Instructions to use tencent/Hy4-preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tencent/Hy4-preview with Transformers:

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

pipe = pipeline("text-generation", model="tencent/Hy4-preview")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tencent/Hy4-preview")
model = AutoModelForCausalLM.from_pretrained("tencent/Hy4-preview", device_map="auto")
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 Settings

vLLM

How to use tencent/Hy4-preview with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tencent/Hy4-preview"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tencent/Hy4-preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tencent/Hy4-preview

SGLang

How to use tencent/Hy4-preview 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 "tencent/Hy4-preview" \
    --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": "tencent/Hy4-preview",
		"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 "tencent/Hy4-preview" \
        --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": "tencent/Hy4-preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tencent/Hy4-preview with Docker Model Runner:
```
docker model run hf.co/tencent/Hy4-preview
```

yiqichen01 commited on 20 days ago

Commit

705d81e

verified ·

1 Parent(s): ce1ba0b

update

Browse files

Files changed (2) hide show

finetune/README.md +0 -1
finetune/README_CN.md +2 -2

finetune/README.md CHANGED Viewed

@@ -143,7 +143,6 @@ The key parameters in the script are as follows:
 - `--max_steps`: Total number of training steps.
 - `--save_steps`: Number of steps between saving checkpoints.
 - `--use_lora`: Whether to use LoRA training. Also accepts `--lora_rank`, `--lora_alpha`, and `--lora_dropout` parameters. By default, LoRA is applied to "q_proj", "k_proj", "v_proj", and "o_proj". To change this, modify the code. Note: **When using LoRA training, only the LoRA weights are saved, not the base model weights.**
-- `--make_moe_param_leaf_module`: When using ZeRO-3 with MoE training, treat the MoE module as a leaf module, i.e., its parameters are not partitioned by ZeRO-3. This option is expected to significantly increase memory usage.
 - `--gradient_checkpointing`: Enable gradient checkpointing.
 - `--learning_rate`: Maximum learning rate during training.
 - `--min_lr`: Minimum learning rate during training.

 - `--max_steps`: Total number of training steps.
 - `--save_steps`: Number of steps between saving checkpoints.
 - `--use_lora`: Whether to use LoRA training. Also accepts `--lora_rank`, `--lora_alpha`, and `--lora_dropout` parameters. By default, LoRA is applied to "q_proj", "k_proj", "v_proj", and "o_proj". To change this, modify the code. Note: **When using LoRA training, only the LoRA weights are saved, not the base model weights.**
 - `--gradient_checkpointing`: Enable gradient checkpointing.
 - `--learning_rate`: Maximum learning rate during training.
 - `--min_lr`: Minimum learning rate during training.

finetune/README_CN.md CHANGED Viewed

@@ -143,8 +143,8 @@ Are you sure you want to continue connecting (yes/no)?
 - `--max_steps`: 训练的总步数
 - `--save_steps`: 每多少个 step 存储一个 checkpoint
 - `--use_lora`: 是否用 lora 训练，同时接收`--lora_rank`，`--lora_alpha`和`--lora_dropout`参数。lora 默认应用于 MLA（Multi-head Latent Attention）投影层："q_a_proj", "q_b_proj", "kv_a_proj_with_mqa", "kv_b_proj", "o_proj"，如果需要改变的话在代码中修改即可。注意：**使用 lora 训练时，只会保存 lora 的权重，而不会保存 base 模型的权重**
-- `--make_moe_param_leaf_module`：当用 zero3 以及 MoE 训练时，将 MoE 模块视作一个 leaf module，即它的参数不进行 zero3 切分，这个选项预计会显著增加显存占用
-- `--gradient_checkpointing`：开启梯度检查点- `--learning_rate`: 训练时的最大学习率
 - `--min_lr`: 训练时的最小学习率
 - `--use_flash_attn`: 开启 flash-attention 进行训练加速

 - `--max_steps`: 训练的总步数
 - `--save_steps`: 每多少个 step 存储一个 checkpoint
 - `--use_lora`: 是否用 lora 训练，同时接收`--lora_rank`，`--lora_alpha`和`--lora_dropout`参数。lora 默认应用于 MLA（Multi-head Latent Attention）投影层："q_a_proj", "q_b_proj", "kv_a_proj_with_mqa", "kv_b_proj", "o_proj"，如果需要改变的话在代码中修改即可。注意：**使用 lora 训练时，只会保存 lora 的权重，而不会保存 base 模型的权重**
+- `--gradient_checkpointing`：开启梯度检查点
+- `--learning_rate`: 训练时的最大学习率
 - `--min_lr`: 训练时的最小学习率
 - `--use_flash_attn`: 开启 flash-attention 进行训练加速