Instructions to use xDAN-AI/xDAN-L1-Chat-RL-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use xDAN-AI/xDAN-L1-Chat-RL-v1 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="xDAN-AI/xDAN-L1-Chat-RL-v1")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("xDAN-AI/xDAN-L1-Chat-RL-v1")
model = AutoModelForCausalLM.from_pretrained("xDAN-AI/xDAN-L1-Chat-RL-v1")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use xDAN-AI/xDAN-L1-Chat-RL-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "xDAN-AI/xDAN-L1-Chat-RL-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "xDAN-AI/xDAN-L1-Chat-RL-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/xDAN-AI/xDAN-L1-Chat-RL-v1

SGLang

How to use xDAN-AI/xDAN-L1-Chat-RL-v1 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 "xDAN-AI/xDAN-L1-Chat-RL-v1" \
    --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": "xDAN-AI/xDAN-L1-Chat-RL-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "xDAN-AI/xDAN-L1-Chat-RL-v1" \
        --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": "xDAN-AI/xDAN-L1-Chat-RL-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use xDAN-AI/xDAN-L1-Chat-RL-v1 with Docker Model Runner:
```
docker model run hf.co/xDAN-AI/xDAN-L1-Chat-RL-v1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Top 1 Performer on MT-bench🏆

**The first top model which is performance at Humanalities, Coding and Writing with 7b. **

xDAN-AI • > Discord • Twitter • Huggingface

Outperformer GPT3.5turbo & Claude-v1

Touch nearby GPT4 on MT-Bench

########## First turn ##########

model	turn	score	size
gpt-4	1	8.95625	-
xDAN-L1-Chat-RL-v1	1	8.87500	7b
xDAN-L2-Chat-RL-v2	1	8.78750	30b
claude-v1	1	8.15000	-
gpt-3.5-turbo	1	8.07500	20b
vicuna-33b-v1.3	1	7.45625	33b
wizardlm-30b	1	7.13125	30b
oasst-sft-7-llama-30b	1	7.10625	30b
Llama-2-70b-chat	1	6.98750	70b

########## Second turn ##########

model	turn	score	size
gpt-4	2	9.025000	-
xDAN-L2-Chat-RL-v2	2	8.087500	30b
xDAN-L1-Chat-RL-v1	2	7.825000	7b
gpt-3.5-turbo	2	7.812500	20b
claude-v1	2	7.650000	-
wizardlm-30b	2	6.887500	30b
vicuna-33b-v1.3	2	6.787500	33b
Llama-2-70b-chat	2	6.725000	70b

########## Average turn##########

model	score	size
gpt-4	8.990625	-
xDAN-L2-Chat-RL-v2	8.437500	30b
xDAN-L1-Chat-RL-v1	8.350000	7b
gpt-3.5-turbo	7.943750	20b
claude-v1	7.900000	-
vicuna-33b-v1.3	7.121875	33b
wizardlm-30b	7.009375	30b
Llama-2-70b-chat	6.856250	70b

LM-Evaluation-Harness

Task	Score
Average	68.38
ARC	66.3
HellaSwag	85.81
MMLU	63.21
TruthfulQA	56.7
Winogrande	78.85
GSM8K	59.44

Prompt Template(Alpaca)

You are a helpful assistant named DAN. You are an expert in worldly knowledge, skilled in employing a probing questioning strategy, and you carefully consider each step before providing answers. \n\n### Instruction:\n{instruction}\n\n### Response:

Dataset:

Selected from OpenOrca
Intel Orca-DPO-Pairs
Privately Crafted Dataset

Training:

SFT with Mixed dataset from OpenOrca & Intel
The DPO-v2 dataset
The DPO-v2 Trainer

Created By xDAN-AI at 2023-12-15

Eval by FastChat: https://github.com/lm-sys/FastChat.git

Disclaimer

We employ rigorous data compliance validation algorithms throughout the training of our language model to ensure the highest level of compliance. However, due to the intricate nature of data and the wide range of potential usage scenarios for the model, we cannot guarantee that it will consistently produce accurate and sensible outputs. Users should be aware of the possibility of the model generating problematic results. Our organization disclaims any responsibility for risks or issues arising from misuse, improper guidance, unlawful usage, misinformation, or subsequent concerns regarding data security.

About xDAN-AI

xDAN-AI represents the forefront of Silicon-Based Life Factory technology. For comprehensive information and deeper insights into our cutting-edge technology and offerings, please visit our website: https://www.xdan.ai.

Downloads last month: 221

Safetensors

Model size

7B params

Tensor type

F32

Model tree for xDAN-AI/xDAN-L1-Chat-RL-v1

Merges

1 model

Quantizations

4 models

xDAN-AI
/

xDAN-L1-Chat-RL-v1