richarderkhov/mlabonne_-_neuralbeagle14-7b-gguf overview

{
  "metadata": {},
  "card_data": {
    "frontmatter": {},
    "hero_image_url": "https://raw.githubusercontent.com/argilla-io/distilabel/main/docs/assets/distilabel-badge-light.png",
    "summary": "**Update 01/16/24: NeuralBeagle14-7B is (probably) the best 7B model you can find! 🎉** NeuralBeagle14-7B is a DPO fine-tune of mlabonne/Beagle14-7B using the argilla/distilabel-intel-orca-dpo-pairs preference dataset and my DPO notebook from this article. It is based on a merge of the following models using LazyMergekit: * fblgit/UNA-TheBeagle-7b-v1, based on jondurbin's repo and jondurbin/bagel-v0.3 * argilla/distilabeled-Marcoro14-7B-slerp, based on mlabonne/Marcoro14-7B-slerp Thanks Argilla for providing the dataset and the training recipe here. 💪 You can try it out in this Space (GGUF Q4_K_M).",
    "quick_links": [],
    "benchmark_table_html": "",
    "readme_markdown": "Quantization made by Richard Erkhov.\n\n[Github](https://github.com/RichardErkhov)\n\n[Discord](https://discord.gg/pvy7H8DZMG)\n\n[Request more models](https://github.com/RichardErkhov/quant_request)\n\n\nNeuralBeagle14-7B - GGUF\n- Model creator: https://huggingface.co/mlabonne/\n- Original model: https://huggingface.co/mlabonne/NeuralBeagle14-7B/\n\n\n| Name | Quant method | Size |\n| ---- | ---- | ---- |\n| [NeuralBeagle14-7B.Q2_K.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q2_K.gguf) | Q2_K | 2.53GB |\n| [NeuralBeagle14-7B.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.IQ3_XS.gguf) | IQ3_XS | 2.81GB |\n| [NeuralBeagle14-7B.IQ3_S.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.IQ3_S.gguf) | IQ3_S | 2.96GB |\n| [NeuralBeagle14-7B.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q3_K_S.gguf) | Q3_K_S | 2.95GB |\n| [NeuralBeagle14-7B.IQ3_M.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.IQ3_M.gguf) | IQ3_M | 3.06GB |\n| [NeuralBeagle14-7B.Q3_K.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q3_K.gguf) | Q3_K | 3.28GB |\n| [NeuralBeagle14-7B.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q3_K_M.gguf) | Q3_K_M | 3.28GB |\n| [NeuralBeagle14-7B.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q3_K_L.gguf) | Q3_K_L | 3.56GB |\n| [NeuralBeagle14-7B.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.IQ4_XS.gguf) | IQ4_XS | 3.67GB |\n| [NeuralBeagle14-7B.Q4_0.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q4_0.gguf) | Q4_0 | 3.83GB |\n| [NeuralBeagle14-7B.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.IQ4_NL.gguf) | IQ4_NL | 3.87GB |\n| [NeuralBeagle14-7B.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q4_K_S.gguf) | Q4_K_S | 3.86GB |\n| [NeuralBeagle14-7B.Q4_K.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q4_K.gguf) | Q4_K | 4.07GB |\n| [NeuralBeagle14-7B.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q4_K_M.gguf) | Q4_K_M | 4.07GB |\n| [NeuralBeagle14-7B.Q4_1.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q4_1.gguf) | Q4_1 | 4.24GB |\n| [NeuralBeagle14-7B.Q5_0.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q5_0.gguf) | Q5_0 | 4.65GB |\n| [NeuralBeagle14-7B.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q5_K_S.gguf) | Q5_K_S | 4.65GB |\n| [NeuralBeagle14-7B.Q5_K.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q5_K.gguf) | Q5_K | 4.78GB |\n| [NeuralBeagle14-7B.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q5_K_M.gguf) | Q5_K_M | 4.78GB |\n| [NeuralBeagle14-7B.Q5_1.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q5_1.gguf) | Q5_1 | 5.07GB |\n| [NeuralBeagle14-7B.Q6_K.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q6_K.gguf) | Q6_K | 5.53GB |\n| [NeuralBeagle14-7B.Q8_0.gguf](https://huggingface.co/RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf/blob/main/NeuralBeagle14-7B.Q8_0.gguf) | Q8_0 | 7.17GB |\n\n\n\n\nOriginal model description:\n---\nlicense: cc-by-nc-4.0\ntags:\n- merge\n- mergekit\n- lazymergekit\n- dpo\n- rlhf\nbase_model: mlabonne/Beagle14-7B\nmodel-index:\n- name: NeuralBeagle14-7B\n  results:\n  - task:\n      type: text-generation\n      name: Text Generation\n    dataset:\n      name: AI2 Reasoning Challenge (25-Shot)\n      type: ai2_arc\n      config: ARC-Challenge\n      split: test\n      args:\n        num_few_shot: 25\n    metrics:\n    - type: acc_norm\n      value: 72.95\n      name: normalized accuracy\n    source:\n      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/NeuralBeagle14-7B\n      name: Open LLM Leaderboard\n  - task:\n      type: text-generation\n      name: Text Generation\n    dataset:\n      name: HellaSwag (10-Shot)\n      type: hellaswag\n      split: validation\n      args:\n        num_few_shot: 10\n    metrics:\n    - type: acc_norm\n      value: 88.34\n      name: normalized accuracy\n    source:\n      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/NeuralBeagle14-7B\n      name: Open LLM Leaderboard\n  - task:\n      type: text-generation\n      name: Text Generation\n    dataset:\n      name: MMLU (5-Shot)\n      type: cais/mmlu\n      config: all\n      split: test\n      args:\n        num_few_shot: 5\n    metrics:\n    - type: acc\n      value: 64.55\n      name: accuracy\n    source:\n      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/NeuralBeagle14-7B\n      name: Open LLM Leaderboard\n  - task:\n      type: text-generation\n      name: Text Generation\n    dataset:\n      name: TruthfulQA (0-shot)\n      type: truthful_qa\n      config: multiple_choice\n      split: validation\n      args:\n        num_few_shot: 0\n    metrics:\n    - type: mc2\n      value: 69.93\n    source:\n      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/NeuralBeagle14-7B\n      name: Open LLM Leaderboard\n  - task:\n      type: text-generation\n      name: Text Generation\n    dataset:\n      name: Winogrande (5-shot)\n      type: winogrande\n      config: winogrande_xl\n      split: validation\n      args:\n        num_few_shot: 5\n    metrics:\n    - type: acc\n      value: 82.4\n      name: accuracy\n    source:\n      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/NeuralBeagle14-7B\n      name: Open LLM Leaderboard\n  - task:\n      type: text-generation\n      name: Text Generation\n    dataset:\n      name: GSM8k (5-shot)\n      type: gsm8k\n      config: main\n      split: test\n      args:\n        num_few_shot: 5\n    metrics:\n    - type: acc\n      value: 70.28\n      name: accuracy\n    source:\n      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/NeuralBeagle14-7B\n      name: Open LLM Leaderboard\n---\n\n![](https://i.imgur.com/89ZAKcn.png)\n\n# 🐶 NeuralBeagle14-7B\n\n**Update 01/16/24: NeuralBeagle14-7B is (probably) the best 7B model you can find! 🎉**\n\nNeuralBeagle14-7B is a DPO fine-tune of [mlabonne/Beagle14-7B](https://huggingface.co/mlabonne/Beagle14-7B) using the [argilla/distilabel-intel-orca-dpo-pairs](https://huggingface.co/datasets/argilla/distilabel-intel-orca-dpo-pairs) preference dataset and my DPO notebook from [this article](https://towardsdatascience.com/fine-tune-a-mistral-7b-model-with-direct-preference-optimization-708042745aac).\n\nIt is based on a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):\n* [fblgit/UNA-TheBeagle-7b-v1](https://huggingface.co/fblgit/UNA-TheBeagle-7b-v1), based on jondurbin's [repo](https://github.com/jondurbin/bagel) and [jondurbin/bagel-v0.3](https://huggingface.co/datasets/jondurbin/bagel-v0.3])\n* [argilla/distilabeled-Marcoro14-7B-slerp](https://huggingface.co/argilla/distilabeled-Marcoro14-7B-slerp), based on [mlabonne/Marcoro14-7B-slerp](https://huggingface.co/mlabonne/Marcoro14-7B-slerp)\n\nThanks [Argilla](https://huggingface.co/argilla) for providing the dataset and the training recipe [here](https://huggingface.co/argilla/distilabeled-Marcoro14-7B-slerp). 💪\n\nYou can try it out in this [Space](https://huggingface.co/spaces/mlabonne/NeuralBeagle14-7B-GGUF-Chat) (GGUF Q4_K_M).\n\n## 🔍 Applications\n\nThis model uses a context window of 8k. It is compatible with different templates, like chatml and Llama's chat template.\n\nCompared to other 7B models, it displays good performance in instruction following and reasoning tasks. It can also be used for RP and storytelling.\n\n## ⚡ Quantized models\n\n* **GGUF**: https://huggingface.co/mlabonne/NeuralBeagle14-7B-GGUF\n* **GPTQ**: https://huggingface.co/TheBloke/NeuralBeagle14-7B-GPTQ\n* **AWQ**: https://huggingface.co/TheBloke/NeuralBeagle14-7B-AWQ\n* **EXL2**: https://huggingface.co/LoneStriker/NeuralBeagle14-7B-8.0bpw-h8-exl2\n\n## 🏆 Evaluation\n\n### Open LLM Leaderboard\n\nNeuralBeagle14-7B ranks first on the Open LLM Leaderboard in the ~7B category.\n\n![](https://i.imgur.com/4nAzJsr.png)\n\nIt has the same average score as Beagle14-7B (\"Show merges\"), which could be due to might be due to an unlucky run.\nI think I might be overexploiting argilla/distilabel-intel-orca-dpo-pairs at this point, since this dataset or its original version are present in multiple models.\nI need to find more high-quality preference data for the next DPO merge.\n\nNote that some models like udkai/Turdus and nfaheem/Marcoroni-7b-DPO-Merge are unfortunately contaminated on purpose (see the very high Winogrande score).\n\n### Nous\n\nThe evaluation was performed using [LLM AutoEval](https://github.com/mlabonne/llm-autoeval) on Nous suite. It is the best 7B model to date.\n\n| Model | Average | AGIEval | GPT4All | TruthfulQA | Bigbench |\n|---|---:|---:|---:|---:|---:|\n| [**mlabonne/NeuralBeagle14-7B**](https://huggingface.co/mlabonne/NeuralBeagle14-7B) [📄](https://gist.github.com/mlabonne/ad0c665bbe581c8420136c3b52b3c15c) | **60.25** | **46.06** | **76.77** | **70.32** | **47.86** |\n| [mlabonne/Beagle14-7B](https://huggingface.co/mlabonne/Beagle14-7B) [📄](https://gist.github.com/mlabonne/f5a5bf8c0827bbec2f05b97cc62d642c) | 59.4 | 44.38 | 76.53 | 69.44 | 47.25 |\n| [mlabonne/NeuralDaredevil-7B](https://huggingface.co/mlabonne/NeuralDaredevil-7B) [📄](https://gist.github.com/mlabonne/cbeb077d1df71cb81c78f742f19f4155) | 59.39 | 45.23 | 76.2 | 67.61 | 48.52 |\n| [argilla/distilabeled-Marcoro14-7B-slerp](https://huggingface.co/argilla/distilabeled-Marcoro14-7B-slerp) [📄](https://gist.github.com/mlabonne/9082c4e59f4d3f3543c5eda3f4807040) | 58.93 | 45.38 | 76.48 | 65.68 | 48.18 |\n| [mlabonne/NeuralMarcoro14-7B](https://huggingface.co/mlabonne/NeuralMarcoro14-7B) [📄](https://gist.github.com/mlabonne/b31572a4711c945a4827e7242cfc4b9d) | 58.4 | 44.59 | 76.17 | 65.94 | 46.9 |\n| [openchat/openchat-3.5-0106](https://huggingface.co/openchat/openchat-3.5-0106) [📄](https://gist.github.com/mlabonne/1afab87b543b0717ec08722cf086dcc3) | 53.71 | 44.17 | 73.72 | 52.53 | 44.4 |\n| [teknium/OpenHermes-2.5-Mistral-7B](https://huggingface.co/teknium/OpenHermes-2.5-Mistral-7B) [📄](https://gist.github.com/mlabonne/88b21dd9698ffed75d6163ebdc2f6cc8) | 52.42 | 42.75 | 72.99 | 52.99 | 40.94 |\n\nYou can find the complete benchmark on [YALL - Yet Another LLM Leaderboard](https://huggingface.co/spaces/mlabonne/Yet_Another_LLM_Leaderboard).\n\n## 💻 Usage\n\n```python\n!pip install -qU transformers accelerate\n\nfrom transformers import AutoTokenizer\nimport transformers\nimport torch\n\nmodel = \"mlabonne/NeuralBeagle14-7B\"\nmessages = [{\"role\": \"user\", \"content\": \"What is a large language model?\"}]\n\ntokenizer = AutoTokenizer.from_pretrained(model)\nprompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)\npipeline = transformers.pipeline(\n    \"text-generation\",\n    model=model,\n    torch_dtype=torch.float16,\n    device_map=\"auto\",\n)\n\noutputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)\nprint(outputs[0][\"generated_text\"])\n```\n\n<p align=\"center\">\n  <a href=\"https://github.com/argilla-io/distilabel\">\n    <img src=\"https://raw.githubusercontent.com/argilla-io/distilabel/main/docs/assets/distilabel-badge-light.png\" alt=\"Built with Distilabel\" width=\"200\" height=\"32\"/>\n  </a>\n</p>\n\n\n",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "endpoints_compatible",
    "region:us",
    "conversational"
  ],
  "likes": 0,
  "downloads": 171,
  "gated": false,
  "private": false,
  "last_modified": "2024-07-25T04:04:23.000Z",
  "created_at": "2024-07-25T00:04:34.000Z",
  "pipeline_tag": "",
  "library_name": ""
}

{
  "_id": "66a19692d118578d3f7f4e66",
  "id": "RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf",
  "modelId": "RichardErkhov/mlabonne_-_NeuralBeagle14-7B-gguf",
  "sha": "c6abb10597b7e7ef7497e72e0b7a1a671a14da3f",
  "createdAt": "2024-07-25T00:04:34.000Z",
  "lastModified": "2024-07-25T04:04:23.000Z",
  "author": "RichardErkhov",
  "downloads": 171,
  "likes": 0,
  "gated": false,
  "private": false,
  "pipeline_tag": "",
  "library_name": "",
  "siblings_count": 24
}