--- title: Qwen2-0.5B-Instruct-FP8 canonical_url: "https://www.modelscope.cn/models/QwenCollection/Qwen2-0.5B-Instruct-FP8" md_url: "https://www.modelscope.cn/models/QwenCollection/Qwen2-0.5B-Instruct-FP8.md" repository: QwenCollection/Qwen2-0.5B-Instruct-FP8 last_updated: 2024-11-21 license: apache-2.0 pipeline_tag: text-generation tasks: - text-generation model_type: - qwen2 architectures: - Qwen2ForCausalLM parameters: 494.0M tensor_type: - BF16 - F32 - F8_E4M3 library_name: - transformer - safetensors - pytorch frameworks: - pytorch inference_backends: - "deploy_task text/emb" - "lmdeploy 0.9.1" - "lmdeploy_turbomind 0.9.1" - "sglang 0.5.2" - "vllm 0.9.2" downloads: 2434 stars: 0 tags: - fp8 - vllm --- # Qwen2-0.5B-Instruct-FP8 > Qwen2-0.5B-Instruct-FP8 - QwenCollection 在 ModelScope 开源的模型。Qwen2-0.5B-Instruct-FP8 QwenCollection/Qwen2-0.5B-Instruct-FP8 是 ModelScope 魔搭社区上的 494.0M 参数text-generation模型，采用 apache-2.0 许可，可用 deploy_task text/emb、lmdeploy 0.9.1、lmdeploy_turbomind 0.9.1 部署。 - **Repository**: QwenCollection/Qwen2-0.5B-Instruct-FP8 - **License**: apache-2.0 - **Tasks**: text-generation - **Parameters**: 494.0M - **Inference backends**: deploy_task text/emb, lmdeploy 0.9.1, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2 - **Tags**: fp8, vllm - **Downloads**: 2434 - **Stars**: 0 - **Last updated**: 2024-11-21 Source: https://www.modelscope.cn/models/QwenCollection/Qwen2-0.5B-Instruct-FP8 --- # Qwen2-0.5B-Instruct-FP8 ## Model Overview - **Model Architecture:** Qwen2 - **Input:** Text - **Output:** Text - **Model Optimizations:** - **Weight quantization:** FP8 - **Activation quantization:** FP8 - **Intended Use Cases:** Intended for commercial and research use in English. Similarly to [Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct), this models is intended for assistant-like chat. - **Out-of-scope:** Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English. - **Release Date:** 6/14/2024 - **Version:** 1.0 - **License(s):** [apache-2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) - **Model Developers:** Neural Magic Quantized version of [Qwen2-0.5B-Instruct](https://huggingface.co/Qwen/Qwen2-0.5B-Instruct). It achieves an average score of 42.94 on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) benchmark (version 1), whereas the unquantized model achieves 42.96. ### Model Optimizations This model was obtained by quantizing the weights and activations of [Qwen2-0.5B-Instruct](https://huggingface.co/Qwen/Qwen2-0.5B-Instruct) to FP8 data type, ready for inference with vLLM >= 0.5.0. 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. Symmetric per-tensor quantization is applied, in which a single linear scaling maps the FP8 representations of the quantized weights and activations. [AutoFP8](https://github.com/neuralmagic/AutoFP8) is used for quantization with 512 sequences of UltraChat. ## 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 vllm import LLM, SamplingParams from transformers import AutoTokenizer model_id = "neuralmagic/Qwen2-0.5B-Instruct-FP8" sampling_params = SamplingParams(temperature=0.6, top_p=0.9, max_tokens=256) tokenizer = AutoTokenizer.from_pretrained(model_id) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Who are you?"}, ] prompts = tokenizer.apply_chat_template(messages, tokenize=False) llm = LLM(model=model_id) outputs = llm.generate(prompts, sampling_params) generated_text = outputs[0].outputs[0].text print(generated_text) ``` vLLM aslo supports OpenAI-compatible serving. See the [documentation](https://docs.vllm.ai/en/latest/) for more details. ## Creation This model was created by applying [AutoFP8 with calibration samples from ultrachat](https://github.com/neuralmagic/AutoFP8/blob/147fa4d9e1a90ef8a93f96fc7d9c33056ddc017a/example_dataset.py), as presented in the code snipet below. Although AutoFP8 was used for this particular model, Neural Magic is transitioning to using [llm-compressor](https://github.com/vllm-project/llm-compressor) which supports several quantization schemes and models not supported by AutoFP8. ```python from datasets import load_dataset from transformers import AutoTokenizer from auto_fp8 import AutoFP8ForCausalLM, BaseQuantizeConfig pretrained_model_dir = "Qwen/Qwen2-0.5B-Instruct" quantized_model_dir = "Qwen2-0.5B-Instruct-FP8" tokenizer = AutoTokenizer.from_pretrained(pretrained_model_dir, use_fast=True, model_max_length=4096) tokenizer.pad_token = tokenizer.eos_token ds = load_dataset("mgoin/ultrachat_2k", split="train_sft").select(range(512)) examples = [tokenizer.apply_chat_template(batch["messages"], tokenize=False) for batch in ds] examples = tokenizer(examples, padding=True, truncation=True, return_tensors="pt").to("cuda") quantize_config = BaseQuantizeConfig( quant_method="fp8", activation_scheme="static" ignore_patterns=["re:.*lm_head"], ) model = AutoFP8ForCausalLM.from_pretrained( pretrained_model_dir, quantize_config=quantize_config ) model.quantize(examples) model.save_quantized(quantized_model_dir) ``` ## Evaluation The model was evaluated on the [OpenLLM](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard) leaderboard tasks (version 1) with the [lm-evaluation-harness](https://github.com/EleutherAI/lm-evaluation-harness/tree/383bbd54bc621086e05aa1b030d8d4d5635b25e6) (commit 383bbd54bc621086e05aa1b030d8d4d5635b25e6) and the [vLLM](https://docs.vllm.ai/en/stable/) engine, using the following command: ``` lm_eval \ --model vllm \ --model_args pretrained="neuralmagic/Qwen2-0.5B-Instruct-FP8",dtype=auto,gpu_memory_utilization=0.4,add_bos_token=True,max_model_len=4096 \ --tasks openllm \ --batch_size auto ``` ### Accuracy #### Open LLM Leaderboard evaluation scores
Benchmark Qwen2-0.5B-Instruct Qwen2-0.5B-Instruct-FP8(this model) Recovery
MMLU (5-shot) 43.87 43.63 99.45%
ARC Challenge (25-shot) 31.74 32.00 100.8%
GSM-8K (5-shot, strict-match) 37.83 36.85 97.40%
Hellaswag (10-shot) 49.45 49.21 99.51%
Winogrande (5-shot) 55.49 56.59 101.9%
TruthfulQA (0-shot) 39.37 39.33 99.89%
Average 42.96 42.94 99.95%