richarderkhov/yanismiraoui_-_yarn-mistral-7b-128k-sharded-gguf overview

{
  "metadata": {},
  "card_data": {
    "frontmatter": {},
    "hero_image_url": "https://raw.githubusercontent.com/jquesnelle/yarn/mistral/data/proofpile-long-small-mistral.csv.png",
    "summary": "Preprint (arXiv) GitHub !yarn",
    "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\nYarn-Mistral-7b-128k-sharded - GGUF\n- Model creator: https://huggingface.co/yanismiraoui/\n- Original model: https://huggingface.co/yanismiraoui/Yarn-Mistral-7b-128k-sharded/\n\n\n| Name | Quant method | Size |\n| ---- | ---- | ---- |\n| [Yarn-Mistral-7b-128k-sharded.Q2_K.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q2_K.gguf) | Q2_K | 2.53GB |\n| [Yarn-Mistral-7b-128k-sharded.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.IQ3_XS.gguf) | IQ3_XS | 0.34GB |\n| [Yarn-Mistral-7b-128k-sharded.IQ3_S.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.IQ3_S.gguf) | IQ3_S | 0.05GB |\n| [Yarn-Mistral-7b-128k-sharded.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q3_K_S.gguf) | Q3_K_S | 0.03GB |\n| [Yarn-Mistral-7b-128k-sharded.IQ3_M.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.IQ3_M.gguf) | IQ3_M | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q3_K.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q3_K.gguf) | Q3_K | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q3_K_M.gguf) | Q3_K_M | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q3_K_L.gguf) | Q3_K_L | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.IQ4_XS.gguf) | IQ4_XS | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q4_0.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q4_0.gguf) | Q4_0 | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.IQ4_NL.gguf) | IQ4_NL | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q4_K_S.gguf) | Q4_K_S | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q4_K.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q4_K.gguf) | Q4_K | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q4_K_M.gguf) | Q4_K_M | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q4_1.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q4_1.gguf) | Q4_1 | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q5_0.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q5_0.gguf) | Q5_0 | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q5_K_S.gguf) | Q5_K_S | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q5_K.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q5_K.gguf) | Q5_K | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q5_K_M.gguf) | Q5_K_M | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q5_1.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q5_1.gguf) | Q5_1 | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q6_K.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q6_K.gguf) | Q6_K | 0.0GB |\n| [Yarn-Mistral-7b-128k-sharded.Q8_0.gguf](https://huggingface.co/RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf/blob/main/Yarn-Mistral-7b-128k-sharded.Q8_0.gguf) | Q8_0 | 0.0GB |\n\n\n\n\nOriginal model description:\n---\ndatasets:\n- emozilla/yarn-train-tokenized-16k-mistral\nmetrics:\n- perplexity\nlibrary_name: transformers\nlicense: apache-2.0\nlanguage:\n- en\n---\n\n\n## This repo contains a SHARDED version of: https://huggingface.co/NousResearch/Yarn-Mistral-7b-128k\n\n### Huge thanks to the publishers for their amazing work, all credits go to them: https://huggingface.co/NousResearch\n\n# Model Card: Nous-Yarn-Mistral-7b-128k\n\n[Preprint (arXiv)](https://arxiv.org/abs/2309.00071)  \n[GitHub](https://github.com/jquesnelle/yarn)\n![yarn](https://raw.githubusercontent.com/jquesnelle/yarn/mistral/data/proofpile-long-small-mistral.csv.png)\n\n## Model Description\n\nNous-Yarn-Mistral-7b-128k is a state-of-the-art language model for long context, further pretrained on long context data for 1500 steps using the YaRN extension method.\nIt is an extension of [Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) and supports a 128k token context window.\n\nTo use, pass `trust_remote_code=True` when loading the model, for example\n\n```python\nmodel = AutoModelForCausalLM.from_pretrained(\"NousResearch/Yarn-Mistral-7b-128k\",\n  use_flash_attention_2=True,\n  torch_dtype=torch.bfloat16,\n  device_map=\"auto\",\n  trust_remote_code=True)\n```\n\nIn addition you will need to use the latest version of `transformers` (until 4.35 comes out)\n```sh\npip install git+https://github.com/huggingface/transformers\n```\n\n## Benchmarks\n\nLong context benchmarks:\n| Model | Context Window | 8k PPL | 16k PPL | 32k PPL | 64k PPL | 128k PPL |\n|-------|---------------:|------:|----------:|-----:|-----:|------------:|\n| [Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) | 8k | 2.96 | - | - | - | - |\n| [Yarn-Mistral-7b-64k](https://huggingface.co/NousResearch/Yarn-Mistral-7b-64k) | 64k | 3.04 | 2.65 | 2.44 | 2.20 | - |\n| [Yarn-Mistral-7b-128k](https://huggingface.co/NousResearch/Yarn-Mistral-7b-128k) | 128k | 3.08 | 2.68 | 2.47 | 2.24 | 2.19 |\n\nShort context benchmarks showing that quality degradation is minimal:\n| Model | Context Window | ARC-c | Hellaswag | MMLU | Truthful QA |\n|-------|---------------:|------:|----------:|-----:|------------:|\n| [Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) | 8k | 59.98 | 83.31 | 64.16 | 42.15 |\n| [Yarn-Mistral-7b-64k](https://huggingface.co/NousResearch/Yarn-Mistral-7b-64k) | 64k | 59.38 | 81.21 | 61.32 | 42.50 |\n| [Yarn-Mistral-7b-128k](https://huggingface.co/NousResearch/Yarn-Mistral-7b-128k) | 128k | 58.87 | 80.58 | 60.64 | 42.46 |\n\n## Collaborators\n\n - [bloc97](https://github.com/bloc97): Methods, paper and evals\n - [@theemozilla](https://twitter.com/theemozilla): Methods, paper, model training, and evals\n - [@EnricoShippole](https://twitter.com/EnricoShippole): Model training\n - [honglu2875](https://github.com/honglu2875): Paper and evals\n\nThe authors would like to thank LAION AI for their support of compute for this model.\nIt was trained on the [JUWELS](https://www.fz-juelich.de/en/ias/jsc/systems/supercomputers/juwels) supercomputer.\n\n",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "arxiv:2309.00071",
    "region:us"
  ],
  "likes": 0,
  "downloads": 92,
  "gated": false,
  "private": false,
  "last_modified": "2024-06-01T05:53:10.000Z",
  "created_at": "2024-06-01T05:40:38.000Z",
  "pipeline_tag": "",
  "library_name": ""
}

{
  "_id": "665ab456c076f34b92ddc03b",
  "id": "RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf",
  "modelId": "RichardErkhov/yanismiraoui_-_Yarn-Mistral-7b-128k-sharded-gguf",
  "sha": "4a73a5b0369ad13cd1caf648a750e18d703ffdcc",
  "createdAt": "2024-06-01T05:40:38.000Z",
  "lastModified": "2024-06-01T05:53:10.000Z",
  "author": "RichardErkhov",
  "downloads": 92,
  "likes": 0,
  "gated": false,
  "private": false,
  "pipeline_tag": "",
  "library_name": "",
  "siblings_count": 24
}