Quantization made by Richard Erkhov.

DataVortexS-10.7B-dpo-v1.1 - GGUF

Model creator: https://huggingface.co/Edentns/
Original model: https://huggingface.co/Edentns/DataVortexS-10.7B-dpo-v1.1/

Name	Quant method	Size
DataVortexS-10.7B-dpo-v1.1.Q2_K.gguf	Q2_K	3.79GB
DataVortexS-10.7B-dpo-v1.1.IQ3_XS.gguf	IQ3_XS	4.21GB
DataVortexS-10.7B-dpo-v1.1.IQ3_S.gguf	IQ3_S	4.44GB
DataVortexS-10.7B-dpo-v1.1.Q3_K_S.gguf	Q3_K_S	4.41GB
DataVortexS-10.7B-dpo-v1.1.IQ3_M.gguf	IQ3_M	4.58GB
DataVortexS-10.7B-dpo-v1.1.Q3_K.gguf	Q3_K	4.91GB
DataVortexS-10.7B-dpo-v1.1.Q3_K_M.gguf	Q3_K_M	4.91GB
DataVortexS-10.7B-dpo-v1.1.Q3_K_L.gguf	Q3_K_L	5.33GB
DataVortexS-10.7B-dpo-v1.1.IQ4_XS.gguf	IQ4_XS	5.5GB
DataVortexS-10.7B-dpo-v1.1.Q4_0.gguf	Q4_0	5.73GB
DataVortexS-10.7B-dpo-v1.1.IQ4_NL.gguf	IQ4_NL	5.8GB
DataVortexS-10.7B-dpo-v1.1.Q4_K_S.gguf	Q4_K_S	5.78GB
DataVortexS-10.7B-dpo-v1.1.Q4_K.gguf	Q4_K	6.1GB
DataVortexS-10.7B-dpo-v1.1.Q4_K_M.gguf	Q4_K_M	6.1GB
DataVortexS-10.7B-dpo-v1.1.Q4_1.gguf	Q4_1	6.35GB
DataVortexS-10.7B-dpo-v1.1.Q5_0.gguf	Q5_0	6.97GB
DataVortexS-10.7B-dpo-v1.1.Q5_K_S.gguf	Q5_K_S	6.97GB
DataVortexS-10.7B-dpo-v1.1.Q5_K.gguf	Q5_K	7.16GB
DataVortexS-10.7B-dpo-v1.1.Q5_K_M.gguf	Q5_K_M	7.16GB
DataVortexS-10.7B-dpo-v1.1.Q5_1.gguf	Q5_1	7.59GB
DataVortexS-10.7B-dpo-v1.1.Q6_K.gguf	Q6_K	8.29GB
DataVortexS-10.7B-dpo-v1.1.Q8_0.gguf	Q8_0	10.74GB

Original model description:

tags: - text-generation license: cc-by-nc-4.0 language: - ko base_model: beomi/OPEN-SOLAR-KO-10.7B pipeline_tag: text-generation

DataVortexS-10.7B-dpo-v1.1

Our Team

Research & Engineering	Product Management
Kwangseok Yang	Seunghyun Choi
Jeongwon Choi	Hyoseok Choi

Model Details

Base Model

beomi/OPEN-SOLAR-KO-10.7B

Trained On

OS: Ubuntu 22.04
GPU: H100 80GB 4ea
transformers: v4.36.2

Instruction format

It follows Alpaca (Chat) format.

E.g.

text = """\
### System:
당신은 사람들이 정보를 찾을 수 있도록 도와주는 인공지능 비서입니다.

### User:
대한민국의 수도는 어디야?

### Assistant:
대한민국의 수도는 서울입니다.

### User:
서울 인구는 총 몇 명이야?
"""

Model Benchmark

Ko LM Eval Harness

Task	0-shot	5-shot	10-shot	50-shot
kobest_boolq	0.915201	0.908687	0.912913	0.912913
kobest_copa	0.815575	0.858739	0.865826	0.876896
kobest_hellaswag	0.510673	0.515149	0.517118	0.517941
kobest_sentineg	0.3517	0.977329	0.992443	0.984886
Average	0.648287	0.814976	0.822075	0.823159

Ko-LLM-Leaderboard

Average	Ko-ARC	Ko-HellaSwag	Ko-MMLU	Ko-TruthfulQA	Ko-CommonGen V2
56.78	54.35	63.44	51.09	53.85	61.16

Implementation Code

This model contains the chat_template instruction format.
You can use the code below.

from transformers import AutoModelForCausalLM, AutoTokenizer

device = "cuda" # the device to load the model onto

model = AutoModelForCausalLM.from_pretrained("Edentns/DataVortexS-10.7B-dpo-v1.1")
tokenizer = AutoTokenizer.from_pretrained("Edentns/DataVortexS-10.7B-dpo-v1.1")

messages = [
    {"role": "system", "content": "당신은 사람들이 정보를 찾을 수 있도록 도와주는 인공지능 비서입니다."},
    {"role": "user", "content": "대한민국의 수도는 어디야?"},
    {"role": "assistant", "content": "대한민국의 수도는 서울입니다."},
    {"role": "user", "content": "서울 인구는 총 몇 명이야?"}
]

encodeds = tokenizer.apply_chat_template(messages, return_tensors="pt")

model_inputs = encodeds.to(device)
model.to(device)

generated_ids = model.generate(model_inputs, max_new_tokens=1000, do_sample=True)
decoded = tokenizer.batch_decode(generated_ids)
print(decoded[0])

License

This model is licensed under the cc-by-nc-4.0. which allows others to share and adapt the model for non-commercial purposes.