ضبط نموذج LLM بتقنية QLoRA خطوة بخطوة (كود كامل)

python -m venv .venv && source .venv/bin/activate
pip install -U torch transformers peft trl bitsandbytes datasets accelerate

{"messages": [
  {"role": "system", "content": "أنت مساعد دعم فني لشركة Mit AI."},
  {"role": "user", "content": "كيف أحجز استشارة؟"},
  {"role": "assistant", "content": "اضغط زر «احجز استشارة مجانية» وعبّئ النموذج، وسنتواصل معك."}
]}

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig

MODEL_ID = "Qwen/Qwen2.5-7B-Instruct"  # أو أي نموذج مفتوح مناسب

bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.bfloat16,
    bnb_4bit_use_double_quant=True,
)

tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
model = AutoModelForCausalLM.from_pretrained(
    MODEL_ID,
    quantization_config=bnb_config,
    device_map="auto",
)

from peft import LoraConfig

peft_config = LoraConfig(
    r=16,                # رتبة المصفوفات: أكبر = سعة أكبر وذاكرة أكثر
    lora_alpha=32,       # معامل التحجيم، غالباً 1–2 ضعف r
    lora_dropout=0.05,
    target_modules="all-linear",
    task_type="CAUSAL_LM",
)

from datasets import load_dataset
from trl import SFTConfig, SFTTrainer

dataset = load_dataset("json", data_files={"train": "train.jsonl", "validation": "val.jsonl"})

training_args = SFTConfig(
    output_dir="out/mit-qlora",
    num_train_epochs=2,
    per_device_train_batch_size=2,
    gradient_accumulation_steps=8,   # حجم دفعة فعلي = 16
    learning_rate=2e-4,
    lr_scheduler_type="cosine",
    warmup_ratio=0.03,
    bf16=True,
    gradient_checkpointing=True,     # يوفّر ذاكرة مقابل زمن أطول
    logging_steps=10,
    eval_strategy="steps",
    eval_steps=100,
    save_steps=100,
    max_length=2048,
)

trainer = SFTTrainer(
    model=model,
    args=training_args,
    train_dataset=dataset["train"],
    eval_dataset=dataset["validation"],
    peft_config=peft_config,
    processing_class=tokenizer,
)
trainer.train()
trainer.save_model("out/mit-qlora/adapter")

messages = [{"role": "user", "content": "كيف أحجز استشارة؟"}]
inputs = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True, return_tensors="pt", return_dict=True
).to(trainer.model.device)

output = trainer.model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(output[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))

from peft import PeftModel

base = AutoModelForCausalLM.from_pretrained(MODEL_ID, torch_dtype=torch.bfloat16, device_map="auto")
merged = PeftModel.from_pretrained(base, "out/mit-qlora/adapter").merge_and_unload()
merged.save_pretrained("out/mit-merged")
tokenizer.save_pretrained("out/mit-merged")