TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2

Instructions to use ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2 with Transformers:

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

pipe = pipeline("text-generation", model="ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2")
model = AutoModelForCausalLM.from_pretrained("ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2

SGLang

How to use ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2 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 "ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2" \
    --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": "ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2",
		"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 "ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2" \
        --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": "ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2 with Docker Model Runner:
```
docker model run hf.co/ArchiveAI/TheSalt-RP-L3-8b-DPO-v0.3.2-e0.14.2
```

An experimental llama3 RP model, stuffed with lots of tokens and then DPO'd for RP. Making it public so people can give me feedback and mess around with it.

Uses ChatML but you have to format it a certain way. The usernames are required, a system prompt is required.

Datasets Used?

Yes.

Prompt Format ( ChatML with usernames. )

<|im_start|>system
{System Prompt}<|im_end|>
<|im_start|>user
{username}: {usertext}<|im_end|>
<|im_start|>assistant
{botname}: {bottext}<|im_end|>
<|im_start|>user
{username}: {usertext}<|im_end|>
<|im_start|>assistant
{botname}:

Disclaimer

Please prompt responsibly and take anything outputted by any Language Model with a huge grain of salt. Thanks!

Downloads last month: 1

Safetensors

Model size

8B params

Tensor type

BF16