Instructions to use bytedance-research/ChatTS-14B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use bytedance-research/ChatTS-14B with Transformers:

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

pipe = pipeline("text-generation", model="bytedance-research/ChatTS-14B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("bytedance-research/ChatTS-14B", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use bytedance-research/ChatTS-14B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "bytedance-research/ChatTS-14B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bytedance-research/ChatTS-14B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/bytedance-research/ChatTS-14B

SGLang

How to use bytedance-research/ChatTS-14B 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 "bytedance-research/ChatTS-14B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bytedance-research/ChatTS-14B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "bytedance-research/ChatTS-14B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bytedance-research/ChatTS-14B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use bytedance-research/ChatTS-14B with Docker Model Runner:
```
docker model run hf.co/bytedance-research/ChatTS-14B
```

Fix BF16 training

#19

by alexanderchemeris - opened Jun 3, 2025

base: refs/heads/main

←

from: refs/pr/19

Discussion Files changed

-1

alexanderchemeris

Jun 3, 2025

For long sequences, this calculation yields an incorrect result due to a lower number of bits in the mantissa of BF16. E.g., for 640 elements, this produces valid_lengths = 639. Converting this early to long solves the issue.

Fix BF16 training3633e9c8

xiezhe24

bytedance-research org Jun 4, 2025

Thank you for pointing out this issue. But I think this issue has already been fixed by line 133:

mask = x[:, :, -1].long()

Do you think there are still issues with the current code?

alexanderchemeris

Jun 4, 2025

Sorry, I didn't notice this recent fix. I think it's equivalent. I'll check and come back.

xiezhe24 changed pull request status to closed Jul 1, 2025

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment