Instructions to use 0xSero/Trinity-337B-W4A16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use 0xSero/Trinity-337B-W4A16 with Transformers:

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

pipe = pipeline("text-generation", model="0xSero/Trinity-337B-W4A16", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("0xSero/Trinity-337B-W4A16", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("0xSero/Trinity-337B-W4A16", trust_remote_code=True)
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 0xSero/Trinity-337B-W4A16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "0xSero/Trinity-337B-W4A16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "0xSero/Trinity-337B-W4A16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/0xSero/Trinity-337B-W4A16

SGLang

How to use 0xSero/Trinity-337B-W4A16 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 "0xSero/Trinity-337B-W4A16" \
    --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": "0xSero/Trinity-337B-W4A16",
		"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 "0xSero/Trinity-337B-W4A16" \
        --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": "0xSero/Trinity-337B-W4A16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use 0xSero/Trinity-337B-W4A16 with Docker Model Runner:
```
docker model run hf.co/0xSero/Trinity-337B-W4A16
```

Trinity-337B-W4A16 / dequant_verification.json

0xSero

W4A16 RTN quantization (216/256 experts, 166GB)

8032f24 verified about 2 months ago

raw

history blame contribute delete

1.57 kB

	{
	"layers": [
	{
	"label": "dense_mlp_L0",
	"layer": "model.layers.0.mlp.gate_proj",
	"snr_db": 8.8,
	"cosine": 0.942961,
	"mae": 0.001491
	},
	{
	"label": "dense_mlp_L2",
	"layer": "model.layers.2.mlp.gate_proj",
	"snr_db": 9.0,
	"cosine": 0.946257,
	"mae": 0.001487
	},
	{
	"label": "dense_mlp_L5",
	"layer": "model.layers.5.mlp.gate_proj",
	"snr_db": 9.1,
	"cosine": 0.946178,
	"mae": 0.001739
	},
	{
	"label": "moe_L6_E0",
	"layer": "model.layers.6.mlp.experts.0.gate_proj",
	"snr_db": 8.6,
	"cosine": 0.936454,
	"mae": 0.001716
	},
	{
	"label": "moe_L6_E100",
	"layer": "model.layers.6.mlp.experts.100.gate_proj",
	"snr_db": 8.6,
	"cosine": 0.936431,
	"mae": 0.001703
	},
	{
	"label": "moe_L30_E0",
	"layer": "model.layers.30.mlp.experts.0.gate_proj",
	"snr_db": 8.6,
	"cosine": 0.936316,
	"mae": 0.001769
	},
	{
	"label": "moe_L30_E100",
	"layer": "model.layers.30.mlp.experts.100.gate_proj",
	"snr_db": 8.6,
	"cosine": 0.93641,
	"mae": 0.001734
	},
	{
	"label": "moe_L59_E0",
	"layer": "model.layers.59.mlp.experts.0.gate_proj",
	"snr_db": 8.6,
	"cosine": 0.936509,
	"mae": 0.001727
	},
	{
	"label": "moe_L59_E100",
	"layer": "model.layers.59.mlp.experts.100.gate_proj",
	"snr_db": 8.6,
	"cosine": 0.936509,
	"mae": 0.001688
	}
	],
	"avg_snr_db": 8.7,
	"avg_cosine": 0.9393
	}