Instructions to use ModalityDance/latent-tts-colar with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ModalityDance/latent-tts-colar with Transformers:

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

pipe = pipeline("text-generation", model="ModalityDance/latent-tts-colar")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, ColarLlama

tokenizer = AutoTokenizer.from_pretrained("ModalityDance/latent-tts-colar")
model = ColarLlama.from_pretrained("ModalityDance/latent-tts-colar")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ModalityDance/latent-tts-colar with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/ModalityDance/latent-tts-colar

SGLang

How to use ModalityDance/latent-tts-colar 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 "ModalityDance/latent-tts-colar" \
    --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": "ModalityDance/latent-tts-colar",
		"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 "ModalityDance/latent-tts-colar" \
        --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": "ModalityDance/latent-tts-colar",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ModalityDance/latent-tts-colar with Docker Model Runner:
```
docker model run hf.co/ModalityDance/latent-tts-colar
```

Add pipeline tag and paper/code links

by nielsr HF Staff - opened Jan 12

base: refs/heads/main

←

from: refs/pr/1

Discussion Files changed

+10

-5

Files changed (1) hide show

README.md +10 -5

README.md CHANGED Viewed

@@ -1,9 +1,11 @@
 ---
-library_name: transformers
-license: mit
 base_model:
 - meta-llama/Llama-3.2-1B-Instruct
 ---
 # CoLaR Model
 <div align="center">
@@ -14,7 +16,10 @@ base_model:
 ## Overview
-**CoLaR** (Compressed Latent Reasoning) is a latent reasoning model based on LLaMA that uses a specialized LatentHead module for generating continuous latent representations. This model is part of the [Parallel Test-Time Scaling for Latent Reasoning Models](https://arxiv.org/abs/2510.07745) framework.
 ## Model Details
@@ -201,7 +206,7 @@ Run evaluation using the provided scripts:
 ## Model Card
-- **Paper**: [Parallel Test-Time Scaling for Latent Reasoning Models](https://arxiv.org/abs/2510.07745)
 - **HuggingFace**: [ModalityDance/latent-tts-colar](https://huggingface.co/ModalityDance/latent-tts-colar)
 - **Benchmarks**: GSM8K Test, GSM8K Hard, MultiArith
@@ -235,4 +240,4 @@ If you use this model, please cite:
       primaryClass={cs.CL},
       url={https://arxiv.org/abs/2505.16552},
 }
-```

 ---
 base_model:
 - meta-llama/Llama-3.2-1B-Instruct
+library_name: transformers
+license: mit
+pipeline_tag: text-generation
 ---
 # CoLaR Model
 <div align="center">
 ## Overview
+**CoLaR** (Compressed Latent Reasoning) is a latent reasoning model based on LLaMA that uses a specialized LatentHead module for generating continuous latent representations. This model is part of the [Parallel Test-Time Scaling for Latent Reasoning Models](https://huggingface.co/papers/2510.07745) framework.
+- **Paper:** [Parallel Test-Time Scaling for Latent Reasoning Models](https://huggingface.co/papers/2510.07745)
+- **Code:** [https://github.com/ModalityDance/LatentTTS](https://github.com/ModalityDance/LatentTTS)
 ## Model Details
 ## Model Card
+- **Paper**: [Parallel Test-Time Scaling for Latent Reasoning Models](https://huggingface.co/papers/2510.07745)
 - **HuggingFace**: [ModalityDance/latent-tts-colar](https://huggingface.co/ModalityDance/latent-tts-colar)
 - **Benchmarks**: GSM8K Test, GSM8K Hard, MultiArith
       primaryClass={cs.CL},
       url={https://arxiv.org/abs/2505.16552},
 }
+```