ئوچۇق مەنبە LLM لار

from openai import OpenAI

# DeepSeek API ئىشلىتىش (OpenAI compatible)
client = OpenAI(
    api_key="your-deepseek-key",
    base_url="https://api.deepseek.com"
)

response = client.chat.completions.create(
    model="deepseek-reasoner",  # R1 مودېلى
    messages=[
        {"role": "user", "content": "بۇ ماتېماتىكا مەسىلىسىنى قەدەممۇ-قەدەم ھەل قىل..."}
    ]
)

print(response.choices[0].message.content)

from transformers import AutoModelForCausalLM, AutoTokenizer

# Qwen مودېلىنى يۈكلەش
model_name = "Qwen/Qwen2.5-7B-Instruct"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)

# سوئال سوراش
messages = [
    {"role": "user", "content": "Python دىكى decorator نى چۈشەندۈر"}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer([text], return_tensors="pt").to(model.device)

outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

# Ollama بىلەن Llama نى ئىجرا قىلىش
ollama pull llama3.3

# سۆھبەت باشلاش
ollama run llama3.3

# API سۈپىتىدە ئىشلىتىش
curl http://localhost:11434/api/generate -d '{
  "model": "llama3.3",
  "prompt": "Explain quantum computing"
}'

# === Ollama بىلەن ===
# ئورنىتىش
curl -fsSL https://ollama.com/install.sh | sh

# مودېل ئىجرا
ollama run qwen2.5:7b

# === vLLM بىلەن ===
pip install vllm

# API server قوزغىتىش
python -m vllm.entrypoints.openai.api_server \
    --model Qwen/Qwen2.5-7B-Instruct \
    --port 8000

# === Docker TGI بىلەن ===
docker run --gpus all -p 8080:80 \
    ghcr.io/huggingface/text-generation-inference:latest \
    --model-id meta-llama/Llama-3.3-70B-Instruct

from unsloth import FastLanguageModel
from trl import SFTTrainer

# مودېلنى يۈكلەش (4-bit quantized)
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="unsloth/Qwen2.5-7B-Instruct",
    max_seq_length=2048,
    load_in_4bit=True,
)

# LoRA adapter قوشۇش
model = FastLanguageModel.get_peft_model(
    model,
    r=16,                    # LoRA rank
    target_modules=["q_proj", "k_proj", "v_proj", "o_proj"],
    lora_alpha=16,
    lora_dropout=0,
    bias="none",
)

# تەربىيىلەش
trainer = SFTTrainer(
    model=model,
    tokenizer=tokenizer,
    train_dataset=dataset,
    max_seq_length=2048,
    dataset_text_field="text",
)

trainer.train()