mlabonne/open-perfectblend
Viewer β’ Updated β’ 1.42M β’ 1.07k β’ 72
How to use SejongKRX/Sejong-Qwen-v6 with Transformers:
# Use a pipeline as a high-level helper
from transformers import pipeline
pipe = pipeline("text-generation", model="SejongKRX/Sejong-Qwen-v6") # Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("SejongKRX/Sejong-Qwen-v6")
model = AutoModelForCausalLM.from_pretrained("SejongKRX/Sejong-Qwen-v6")How to use SejongKRX/Sejong-Qwen-v6 with vLLM:
# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SejongKRX/Sejong-Qwen-v6"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "SejongKRX/Sejong-Qwen-v6",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'docker model run hf.co/SejongKRX/Sejong-Qwen-v6
How to use SejongKRX/Sejong-Qwen-v6 with SGLang:
# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
--model-path "SejongKRX/Sejong-Qwen-v6" \
--host 0.0.0.0 \
--port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "SejongKRX/Sejong-Qwen-v6",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'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 "SejongKRX/Sejong-Qwen-v6" \
--host 0.0.0.0 \
--port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "SejongKRX/Sejong-Qwen-v6",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'How to use SejongKRX/Sejong-Qwen-v6 with Unsloth Studio:
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for SejongKRX/Sejong-Qwen-v6 to start chatting
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for SejongKRX/Sejong-Qwen-v6 to start chatting
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for SejongKRX/Sejong-Qwen-v6 to start chatting
pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
model_name="SejongKRX/Sejong-Qwen-v6",
max_seq_length=2048,
)How to use SejongKRX/Sejong-Qwen-v6 with Docker Model Runner:
docker model run hf.co/SejongKRX/Sejong-Qwen-v6
!pip install transformers einops accelerate
!pip install qwen
!pip install unsloth
from transformers import AutoTokenizer, AutoModelForCausalLM
# ν ν¬λμ΄μ μ λͺ¨λΈ λ‘λ
tokenizer = AutoTokenizer.from_pretrained(
"SejongKRX/Sejong-Qwen-v6",
trust_remote_code=True,
use_fast=False
)
model = AutoModelForCausalLM.from_pretrained(
"SejongKRX/Sejong-Qwen-v6",
trust_remote_code=True
)
# μ
λ ₯ ν
μ€νΈ
input_text = """
λ€μ μ€ ννμ μκ°κ°μΉμ κ΄ν μ€λͺ
μΌλ‘ μ³μ§ μμ κ²μ 무μμΈκ°?
A. μ 볡리μ κ²½μ°, λ§€μ μ μ©λλ μ΄μμ¨μ μ°κ° λͺ
λͺ© μ΄μμ¨μ 1/12λ‘ λλμ΄ μ°μΆνλ€.
B. ν¬μ μκΈ λ° κΈ°ν μ‘°κ±΄μ΄ λμΌν κ²½μ°, λ¨λ¦¬ λ°©μλ³΄λ€ λ³΅λ¦¬ λ°©μμμ λ°μνλ μ΄μκ° λ ν¬λ€.
C. μΌμλΆλ‘ μ§κΈλ κΈμ‘μ νμ¬ κ°μΉλ λ―Έλ κ°μΉλ₯Ό μΌμ κΈ°κ° λμ ν μΈμ¨μ μ μ©ν΄ μ°μΆν μ μλ€.
D. 1,000,000μμ μ° 5% λ³΅λ¦¬λ‘ 2λ
λμ μμΉνμ κ²½μ°, λ§κΈ°μ λ°μ μΈμ μ΄μλ 100,000μμ΄λ€.
### μ λ΅:
"""
inputs = tokenizer(input_text, return_tensors="pt")
# λͺ¨λΈμ μ¬μ©νμ¬ ν
μ€νΈ μμ±
output = model.generate(**inputs, max_new_tokens=1500)
# κ²°κ³Ό λμ½λ©
generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
print(generated_text)
output:
λ€μ μ€ ννμ μκ°κ°μΉμ κ΄ν μ€λͺ
μΌλ‘ μ³μ§ μμ κ²μ 무μμΈκ°?
A. μ 볡리μ κ²½μ°, λ§€μ μ μ©λλ μ΄μμ¨μ μ°κ° λͺ
λͺ© μ΄μμ¨μ 1/12λ‘ λλμ΄ μ°μΆνλ€.
B. ν¬μ μκΈ λ° κΈ°ν μ‘°κ±΄μ΄ λμΌν κ²½μ°, λ¨λ¦¬ λ°©μλ³΄λ€ λ³΅λ¦¬ λ°©μμμ λ°μνλ μ΄μκ° λ ν¬λ€.
C. μΌμλΆλ‘ μ§κΈλ κΈμ‘μ νμ¬ κ°μΉλ λ―Έλ κ°μΉλ₯Ό μΌμ κΈ°κ° λμ ν μΈμ¨μ μ μ©ν΄ μ°μΆν μ μλ€.
D. 1,000,000μμ μ° 5% λ³΅λ¦¬λ‘ 2λ
λμ μμΉνμ κ²½μ°, λ§κΈ°μ λ°μ μΈμ μ΄μλ 100,000μμ΄λ€.
### μ λ΅:
D
λ³Έ λͺ¨λΈμ λ€μν μΆμ²μ λ°μ΄ν°(mlabonne/open-perfectblend, Wikipedia, νκ΅μνμ 곡곡 λ°μ΄ν° λ±)λ₯Ό νμ©νμ¬ νμ΅λμμΌλ©°, λͺ¨λ λ°μ΄ν°λ μ μκΆ λ° μ¬μ© μ μ± μ λ°λΌ μ μ ν μ¬μ©λμμ΅λλ€.
This qwen2 model was trained 2x faster with Unsloth and Huggingface's TRL library.