Instructions to use nvidia/Hymba-1.5B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nvidia/Hymba-1.5B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="nvidia/Hymba-1.5B-Instruct", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("nvidia/Hymba-1.5B-Instruct", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use nvidia/Hymba-1.5B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nvidia/Hymba-1.5B-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/Hymba-1.5B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nvidia/Hymba-1.5B-Instruct

SGLang

How to use nvidia/Hymba-1.5B-Instruct 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 "nvidia/Hymba-1.5B-Instruct" \
    --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": "nvidia/Hymba-1.5B-Instruct",
		"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 "nvidia/Hymba-1.5B-Instruct" \
        --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": "nvidia/Hymba-1.5B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nvidia/Hymba-1.5B-Instruct with Docker Model Runner:
```
docker model run hf.co/nvidia/Hymba-1.5B-Instruct
```

YongganFu commited on Dec 19, 2024

Commit

06f4e52

verified ·

1 Parent(s): e9aa8ad

Update modeling_hymba.py

Browse files

Files changed (1) hide show

modeling_hymba.py +0 -70

modeling_hymba.py CHANGED Viewed

@@ -1714,76 +1714,6 @@ class HymbaBlock(nn.Module):
                     if ssm_state is not None and cache_params is not None:
                         cache_params.ssm_states[self.layer_idx].copy_(ssm_state)
-            # if use_precomputed_states and self.layer_idx==31:
-            # # except Exception as e:
-            #     print("\n\n\n\n")
-            #     # print(e)
-            #     print(f"use_precomputed_states {use_precomputed_states}; {index} {self.D}, {time_proj_bias} ")
-            #     print(f"{self.D[index]} ")
-            #     # cache_params.ssm_states[self.layer_idx],
-            #     #         hidden_states[..., 0],
-            #     #         discrete_time_step[..., 0],
-            #     #         A,
-            #     #         B[:, 0],
-            #     #         C[:, 0],
-            #     #         self.D[index],
-            #     #         gate[..., 0],
-            #     #         time_proj_bias,
-            #     print("=== Variable Values ===")
-            #     try:
-            #         print(f"cache_params.ssm_states[{self.layer_idx}]: {cache_params.ssm_states[self.layer_idx]}")
-            #         print(f"{cache_params.ssm_states[self.layer_idx].shape}")
-            #     except Exception as e:
-            #         print(f"Error accessing cache_params.ssm_states[{self.layer_idx}]: {e}")
-            #     try:
-            #         print(f"hidden_states[..., 0]: {hidden_states[..., 0]}")
-            #         print(f"hidden_states[..., 0] shape: {hidden_states[..., 0].shape}")
-            #     except Exception as e:
-            #         print(f"Error accessing hidden_states[..., 0]: {e}")
-            #     try:
-            #         print(f"discrete_time_step[..., 0]: {discrete_time_step[..., 0]}")
-            #         print(f"discrete_time_step[..., 0].shape: {discrete_time_step[..., 0].shape}")
-            #     except Exception as e:
-            #         print(f"Error accessing discrete_time_step[..., 0]: {e}")
-            #     try:
-            #         print(f"A: {A}")
-            #         print(f"A.shape: {A.shape}")
-            #     except Exception as e:
-            #         print(f"Error accessing A: {e}")
-            #     try:
-            #         print(f"B[:, 0]: {B[:, 0].shape}")
-            #         print(f"B[:, 0].shape: {B[:, 0].shape}")
-            #     except Exception as e:
-            #         print(f"Error accessing B[:, 0]: {e}")
-            #     try:
-            #         print(f"C[:, 0]: {C[:, 0]}")
-            #         print(f"C[:, 0].shape: {C[:, 0].shape}")
-            #     except Exception as e:
-            #         print(f"Error accessing C[:, 0]: {e}")
-            #     try:
-            #         print(f"D[index]: {self.D[index]}")
-            #         print(f"D[index].shape: {self.D[index].shape}")
-            #     except Exception as e:
-            #         print(f"Error accessing D[{index}]: {e}")
-            #     try:
-            #         print(f"gate[..., 0]: {gate[..., 0]}")
-            #         print(f"gate[..., 0].shape: {gate[..., 0].shape}")
-            #     except Exception as e:
-            #         print(f"Error accessing gate[..., 0]: {e}")
-            #     try:
-            #         print(f"time_proj_bias: {time_proj_bias}")
-            #     except Exception as e:
-            #         print(f"Error accessing time_proj_bias: {e}")
-            #     print("\n\n\n\n")
             scan_outputs = scan_outputs.transpose(1, 2)

                     if ssm_state is not None and cache_params is not None:
                         cache_params.ssm_states[self.layer_idx].copy_(ssm_state)
             scan_outputs = scan_outputs.transpose(1, 2)