README.md

---
license: apache-2.0
language:
- en
tags:
- deepseek
- fp8
- vllm
base_model: deepseek-ai/DeepSeek-R1-Distill-Llama-70B
library_name: transformers
---

# DeepSeek-R1-Distill-Llama-70B-FP8-Dynamic

## Model Overview
- **Model Architecture:** DeepSeek-R1-Distill-Llama-70B
  - **Input:** Text
  - **Output:** Text
- **Model Optimizations:**
  - **Weight quantization:** FP8
  - **Activation quantization:** FP8
- **Release Date:** 3/1/2025
- **Version:** 1.0
- **Model Developers:** Neural Magic

Quantized version of [DeepSeek-R1-Distill-Llama-70B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B).
It achieves an average score of 76.52 on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) benchmark (version 1), whereas the unquantized model achieves 76.49.

### Model Optimizations

This model was obtained by quantizing the weights and activations to FP8 data type, ready for inference with vLLM.
This optimization reduces the number of bits per parameter from 16 to 8, reducing the disk size and GPU memory requirements by approximately 50%. Only the weights and activations of the linear operators within transformers blocks are quantized.

## Deployment

### Use with vLLM

This model can be deployed efficiently using the [vLLM](https://docs.vllm.ai/en/latest/) backend, as shown in the example below.

```python
from transformers import AutoTokenizer
from vllm import LLM, SamplingParams

max_model_len, tp_size = 4096, 1
model_name = "nm-testing/DeepSeek-R1-Distill-Llama-70B-FP8-Dynamic"
tokenizer = AutoTokenizer.from_pretrained(model_name)
llm = LLM(model=model_name, tensor_parallel_size=tp_size, max_model_len=max_model_len, trust_remote_code=True)
sampling_params = SamplingParams(temperature=0.3, max_tokens=256, stop_token_ids=[tokenizer.eos_token_id])

messages_list = [
    [{"role": "user", "content": "Who are you? Please respond in pirate speak!"}],
]

prompt_token_ids = [tokenizer.apply_chat_template(messages, add_generation_prompt=True) for messages in messages_list]

outputs = llm.generate(prompt_token_ids=prompt_token_ids, sampling_params=sampling_params)

generated_text = [output.outputs[0].text for output in outputs]
print(generated_text)
```

vLLM also supports OpenAI-compatible serving. See the [documentation](https://docs.vllm.ai/en/latest/) for more details.

## Creation

This model was created with [llm-compressor](https://github.com/vllm-project/llm-compressor) by running the code snippet below.


```python
import argparse
from transformers import AutoModelForCausalLM, AutoTokenizer
from llmcompressor.modifiers.quantization import QuantizationModifier
from llmcompressor.transformers import oneshot
import os

def main():
    parser = argparse.ArgumentParser(description='Quantize a transformer model to FP8')
    parser.add_argument('--model_id', type=str, required=True,
                        help='The model ID from HuggingFace (e.g., "meta-llama/Meta-Llama-3-8B-Instruct")')
    parser.add_argument('--save_path', type=str, default='.',
                        help='Custom path to save the quantized model. If not provided, will use model_name-FP8-dynamic')
    args = parser.parse_args()

    # Load model
    model = AutoModelForCausalLM.from_pretrained(
        args.model_id, device_map="auto", torch_dtype="auto", trust_remote_code=True,
    )
    tokenizer = AutoTokenizer.from_pretrained(args.model_id)

    # Configure the quantization algorithm and scheme
    recipe = QuantizationModifier(
        targets="Linear", scheme="FP8_DYNAMIC", ignore=["lm_head"]
    )

    # Apply quantization
    oneshot(model=model, recipe=recipe)

    save_path = os.path.join(args.save_path, args.model_id.split("/")[1] + "-FP8-dynamic")
    os.makedirs(save_path, exist_ok=True)

    # Save to disk in compressed-tensors format
    model.save_pretrained(save_path)
    tokenizer.save_pretrained(save_path)
    print(f"Model and tokenizer saved to: {save_path}")

if __name__ == "__main__":
    main()
```

## Evaluation

The model was evaluated on OpenLLM Leaderboard [V1](https://huggingface.co/spaces/open-llm-leaderboard-old/open_llm_leaderboard) and [V2](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/), using the following commands:

OpenLLM Leaderboard V1:
```
lm_eval \
  --model vllm \
  --model_args pretrained="nm-testing/DeepSeek-R1-Distill-Llama-70B-FP8-Dynamic",dtype=auto,add_bos_token=True,max_model_len=4096,tensor_parallel_size=2,gpu_memory_utilization=0.8,enable_chunked_prefill=True,trust_remote_code=True \
  --tasks openllm \
  --write_out \
  --batch_size auto \
  --output_path output_dir \
  --show_config
```

OpenLLM Leaderboard V2:
```
lm_eval \
  --model vllm \
  --model_args pretrained="nm-testing/DeepSeek-R1-Distill-Llama-70B-FP8-Dynamic",dtype=auto,add_bos_token=False,max_model_len=4096,tensor_parallel_size=2,gpu_memory_utilization=0.8,enable_chunked_prefill=True,trust_remote_code=True \
  --apply_chat_template \
  --fewshot_as_multiturn \
  --tasks leaderboard \
  --write_out \
  --batch_size auto \
  --output_path output_dir \
  --show_config

```

### Accuracy

#### OpenLLM Leaderboard V1 evaluation scores

| Metric                                   | deepseek-ai/DeepSeek-R1-Distill-Llama-70B             | nm-testing/DeepSeek-R1-Distill-Llama-70B-FP8-Dynamic |
|-----------------------------------------|:---------------------------------:|:-------------------------------------------:|
| ARC-Challenge (Acc-Norm, 25-shot)       |    66.38                          |   66.38                                     |
| GSM8K (Strict-Match, 5-shot)            |    92.87                         |    93.25                                     |
| HellaSwag (Acc-Norm, 10-shot)           |    85.41                          |   85.40                                     |
| MMLU (Acc, 5-shot)                      |    79.02                          |   78.84                                     |
| TruthfulQA (MC2, 0-shot)                |    57.24                          |   57.54                                     |
| Winogrande (Acc, 5-shot)                |    78.06                          |   77.74                                     |
| **Average Score**                       | **76.49**                        | **76.52**                                   |
| **Recovery (%)**                            | **100.00**                       | **100.03**                                   |

#### OpenLLM Leaderboard V2 evaluation scores


| Metric                                                   | deepseek-ai/DeepSeek-R1-Distill-Llama-70B             | nm-testing/DeepSeek-R1-Distill-Llama-70B-FP8-Dynamic |
|---------------------------------------------------------|:---------------------------------:|:-------------------------------------------:|
| IFEval (Inst-and-Prompt Level Strict Acc, 0-shot)       |    43.51                         |    42.47                                    |
| BBH (Acc-Norm, 3-shot)                                  |    35.30                         |    33.66                                    |
| MMLU-Pro (Acc, 5-shot)                                  |    41.35                         |    41.05                                    |
| **Average Score**                                       | **40.05**                        | **39.06**                                   |
| **Recovery (%)**                                        | **100.00**                  | **97.53**                                   |
| Math-Hard (Exact-Match, 4-shot)                         |    5.55                         |    9.03                                    |
| GPQA (Acc-Norm, 0-shot)                                 |    1.64                         |    1.58                                    |
| MUSR (Acc-Norm, 0-shot)                                 |    13.28                         |    13.80                                    |

Results on Math-Hard, GPQA, and MUSR are not considred for accuracy recovery calculation because the unquantized model has close to random prediction accuracy which doesn't provide a reliable baseline for recovery calculation.