Instructions to use TwT-6/cr-model with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TwT-6/cr-model with Transformers:

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

pipe = pipeline("text-generation", model="TwT-6/cr-model")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("TwT-6/cr-model")
model = AutoModelForCausalLM.from_pretrained("TwT-6/cr-model")
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 TwT-6/cr-model with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TwT-6/cr-model"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TwT-6/cr-model",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/TwT-6/cr-model

SGLang

How to use TwT-6/cr-model 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 "TwT-6/cr-model" \
    --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": "TwT-6/cr-model",
		"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 "TwT-6/cr-model" \
        --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": "TwT-6/cr-model",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use TwT-6/cr-model with Docker Model Runner:
```
docker model run hf.co/TwT-6/cr-model
```

My model is a state-of-the-art language processing AI designed to understand and generate human-like text. It leverages deep learning algorithms to engage in a wide range of language tasks, providing users with information, recommendations, and even casual conversation. With a broad knowledge base and nuanced understanding of context, my capabilities enable me to assist with various inquiries and perform complex language-based tasks effectively.

How to use？

from transformers import AutoModelForCausalLM, AutoTokenizer

from transformers.generation import GenerationConfig

import torch

model = AutoModelForCausalLM.from_pretrained( 'TwT-6/cr-model', attn_implementation="flash_attention_2", trust_remote_code=True, torch_dtype=torch.bfloat16, device_map="auto").eval()

tokenizer = AutoTokenizer.from_pretrained('TwT-6/cr-model', trust_remote_code=True)

inputs = '你好'

inputs = f'<|omni_start|>### User:\n{inputs}\n\n### Assistant:\n'

inputs = tokenizer(inputs, return_tensors="pt").to('cuda')

output_ids = model.generate(**inputs)[0].cpu()

output = tokenizer.decode(output_ids[inputs.input_ids.shape[-1]:])

print(output)

你好！很高兴见到你。有什么我可以帮助你的吗

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	68.09
AI2 Reasoning Challenge (25-Shot)	57.85
HellaSwag (10-Shot)	81.66
MMLU (5-Shot)	68.73
TruthfulQA (0-shot)	58.20
Winogrande (5-shot)	76.24
GSM8k (5-shot)	65.88

Downloads last month: 38

Safetensors

Model size

14B params

Tensor type

BF16

Evaluation results

normalized accuracy on AI2 Reasoning Challenge (25-Shot)
test set Open LLM Leaderboard

57.850
normalized accuracy on HellaSwag (10-Shot)
validation set Open LLM Leaderboard

81.660
accuracy on MMLU (5-Shot)
test set Open LLM Leaderboard

68.730
mc2 on TruthfulQA (0-shot)
validation set Open LLM Leaderboard

58.200
accuracy on Winogrande (5-shot)
validation set Open LLM Leaderboard

76.240
accuracy on GSM8k (5-shot)
test set Open LLM Leaderboard

65.880