Instructions to use ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16") model = AutoModelForCausalLM.from_pretrained("ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16") messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=40) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16
- SGLang
How to use ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16 with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16 with Docker Model Runner:
docker model run hf.co/ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16
use
how to use ?
where code example
import transformers
import torch
model_id = "ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16"
tokenizer = transformers.AutoTokenizer.from_pretrained(model_id, use_fast=True)
pipeline = transformers.pipeline(
"text-generation",
model=model_id,
tokenizer=tokenizer,
model_kwargs={"torch_dtype": torch.bfloat16},
device_map="auto",
)
messages = [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Who is Napoleon Bonaparte?"},
]
outputs = pipeline(
messages,
max_new_tokens=128, # Reduced number of tokens
)
print(outputs[0]["generated_text"])
35mint
not work
it need gpu
i use colab
not gpu
40mint
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
تحديد معرف النموذج
model_id = "ISTA-DASLab/Meta-Llama-3-8B-Instruct-AQLM-2Bit-1x16"
تحميل الـ tokenizer والموديل
tokenizer = AutoTokenizer.from_pretrained(model_id)
تعيين pad_token إلى eos_token إذا كان غير موجود
if tokenizer.pad_token is None:
tokenizer.pad_token = tokenizer.eos_token
تحميل النموذج مع تخفيض استخدام الذاكرة
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16, low_cpu_mem_usage=True)
وضع النموذج في طور التقييم
model.eval()
إدخال نص للسؤال
prompt = "ما هي عاصمة فرنسا؟"
ترميز المدخلات مع تضمين attention_mask وتعيين pad_token_id
inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True, max_length=128)
وضع البيانات على وحدة المعالجة المركزية (CPU)
inputs = {key: value.to("cpu") for key, value in inputs.items()}
توليد الإجابة مع تحسينات السرعة وتقليل استخدام الذاكرة
with torch.inference_mode(): # أسرع من no_grad
outputs = model.generate(
inputs["input_ids"],
attention_mask=inputs["attention_mask"], # تمرير attention_mask
max_length=50, # تحديد الحد الأقصى لطول النص الناتج لتسريع التوليد
pad_token_id=tokenizer.pad_token_id, # تعيين pad_token_id
num_beams=1, # استخدام شعاع واحد لتقليل الحسابات وتسريع التوليد
do_sample=True, # تقليل الاحتمالات باستخدام العينة
temperature=0.7 # تحسين تنوع الإجابات مع الحفاظ على السرعة
)
فك ترميز الناتج
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)
on cloab without gpu
not work