Instructions to use syscall42/nemotron-twotower-nvfp4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use syscall42/nemotron-twotower-nvfp4 with Transformers:

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

pipe = pipeline("text-generation", model="syscall42/nemotron-twotower-nvfp4")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("syscall42/nemotron-twotower-nvfp4", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use syscall42/nemotron-twotower-nvfp4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "syscall42/nemotron-twotower-nvfp4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "syscall42/nemotron-twotower-nvfp4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/syscall42/nemotron-twotower-nvfp4

SGLang

How to use syscall42/nemotron-twotower-nvfp4 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 "syscall42/nemotron-twotower-nvfp4" \
    --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": "syscall42/nemotron-twotower-nvfp4",
		"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 "syscall42/nemotron-twotower-nvfp4" \
        --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": "syscall42/nemotron-twotower-nvfp4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use syscall42/nemotron-twotower-nvfp4 with Docker Model Runner:
```
docker model run hf.co/syscall42/nemotron-twotower-nvfp4
```

nemotron-twotower-nvfp4 / hf_quant_config.json

syscall42

Add files using upload-large-folder tool

f430b91 verified 3 days ago

Raw

History Blame Contribute Delete

7.93 kB

	{
	"producer": {
	"name": "modelopt",
	"version": "0.46.0.dev129+g973cb09cb"
	},
	"quantization": {
	"quant_algo": "NVFP4",
	"kv_cache_quant_algo": null,
	"group_size": 16,
	"exclude_modules": [
	"context_lm_head",
	"context_tower.embeddings",
	"context_tower.layers.0*",
	"context_tower.layers.1.mixer.gate",
	"context_tower.layers.1.mixer.shared_experts*",
	"context_tower.layers.10.mixer.gate",
	"context_tower.layers.10.mixer.shared_experts*",
	"context_tower.layers.11*",
	"context_tower.layers.12*",
	"context_tower.layers.13.mixer.gate",
	"context_tower.layers.13.mixer.shared_experts*",
	"context_tower.layers.14*",
	"context_tower.layers.15.mixer.gate",
	"context_tower.layers.15.mixer.shared_experts*",
	"context_tower.layers.16*",
	"context_tower.layers.17.mixer.gate",
	"context_tower.layers.17.mixer.shared_experts*",
	"context_tower.layers.18*",
	"context_tower.layers.19*",
	"context_tower.layers.2.*",
	"context_tower.layers.20.mixer.gate",
	"context_tower.layers.20.mixer.shared_experts*",
	"context_tower.layers.21*",
	"context_tower.layers.22.mixer.gate",
	"context_tower.layers.22.mixer.shared_experts*",
	"context_tower.layers.23*",
	"context_tower.layers.24.mixer.gate",
	"context_tower.layers.24.mixer.shared_experts*",
	"context_tower.layers.25*",
	"context_tower.layers.26*",
	"context_tower.layers.27.mixer.gate",
	"context_tower.layers.27.mixer.shared_experts*",
	"context_tower.layers.28*",
	"context_tower.layers.29.mixer.gate",
	"context_tower.layers.29.mixer.shared_experts*",
	"context_tower.layers.3.mixer.gate",
	"context_tower.layers.3.mixer.shared_experts*",
	"context_tower.layers.30*",
	"context_tower.layers.31.mixer.gate",
	"context_tower.layers.31.mixer.shared_experts*",
	"context_tower.layers.32*",
	"context_tower.layers.33*",
	"context_tower.layers.34.mixer.gate",
	"context_tower.layers.34.mixer.shared_experts*",
	"context_tower.layers.35*",
	"context_tower.layers.36.mixer.gate",
	"context_tower.layers.36.mixer.shared_experts*",
	"context_tower.layers.37*",
	"context_tower.layers.38.mixer.gate",
	"context_tower.layers.38.mixer.shared_experts*",
	"context_tower.layers.39*",
	"context_tower.layers.4.*",
	"context_tower.layers.40.mixer.gate",
	"context_tower.layers.40.mixer.shared_experts*",
	"context_tower.layers.41*",
	"context_tower.layers.42*",
	"context_tower.layers.43.mixer.gate",
	"context_tower.layers.43.mixer.shared_experts*",
	"context_tower.layers.44*",
	"context_tower.layers.45.mixer.gate",
	"context_tower.layers.45.mixer.shared_experts*",
	"context_tower.layers.46*",
	"context_tower.layers.47.mixer.gate",
	"context_tower.layers.47.mixer.shared_experts*",
	"context_tower.layers.48*",
	"context_tower.layers.49.mixer.gate",
	"context_tower.layers.49.mixer.shared_experts*",
	"context_tower.layers.5.*",
	"context_tower.layers.50*",
	"context_tower.layers.51.mixer.gate",
	"context_tower.layers.51.mixer.shared_experts*",
	"context_tower.layers.6.mixer.gate",
	"context_tower.layers.6.mixer.shared_experts*",
	"context_tower.layers.7*",
	"context_tower.layers.8.mixer.gate",
	"context_tower.layers.8.mixer.shared_experts*",
	"context_tower.layers.9*",
	"denoiser_tower.embeddings",
	"denoiser_tower.layers.0*",
	"denoiser_tower.layers.1.mixer.gate",
	"denoiser_tower.layers.1.mixer.shared_experts*",
	"denoiser_tower.layers.10.mixer.gate",
	"denoiser_tower.layers.10.mixer.shared_experts*",
	"denoiser_tower.layers.11*",
	"denoiser_tower.layers.12*",
	"denoiser_tower.layers.13.mixer.gate",
	"denoiser_tower.layers.13.mixer.shared_experts*",
	"denoiser_tower.layers.14*",
	"denoiser_tower.layers.15.mixer.gate",
	"denoiser_tower.layers.15.mixer.shared_experts*",
	"denoiser_tower.layers.16*",
	"denoiser_tower.layers.17.mixer.gate",
	"denoiser_tower.layers.17.mixer.shared_experts*",
	"denoiser_tower.layers.18*",
	"denoiser_tower.layers.19*",
	"denoiser_tower.layers.2.*",
	"denoiser_tower.layers.20.mixer.gate",
	"denoiser_tower.layers.20.mixer.shared_experts*",
	"denoiser_tower.layers.21*",
	"denoiser_tower.layers.22.mixer.gate",
	"denoiser_tower.layers.22.mixer.shared_experts*",
	"denoiser_tower.layers.23*",
	"denoiser_tower.layers.24.mixer.gate",
	"denoiser_tower.layers.24.mixer.shared_experts*",
	"denoiser_tower.layers.25*",
	"denoiser_tower.layers.26*",
	"denoiser_tower.layers.27.mixer.gate",
	"denoiser_tower.layers.27.mixer.shared_experts*",
	"denoiser_tower.layers.28*",
	"denoiser_tower.layers.29.mixer.gate",
	"denoiser_tower.layers.29.mixer.shared_experts*",
	"denoiser_tower.layers.3.mixer.gate",
	"denoiser_tower.layers.3.mixer.shared_experts*",
	"denoiser_tower.layers.30*",
	"denoiser_tower.layers.31.mixer.gate",
	"denoiser_tower.layers.31.mixer.shared_experts*",
	"denoiser_tower.layers.32*",
	"denoiser_tower.layers.33*",
	"denoiser_tower.layers.34.mixer.gate",
	"denoiser_tower.layers.34.mixer.shared_experts*",
	"denoiser_tower.layers.35*",
	"denoiser_tower.layers.36.mixer.gate",
	"denoiser_tower.layers.36.mixer.shared_experts*",
	"denoiser_tower.layers.37*",
	"denoiser_tower.layers.38.mixer.gate",
	"denoiser_tower.layers.38.mixer.shared_experts*",
	"denoiser_tower.layers.39*",
	"denoiser_tower.layers.4.*",
	"denoiser_tower.layers.40.mixer.gate",
	"denoiser_tower.layers.40.mixer.shared_experts*",
	"denoiser_tower.layers.41*",
	"denoiser_tower.layers.42*",
	"denoiser_tower.layers.43.mixer.gate",
	"denoiser_tower.layers.43.mixer.shared_experts*",
	"denoiser_tower.layers.44*",
	"denoiser_tower.layers.45.mixer.gate",
	"denoiser_tower.layers.45.mixer.shared_experts*",
	"denoiser_tower.layers.46*",
	"denoiser_tower.layers.47.mixer.gate",
	"denoiser_tower.layers.47.mixer.shared_experts*",
	"denoiser_tower.layers.48*",
	"denoiser_tower.layers.49.mixer.gate",
	"denoiser_tower.layers.49.mixer.shared_experts*",
	"denoiser_tower.layers.5.*",
	"denoiser_tower.layers.50*",
	"denoiser_tower.layers.51.mixer.gate",
	"denoiser_tower.layers.51.mixer.shared_experts*",
	"denoiser_tower.layers.6.mixer.gate",
	"denoiser_tower.layers.6.mixer.shared_experts*",
	"denoiser_tower.layers.7*",
	"denoiser_tower.layers.8.mixer.gate",
	"denoiser_tower.layers.8.mixer.shared_experts*",
	"denoiser_tower.layers.9*",
	"lm_head",
	"t_block*",
	"t_embedder*"
	]
	}
	}