Instructions to use WaveCut/YuE2-3B-OrbitQuant-W4A4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use WaveCut/YuE2-3B-OrbitQuant-W4A4 with Transformers:

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

pipe = pipeline("text-generation", model="WaveCut/YuE2-3B-OrbitQuant-W4A4")

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("WaveCut/YuE2-3B-OrbitQuant-W4A4", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use WaveCut/YuE2-3B-OrbitQuant-W4A4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "WaveCut/YuE2-3B-OrbitQuant-W4A4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WaveCut/YuE2-3B-OrbitQuant-W4A4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/WaveCut/YuE2-3B-OrbitQuant-W4A4

SGLang

How to use WaveCut/YuE2-3B-OrbitQuant-W4A4 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 "WaveCut/YuE2-3B-OrbitQuant-W4A4" \
    --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": "WaveCut/YuE2-3B-OrbitQuant-W4A4",
		"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 "WaveCut/YuE2-3B-OrbitQuant-W4A4" \
        --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": "WaveCut/YuE2-3B-OrbitQuant-W4A4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use WaveCut/YuE2-3B-OrbitQuant-W4A4 with Docker Model Runner:
```
docker model run hf.co/WaveCut/YuE2-3B-OrbitQuant-W4A4
```

YuE2-3B-OrbitQuant-W4A4 / src /kernel-source /orbitquant-native /torch-ext /torch_binding.cpp

WaveCut

Update packed inference to OrbitQuant0.9.6 and native1.0.3

f0c91ec verified 2 days ago

Raw

History Blame Contribute Delete

2.96 kB

	#include "registration.h"
	#include "torch_binding.h"

	#if defined(CPU_KERNEL)
	#include <torch/csrc/stable/library.h>

	STABLE_TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, ops) {
	ops.def(
	"matmul_packed_weight(Tensor! out, Tensor x, Tensor packed_weight_indices, "
	"Tensor row_norms, Tensor centroids, Tensor bias, bool has_bias, int bits, "
	"int out_features, int in_features, int block_m, int block_n, int block_k) -> ()");
	ops.def(
	"quantize_activations_cpu(Tensor! out, Tensor x, Tensor permutation, "
	"Tensor signs, Tensor centroids, Tensor boundaries, float eps, "
	"float inv_sqrt_block, int block_size) -> ()");
	ops.def(
	"matmul_packed_adaln_int4_cpu(Tensor! out, Tensor x, Tensor packed_weight, "
	"Tensor scales, Tensor bias, bool has_bias, int out_features, "
	"int in_features, int group_size) -> ()");
	}

	STABLE_TORCH_LIBRARY_IMPL_EXPAND(TORCH_EXTENSION_NAME, CPU, ops) {
	ops.impl("matmul_packed_weight", TORCH_BOX(&matmul_packed_weight));
	ops.impl("quantize_activations_cpu", TORCH_BOX(&quantize_activations_cpu));
	ops.impl(
	"matmul_packed_adaln_int4_cpu",
	TORCH_BOX(&matmul_packed_adaln_int4_cpu));
	}
	#else
	#include <torch/library.h>

	TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, ops) {
	ops.def(
	"matmul_packed_weight(Tensor! out, Tensor x, Tensor packed_weight_indices, "
	"Tensor row_norms, Tensor centroids, Tensor bias, bool has_bias, int bits, "
	"int out_features, int in_features, int block_m, int block_n, int block_k) -> ()");
	#if defined(CUDA_KERNEL)
	ops.impl("matmul_packed_weight", torch::kCUDA, &matmul_packed_weight);
	ops.def(
	"matmul_packed_w4a4_int8(Tensor! out, Tensor packed_activations, "
	"Tensor packed_weight_indices, Tensor token_norms, Tensor row_norms, "
	"Tensor activation_codes, Tensor weight_codes, Tensor bias, bool has_bias, "
	"float activation_scale, float weight_scale, int out_features, "
	"int in_features, int tile_m, int tile_n, bool async_packed, "
	"bool weight_k_major) -> ()");
	ops.impl("matmul_packed_w4a4_int8", torch::kCUDA, &matmul_packed_w4a4_int8);
	ops.def(
	"quantize_activations_packed_w4(Tensor! packed_out, Tensor! norms_out, "
	"Tensor x, Tensor permutation, Tensor signs, Tensor boundaries, float eps, "
	"float inv_sqrt_block, int threads) -> ()");
	ops.impl(
	"quantize_activations_packed_w4",
	torch::kCUDA,
	&quantize_activations_packed_w4);
	ops.def(
	"quantize_activations_int8(Tensor! int8_out, Tensor! norms_out, Tensor x, "
	"Tensor permutation, Tensor signs, Tensor boundaries, Tensor codes, float eps, "
	"float inv_sqrt_block, int threads) -> ()");
	ops.impl(
	"quantize_activations_int8",
	torch::kCUDA,
	&quantize_activations_int8);
	#elif defined(METAL_KERNEL)
	ops.impl("matmul_packed_weight", torch::kMPS, &matmul_packed_weight);
	#endif
	}
	#endif

	REGISTER_EXTENSION(TORCH_EXTENSION_NAME)