--- license: mit tags: - deepseek - fp8 - vllm base_model: deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B library_name: transformers --- # DeepSeek-R1-Distill-Qwen-1.5B-FP8-dynamic ## Model Overview - **Model Architecture:** Qwen2ForCausalLM - **Input:** Text - **Output:** Text - **Model Optimizations:** - **Weight quantization:** FP8 - **Activation quantization:** FP8 - **Release Date:** 2/5/2025 - **Version:** 1.0 - **Model Developers:** Neural Magic Quantized version of [DeepSeek-R1-Distill-Qwen-1.5B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B). ### Model Optimizations This model was obtained by quantizing the weights and activations of [DeepSeek-R1-Distill-Qwen-1.5B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B) to FP8 data type. 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. Weights are quantized using a symmetric per-channel scheme, whereas quantizations are quantized using a symmetric per-token scheme. [LLM Compressor](https://github.com/vllm-project/llm-compressor) is used for quantization. ## 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 number_gpus = 1 model_name = "neuralmagic/DeepSeek-R1-Distill-Qwen-1.5B-dynamic" tokenizer = AutoTokenizer.from_pretrained(model_name) sampling_params = SamplingParams(temperature=0.6, max_tokens=256, stop_token_ids=[tokenizer.eos_token_id]) llm = LLM(model=model_name, tensor_parallel_size=number_gpus, trust_remote_code=True) 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 from transformers import AutoModelForCausalLM, AutoTokenizer from llmcompressor.modifiers.quantization import QuantizationModifier from llmcompressor.transformers import oneshot import os # Load model model_stub = "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B" model_name = model_stub.split("/")[-1] model = AutoModelForCausalLM.from_pretrained( model_stub, torch_dtype="auto", ) tokenizer = AutoTokenizer.from_pretrained(model_stub) # Configure the quantization algorithm and scheme recipe = QuantizationModifier( targets="Linear", scheme="FP8_DYNAMIC", ignore=["lm_head"], ) # Apply quantization oneshot( model=model, recipe=recipe, ) # Save to disk in compressed-tensors format save_path = model_name + "-FP8-dynamic model.save_pretrained(save_path) tokenizer.save_pretrained(save_path) print(f"Model and tokenizer saved to: {save_path}") ``` ## 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="neuralmagic/DeepSeek-R1-Distill-Qwen-1.5B-FP8-dynamic",dtype=auto,max_model_len=4096,tensor_parallel_size=1,enable_chunked_prefill=True \ --tasks openllm \ --write_out \ --batch_size auto \ --output_path output_dir \ --show_config ``` OpenLLM Leaderboard V2: ``` lm_eval \ --model vllm \ --model_args pretrained="neuralmagic/DeepSeek-R1-Distill-Qwen-1.5B-FP8-dynamic",dtype=auto,max_model_len=4096,tensor_parallel_size=1,enable_chunked_prefill=True \ --apply_chat_template \ --fewshot_as_multiturn \ --tasks leaderboard \ --write_out \ --batch_size auto \ --output_path output_dir \ --show_config ``` ### Accuracy
Category Metric deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B neuralmagic/DeepSeek-R1-Distill-Qwen-1.5B-FP8-dynamic Recovery
OpenLLM V1 ARC-Challenge (Acc-Norm, 25-shot) 37.02 37.71 101.4%
GSM8K (Strict-Match, 5-shot) 69.98 68.99 98.6%
HellaSwag (Acc-Norm, 10-shot) 43.86 43.61 99.4%
MMLU (Acc, 5-shot) 37.38 37.22 99.6%
TruthfulQA (MC2, 0-shot) 45.21 44.77 99.0%
Winogrande (Acc, 5-shot) 54.30 54.62 100.6%
Average Score 47.99 47.82 99.7%
OpenLLM V2 IFEval (Inst Level Strict Acc, 0-shot) 34.37 34.91 101.6%
BBH (Acc-Norm, 3-shot) 34.44 34.40 99.9%
Math-Hard (Exact-Match, 4-shot) 0.00 0.00 ---
GPQA (Acc-Norm, 0-shot) 24.67 25.16 102.0%
MUSR (Acc-Norm, 0-shot) 35.82 36.61 102.2%
MMLU-Pro (Acc, 5-shot) 11.80 11.69 99.1%
Average Score 23.52 23.79 101.2%
Coding HumanEval (pass@1) 37.90 36.40 96.0%
HumanEval (pass@10) 61.30 61.30 100.0%
HumanEval+ (pass@10) 33.00 32.60 98.8%
HumanEval+ (pass@10) 55.90 56.30 100.7%