Instructions to use RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic with Transformers:

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

pipe = pipeline("text-generation", model="RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic")
model = AutoModelForCausalLM.from_pretrained("RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic")
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 RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic

SGLang

How to use RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic 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 "RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic" \
    --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": "RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic",
		"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 "RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic" \
        --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": "RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic with Docker Model Runner:
```
docker model run hf.co/RedHatAI/DeepSeek-R1-Distill-Llama-8B-FP8-dynamic
```

nm-research commited on Feb 6, 2025

Commit

14375d3

verified ·

1 Parent(s): 105bf8d

add v1 evals

Browse files

Files changed (1) hide show

README.md +8 -8

README.md CHANGED Viewed

@@ -141,14 +141,14 @@ lm_eval \
 | Metric                                   | deepseek-ai/DeepSeek-R1-Distill-Llama-8B             | neuralmagic-ent/DeepSeek-R1-Distill-Llama-8B-FP8-Dynamic |
 |-----------------------------------------|:---------------------------------:|:-------------------------------------------:|
-| ARC-Challenge (Acc-Norm, 25-shot)       |                             |                                        |
-| GSM8K (Strict-Match, 5-shot)            |          83.62                   |                                        |
-| HellaSwag (Acc-Norm, 10-shot)           |          77.09                   |             77.13                      |
-| MMLU (Acc, 5-shot)                      |                            |                                     |
-| TruthfulQA (MC2, 0-shot)                |          50.84                   |             50.61                      |
-| Winogrande (Acc, 5-shot)                |          68.03                   |             66.93                           |
-| **Average Score**                       | ****                        | ****                                   |
-| **Recovery (%)**                            | **100.00**                       | ****                                   |
 #### OpenLLM Leaderboard V2 evaluation scores

 | Metric                                   | deepseek-ai/DeepSeek-R1-Distill-Llama-8B             | neuralmagic-ent/DeepSeek-R1-Distill-Llama-8B-FP8-Dynamic |
 |-----------------------------------------|:---------------------------------:|:-------------------------------------------:|
+| ARC-Challenge (Acc-Norm, 25-shot)       |          45.05                   |               44.88                         |
+| GSM8K (Strict-Match, 5-shot)            |          62.77                   |              61.49                          |
+| HellaSwag (Acc-Norm, 10-shot)           |          76.78                   |             76.68                     |
+| MMLU (Acc, 5-shot)                      |          55.65                  |               55.82                      |
+| TruthfulQA (MC2, 0-shot)                |          50.55                   |             49.92                      |
+| Winogrande (Acc, 5-shot)                |          68.51                   |             67.72                           |
+| **Average Score**                       | **59.88**                        | **59.42**                                   |
+| **Recovery (%)**                            | **100.00**                       | **99.22**                                   |
 #### OpenLLM Leaderboard V2 evaluation scores