--- license: llama2 language: - ko --- # Llama-2-ko-7b-ggml Llama-2-ko-7b-ggml 은 [beomi/llama-2-ko-7b](https://huggingface.co/beomi/llama-2-ko-7b) 의 **GGML** 포맷 모델입니다. - Llama2 tokenizer 에 [beomi/llama-2-ko-7b](https://huggingface.co/beomi/llama-2-ko-7b) 에서 사용된 한국어 Additaional Token 을 반영하여 생성했습니다. - **GGML** 포맷 모델은 [llama.cpp](https://github.com/ggerganov/llama.cpp) 를 사용하여 C/C++ 기반으로 Inference 합니다. - **GGML** 포맷 모델은 비교적 낮은 사양의 컴퓨팅 자원에서도 Inference 가능합니다. ( 예: 4비트 양자화 모델 (q4) 은 CPU,7-8GB RAM 환경에서 Inference 가능 ) - [llama.cpp](https://github.com/ggerganov/llama.cpp) 의 Python Binding 패키지인 [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) 을 사용하면 python 환경에서도 Inference 가능합니다. 참고로, [Llama-2-ko-7b-chat-ggml](https://huggingface.co/StarFox7/Llama-2-ko-7B-chat-ggml) 에서 [beomi/llama-2-ko-7b](https://huggingface.co/beomi/llama-2-ko-7b) 에 [nlpai-lab/kullm-v2](https://huggingface.co/datasets/nlpai-lab/kullm-v2) 을 추가 학습한 [kfkas/Llama-2-ko-7b-chat](https://huggingface.co/kfkas/Llama-2-ko-7b-Chat) 의 **GGML** 포맷 모델을 찾을 수 있습니다. --- # 양자화 (Quantization) 이 Repository 에는 [llama.cpp](https://github.com/ggerganov/llama.cpp) 에서 제공하는 quantization method 를 적용한 f16, q4_0, q4_1, q5_0, q5_1, q8_0 모델을 포함합니다. 각 모델의 File Size 는 다음과 같습니다. | Model| Measure |q4_0|q4_1|q5_0|q5_1|q8_0|f16|f32| |------|---------|------|------|------|-----|----------|------|-------------| | 7B |file size|3.9G | 4.3G | 4.7G | 5.2G | 7.2G |13.5G|27.4G| --- # Inference Code Example (Python) 다음은 Inference 를 위한 간단한 Example Code 입니다. [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) 그리고 이 Repository 의 Llama-2-ko-7b-ggml-q4_0.bin 가 필요합니다. ```python # !pip install llama-cpp-python # llama-cpp-python 이 설치되어 있지 않다면 주석을 해제하여 설치합니다. from llama_cpp import Llama llm = Llama(model_path = 'Llama-2-ko-7b-ggml-q4_0.bin', n_ctx=1024, # n_gpu_layers=1 #gpu 가속을 원하는 경우 주석을 해제하고 Metal(Apple M1) 은 1, Cuda(Nvidia) 는 Video RAM Size 를 고려하여 적정한 수치를 입력합니다.) ) output = llm("Q: 인생에 대해서 설명하시오. A: ", max_tokens=1024, stop=["Q:", "\n"], echo=True) print( output['choices'][0]['text'].replace('▁',' ') ) #출력 결과 ''' Q: 인생에 대해서 설명하시오. A: 20대에는 모든 것을 할 수 있는 시기로, 자신이 하고 싶은 일과 하고 싶은 공부를 선택해 공부할 수 있고 자신이 이루고 싶은 것들을 성취하고 꿈꿀 수 있는 시기라고 생각했습니다. 이러한 이유로 20대의 저를 설명하라고 한다면 '꿈이 많은 젊은이'가 가장 어울릴 듯합니다. 어렸을 때는 마냥 어른이 되고 싶었고, 중학생 때에는 빨리 고등학교에 올라가고 싶었습니다. 고등학교에 올라가서도 저는 대학에 진학하고 싶은 마음에 공부를 열심히 하였습니다. 대학에 입학한 후에도 저의 공부는 계속되었습니다. 하지만 2학년 정도 되었을 때, '내가 하고 있는 것이 정말 내가 하고 싶은 일일까?' 라는 생각이 들기 시작했습니다. 이런 고민 끝에 저는 방황을 하였고 결국 제가 정말 원하는 일이 무엇인지 몰랐기 때문에 학교를 그만두기로 결심하게 되었습니다. 저는 인생의 목표는 행복이라고 생각했고, 행복한 삶을 살기 위해 모든 것을 할 수 있다는 20대에 제가 하고 싶었던 일을 선택해 제 자신을 발전시키고 성취해가며 꿈을 이루려고 하였습니다. 그래서 저는 '내가 하고 싶은 일과 내가 잘 할 수 있는 일'을 찾기 위해 여러 곳을 둘러보고 경험하였습니다. 여러 곳을 둘러보다 보니 저는 자신이 하고 싶은 일을 찾아내고, 잘 할 수 있다는 자신감을 가지게 되었습니다. 그 후 저는 제가 좋아하는 것을 찾고 성취해가며 꿈을 이루기 위해 노력하고 있습니다. ''' ``` --- > Below is the original model card of the Llama-2 model. # **Llama 2** Llama 2 is a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. This is the repository for the 7B pretrained model, converted for the Hugging Face Transformers format. Links to other models can be found in the index at the bottom. ## Model Details *Note: Use of this model is governed by the Meta license. In order to download the model weights and tokenizer, please visit the [website](https://ai.meta.com/resources/models-and-libraries/llama-downloads/) and accept our License before requesting access here.* Meta developed and publicly released the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama-2-Chat, are optimized for dialogue use cases. Llama-2-Chat models outperform open-source chat models on most benchmarks we tested, and in our human evaluations for helpfulness and safety, are on par with some popular closed-source models like ChatGPT and PaLM. **Model Developers** Meta **Variations** Llama 2 comes in a range of parameter sizes — 7B, 13B, and 70B — as well as pretrained and fine-tuned variations. **Input** Models input text only. **Output** Models generate text only. **Model Architecture** Llama 2 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align to human preferences for helpfulness and safety. ||Training Data|Params|Content Length|GQA|Tokens|LR| |---|---|---|---|---|---|---| |Llama 2|*A new mix of publicly available online data*|7B|4k|✗|2.0T|3.0 x 10-4| |Llama 2|*A new mix of publicly available online data*|13B|4k|✗|2.0T|3.0 x 10-4| |Llama 2|*A new mix of publicly available online data*|70B|4k|✔|2.0T|1.5 x 10-4| *Llama 2 family of models.* Token counts refer to pretraining data only. All models are trained with a global batch-size of 4M tokens. Bigger models - 70B -- use Grouped-Query Attention (GQA) for improved inference scalability. **Model Dates** Llama 2 was trained between January 2023 and July 2023. **Status** This is a static model trained on an offline dataset. Future versions of the tuned models will be released as we improve model safety with community feedback. **License** A custom commercial license is available at: [https://ai.meta.com/resources/models-and-libraries/llama-downloads/](https://ai.meta.com/resources/models-and-libraries/llama-downloads/) **Research Paper** ["Llama-2: Open Foundation and Fine-tuned Chat Models"](arxiv.org/abs/2307.09288) ## Intended Use **Intended Use Cases** Llama 2 is intended for commercial and research use in English. Tuned models are intended for assistant-like chat, whereas pretrained models can be adapted for a variety of natural language generation tasks. To get the expected features and performance for the chat versions, a specific formatting needs to be followed, including the `INST` and `<>` tags, `BOS` and `EOS` tokens, and the whitespaces and breaklines in between (we recommend calling `strip()` on inputs to avoid double-spaces). See our reference code in github for details: [`chat_completion`](https://github.com/facebookresearch/llama/blob/main/llama/generation.py#L212). **Out-of-scope Uses** Use in any manner that violates applicable laws or regulations (including trade compliance laws).Use in languages other than English. Use in any other way that is prohibited by the Acceptable Use Policy and Licensing Agreement for Llama 2. ## Hardware and Software **Training Factors** We used custom training libraries, Meta's Research Super Cluster, and production clusters for pretraining. Fine-tuning, annotation, and evaluation were also performed on third-party cloud compute. **Carbon Footprint** Pretraining utilized a cumulative 3.3M GPU hours of computation on hardware of type A100-80GB (TDP of 350-400W). Estimated total emissions were 539 tCO2eq, 100% of which were offset by Meta’s sustainability program. ||Time (GPU hours)|Power Consumption (W)|Carbon Emitted(tCO2eq)| |---|---|---|---| |Llama 2 7B|184320|400|31.22| |Llama 2 13B|368640|400|62.44| |Llama 2 70B|1720320|400|291.42| |Total|3311616||539.00| **CO2 emissions during pretraining.** Time: total GPU time required for training each model. Power Consumption: peak power capacity per GPU device for the GPUs used adjusted for power usage efficiency. 100% of the emissions are directly offset by Meta's sustainability program, and because we are openly releasing these models, the pretraining costs do not need to be incurred by others. ## Training Data **Overview** Llama 2 was pretrained on 2 trillion tokens of data from publicly available sources. The fine-tuning data includes publicly available instruction datasets, as well as over one million new human-annotated examples. Neither the pretraining nor the fine-tuning datasets include Meta user data. **Data Freshness** The pretraining data has a cutoff of September 2022, but some tuning data is more recent, up to July 2023. ## Evaluation Results In this section, we report the results for the Llama 1 and Llama 2 models on standard academic benchmarks.For all the evaluations, we use our internal evaluations library. |Model|Size|Code|Commonsense Reasoning|World Knowledge|Reading Comprehension|Math|MMLU|BBH|AGI Eval| |---|---|---|---|---|---|---|---|---|---| |Llama 1|7B|14.1|60.8|46.2|58.5|6.95|35.1|30.3|23.9| |Llama 1|13B|18.9|66.1|52.6|62.3|10.9|46.9|37.0|33.9| |Llama 1|33B|26.0|70.0|58.4|67.6|21.4|57.8|39.8|41.7| |Llama 1|65B|30.7|70.7|60.5|68.6|30.8|63.4|43.5|47.6| |Llama 2|7B|16.8|63.9|48.9|61.3|14.6|45.3|32.6|29.3| |Llama 2|13B|24.5|66.9|55.4|65.8|28.7|54.8|39.4|39.1| |Llama 2|70B|**37.5**|**71.9**|**63.6**|**69.4**|**35.2**|**68.9**|**51.2**|**54.2**| **Overall performance on grouped academic benchmarks.** *Code:* We report the average pass@1 scores of our models on HumanEval and MBPP. *Commonsense Reasoning:* We report the average of PIQA, SIQA, HellaSwag, WinoGrande, ARC easy and challenge, OpenBookQA, and CommonsenseQA. We report 7-shot results for CommonSenseQA and 0-shot results for all other benchmarks. *World Knowledge:* We evaluate the 5-shot performance on NaturalQuestions and TriviaQA and report the average. *Reading Comprehension:* For reading comprehension, we report the 0-shot average on SQuAD, QuAC, and BoolQ. *MATH:* We report the average of the GSM8K (8 shot) and MATH (4 shot) benchmarks at top 1. |||TruthfulQA|Toxigen| |---|---|---|---| |Llama 1|7B|27.42|23.00| |Llama 1|13B|41.74|23.08| |Llama 1|33B|44.19|22.57| |Llama 1|65B|48.71|21.77| |Llama 2|7B|33.29|**21.25**| |Llama 2|13B|41.86|26.10| |Llama 2|70B|**50.18**|24.60| **Evaluation of pretrained LLMs on automatic safety benchmarks.** For TruthfulQA, we present the percentage of generations that are both truthful and informative (the higher the better). For ToxiGen, we present the percentage of toxic generations (the smaller the better). |||TruthfulQA|Toxigen| |---|---|---|---| |Llama-2-Chat|7B|57.04|**0.00**| |Llama-2-Chat|13B|62.18|**0.00**| |Llama-2-Chat|70B|**64.14**|0.01| **Evaluation of fine-tuned LLMs on different safety datasets.** Same metric definitions as above. ## Ethical Considerations and Limitations Llama 2 is a new technology that carries risks with use. Testing conducted to date has been in English, and has not covered, nor could it cover all scenarios. For these reasons, as with all LLMs, Llama 2’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate, biased or other objectionable responses to user prompts. Therefore, before deploying any applications of Llama 2, developers should perform safety testing and tuning tailored to their specific applications of the model. Please see the Responsible Use Guide available at [https://ai.meta.com/llama/responsible-use-guide/](https://ai.meta.com/llama/responsible-use-guide) ## Reporting Issues Please report any software “bug,” or other problems with the models through one of the following means: - Reporting issues with the model: [github.com/facebookresearch/llama](http://github.com/facebookresearch/llama) - Reporting problematic content generated by the model: [developers.facebook.com/llama_output_feedback](http://developers.facebook.com/llama_output_feedback) - Reporting bugs and security concerns: [facebook.com/whitehat/info](http://facebook.com/whitehat/info) ## Llama Model Index |Model|Llama2|Llama2-hf|Llama2-chat|Llama2-chat-hf| |---|---|---|---|---| |7B| [Link](https://huggingface.co/llamaste/Llama-2-7b) | [Link](https://huggingface.co/llamaste/Llama-2-7b-hf) | [Link](https://huggingface.co/llamaste/Llama-2-7b-chat) | [Link](https://huggingface.co/llamaste/Llama-2-7b-chat-hf)| |13B| [Link](https://huggingface.co/llamaste/Llama-2-13b) | [Link](https://huggingface.co/llamaste/Llama-2-13b-hf) | [Link](https://huggingface.co/llamaste/Llama-2-13b-chat) | [Link](https://huggingface.co/llamaste/Llama-2-13b-hf)| |70B| [Link](https://huggingface.co/llamaste/Llama-2-70b) | [Link](https://huggingface.co/llamaste/Llama-2-70b-hf) | [Link](https://huggingface.co/llamaste/Llama-2-70b-chat) | [Link](https://huggingface.co/llamaste/Llama-2-70b-hf)|