruct模型的AWQ 4位量化版本，可在不同框架下快速进行推理，同时介绍了量化过程的复现方法。

🚀 快速开始

本量化模型支持在transformers、autoawq、text-generation-inference和vLLM等不同解决方案中使用。在运行Llama 3.3 70B Instruct AWQ的INT4推理时，加载模型检查点大约需要35 GiB的VRAM，且不包括KV缓存或CUDA图，因此需要确保有足够的VRAM可用。

✨ 主要特性

• 多语言支持：支持英语、法语、意大利语、葡萄牙语、印地语、西班牙语、泰语和德语等多种语言。
• 量化优化：使用AutoAWQ将模型从FP16量化到INT4，减少内存占用，提高推理速度。
• 多框架兼容：可在transformers、autoawq、text-generation-inference和vLLM等框架中使用。

📦 安装指南

使用transformers或autoawq运行推理

pip install -q --upgrade transformers autoawq accelerate

使用text-generation-inference运行推理

pip install -q --upgrade huggingface_hub
huggingface-cli login

使用vLLM运行推理

需安装Docker，参考安装说明。

💻 使用示例

基础用法

使用transformers运行推理

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, AwqConfig

model_id = "ibnzterrell/Meta-Llama-3.3-70B-Instruct-AWQ-INT4"
quantization_config = AwqConfig(
    bits=4,
    fuse_max_seq_len=512, # Note: Update this as per your use-case
    do_fuse=True,
)

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
  model_id,
  torch_dtype=torch.float16,
  low_cpu_mem_usage=True,
  device_map="auto",
  quantization_config=quantization_config
)

prompt = [
  {"role": "system", "content": "You are a helpful assistant, that responds as a pirate."},
  {"role": "user", "content": "What's Deep Learning?"},
]
inputs = tokenizer.apply_chat_template(
  prompt,
  tokenize=True,
  add_generation_prompt=True,
  return_tensors="pt",
  return_dict=True,
).to("cuda")

outputs = model.generate(**inputs, do_sample=True, max_new_tokens=256)
print(tokenizer.batch_decode(outputs[:, inputs['input_ids'].shape[1]:], skip_special_tokens=True)[0])

使用autoawq运行推理

import torch
from awq import AutoAWQForCausalLM
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "ibnzterrell/Meta-Llama-3.3-70B-Instruct-AWQ-INT4"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoAWQForCausalLM.from_pretrained(
  model_id,
  torch_dtype=torch.float16,
  low_cpu_mem_usage=True,
  device_map="auto",
)

prompt = [
  {"role": "system", "content": "You are a helpful assistant, that responds as a pirate."},
  {"role": "user", "content": "What's Deep Learning?"},
]
inputs = tokenizer.apply_chat_template(
  prompt,
  tokenize=True,
  add_generation_prompt=True,
  return_tensors="pt",
  return_dict=True,
).to("cuda")

outputs = model.generate(**inputs, do_sample=True, max_new_tokens=256)
print(tokenizer.batch_decode(outputs[:, inputs['input_ids'].shape[1]:], skip_special_tokens=True)[0])

高级用法

使用text-generation-inference运行推理

docker run --gpus all --shm-size 1g -ti -p 8080:80 \
  -v hf_cache:/data \
  -e MODEL_ID=ibnzterrell/Meta-Llama-3.3-70B-Instruct-AWQ-INT4 \
  -e NUM_SHARD=4 \
  -e QUANTIZE=awq \
  -e HF_TOKEN=$(cat ~/.cache/huggingface/token) \
  -e MAX_INPUT_LENGTH=4000 \
  -e MAX_TOTAL_TOKENS=4096 \
  ghcr.io/huggingface/text-generation-inference:2.2.0

发送请求到部署的TGI端点：

curl 0.0.0.0:8080/v1/chat/completions \
  -X POST \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "tgi",
    "messages": [
      {
        "role": "system",
        "content": "You are a helpful assistant."
      },
      {
        "role": "user",
        "content": "What is Deep Learning?"
      }
    ],
    "max_tokens": 128
  }'

使用vLLM运行推理

docker run --runtime nvidia --gpus all --ipc=host -p 8000:8000 \
  -v hf_cache:/root/.cache/huggingface \
  vllm/vllm-openai:latest \
  --model ibnzterrell/Meta-Llama-3.3-70B-Instruct-AWQ-INT4 \
  --tensor-parallel-size 4 \
  --max-model-len 4096

发送请求到部署的vLLM端点：

curl 0.0.0.0:8000/v1/chat/completions \
  -X POST \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "ibnzterrell/Meta-Llama-3.3-70B-Instruct-AWQ-INT4",
    "messages": [
      {
        "role": "system",
        "content": "You are a helpful assistant."
      },
      {
        "role": "user",
        "content": "What is Deep Learning?"
      }
    ],
    "max_tokens": 128
  }'

📚 详细文档

原始模型信息

Meta Llama 3.3是一个70B参数的多语言大语言模型（LLM），经过预训练和指令调优，适用于多语言对话场景，在常见行业基准测试中表现优于许多开源和闭源聊天模型。

量化复现信息

要使用AutoAWQ对Llama 3.3 70B Instruct进行量化，需要使用至少有足够CPU RAM（约140GiB）来容纳整个模型的实例，以及具有40GiB VRAM的NVIDIA GPU。

pip install -q --upgrade transformers autoawq accelerate

量化代码示例：

from awq import AutoAWQForCausalLM
from transformers import AutoTokenizer
import torch

# Empty Cache
torch.cuda.empty_cache()

# Memory Limits - Set this according to your hardware limits
max_memory = {0: "22GiB", 1: "22GiB", "cpu": "160GiB"}

model_path = "meta-llama/Llama-3.3-70B-Instruct"
quant_path = "ibnzterrell/Meta-Llama-3.3-70B-Instruct-AWQ-INT4"
quant_config = {
  "zero_point": True,
  "q_group_size": 128,
  "w_bit": 4,
  "version": "GEMM"
  
}

# Load model - Note: while this loads the layers into the CPU, the GPUs (and the VRAM) are still required for quantization! (Verified with nvida-smi)
model = AutoAWQForCausalLM.from_pretrained(
    model_path,
    use_cache=False,
    max_memory=max_memory,
    device_map="cpu"
)

tokenizer = AutoTokenizer.from_pretrained(model_path)

# Quantize
model.quantize(
    tokenizer,
    quant_config=quant_config
)

# Save quantized model
model.save_quantized(quant_path)
tokenizer.save_pretrained(quant_path)

print(f'Model is quantized and saved at "{quant_path}"')

🔧 技术细节

本模型使用AutoAWQ将Llama 3.3 70B Instruct模型从FP16量化到INT4，采用GEMM内核，零点位量化，组大小为128。量化过程在单个节点上进行，该节点配备了Intel Xeon CPU E5 - 2699A v4 @ 2.40GHz、256GB的RAM和2块NVIDIA RTX 3090（每块24GB VRAM，总共48GB VRAM）。

📄 许可证

本项目使用的模型遵循llama3.3许可证。