quantfactory/llama-3.1-8b-arliai-rpmax-v1.2-gguf overview

{
  "metadata": {},
  "card_data": {
    "license": "llama3.1",
    "frontmatter": {},
    "hero_image_url": "https://lh7-rt.googleusercontent.com/docsz/AD_4nXeiuCm7c8lEwEJuRey9kiVZsRn2W-b4pWlu3-X534V3YmVuVc2ZL-NXg2RkzSOOS2JXGHutDuyyNAUtdJI65jGTo8jT9Y99tMi4H4MqL44Uc5QKG77B0d6-JfIkZHFaUA71-RtjyYZWVIhqsNZcx8-OMaA?key=xt3VSDoCbmTY7o-cwwOFwQ",
    "summary": "This is quantized version of ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.2 created using llama.cpp # Original Model Card # Llama-3.1-8B-ArliAI-RPMax-v1.2 =====================================",
    "quick_links": [],
    "benchmark_table_html": "",
    "readme_markdown": "\n---\n\nlicense: llama3.1\n\n---\n\n[![QuantFactory Banner](https://lh7-rt.googleusercontent.com/docsz/AD_4nXeiuCm7c8lEwEJuRey9kiVZsRn2W-b4pWlu3-X534V3YmVuVc2ZL-NXg2RkzSOOS2JXGHutDuyyNAUtdJI65jGTo8jT9Y99tMi4H4MqL44Uc5QKG77B0d6-JfIkZHFaUA71-RtjyYZWVIhqsNZcx8-OMaA?key=xt3VSDoCbmTY7o-cwwOFwQ)](https://hf.co/QuantFactory)\n\n\n# QuantFactory/Llama-3.1-8B-ArliAI-RPMax-v1.2-GGUF\nThis is quantized version of [ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.2](https://huggingface.co/ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.2) created using llama.cpp\n\n# Original Model Card\n\n\n# Llama-3.1-8B-ArliAI-RPMax-v1.2\n=====================================\n\n## RPMax Series Overview\n\n| [2B](https://huggingface.co/ArliAI/Gemma-2-2B-ArliAI-RPMax-v1.1) |\n[3.8B](https://huggingface.co/ArliAI/Phi-3.5-mini-3.8B-ArliAI-RPMax-v1.1) | \n[8B](https://huggingface.co/ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.2) | \n[9B](https://huggingface.co/ArliAI/Gemma-2-9B-ArliAI-RPMax-v1.1) |\n[12B](https://huggingface.co/ArliAI/Mistral-Nemo-12B-ArliAI-RPMax-v1.2) | \n[20B](https://huggingface.co/ArliAI/InternLM2_5-20B-ArliAI-RPMax-v1.1) | \n[22B](https://huggingface.co/ArliAI/Mistral-Small-22B-ArliAI-RPMax-v1.1) | \n[70B](https://huggingface.co/ArliAI/Llama-3.1-70B-ArliAI-RPMax-v1.1) | \n\nRPMax is a series of models that are trained on a diverse set of curated creative writing and RP datasets with a focus on variety and deduplication. This model is designed to be highly creative and non-repetitive by making sure no two entries in the dataset have repeated characters or situations, which makes sure the model does not latch on to a certain personality and be capable of understanding and acting appropriately to any characters or situations.\n\nEarly tests by users mentioned that these models does not feel like any other RP models, having a different style and generally doesn't feel in-bred.\n\nYou can access the model at https://arliai.com and ask questions at https://www.reddit.com/r/ArliAI/\n\nWe also have a models ranking page at https://www.arliai.com/models-ranking\n\nAsk questions in our new Discord Server! https://discord.com/invite/t75KbPgwhk\n\n## Model Description\n\nLlama-3.1-8B-ArliAI-RPMax-v1.2 is a variant of the Meta-Llama-3.1-8B model.\n\nv1.2 update is a retrain using an incremental improvement of the RPMax dataset which dedups the dataset even more and better filtering to cutout irrelevant description text that came from card sharing sites.\n\n### Specs\n\n* **Context Length**: 128K\n* **Parameters**: 8B\n\n### Training Details\n\n* **Sequence Length**: 8192\n* **Training Duration**: Approximately 1 day on 2x3090Ti\n* **Epochs**: 1 epoch training for minimized repetition sickness\n* **LORA**: 64-rank 128-alpha, resulting in ~2% trainable weights\n* **Learning Rate**: 0.00001\n* **Gradient accumulation**: Very low 32 for better learning.\n\n## Quantization\n\nThe model is available in quantized formats:\n\nWe recommend using full weights or GPTQ\n\n* **FP16**: https://huggingface.co/ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.2\n* **GGUF**: https://huggingface.co/ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.2-GGUF\n\n## Suggested Prompt Format\n\nLlama 3 Instruct Format\n\nExample:\n```\n<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nYou are [character]. You have a personality of [personality description]. [Describe scenario]<|eot_id|><|start_header_id|>user<|end_header_id|>\n\n{{ user_message_1 }}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n{{ model_answer_1 }}<|eot_id|><|start_header_id|>user<|end_header_id|>\n\n{{ user_message_2 }}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n```\n\n",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "license:llama3.1",
    "endpoints_compatible",
    "region:us",
    "conversational"
  ],
  "likes": 2,
  "downloads": 149,
  "gated": false,
  "private": false,
  "last_modified": "2024-10-16T04:52:31.000Z",
  "created_at": "2024-10-16T04:10:08.000Z",
  "pipeline_tag": "",
  "library_name": ""
}

{
  "_id": "670f3ca04a96858b071618f9",
  "id": "QuantFactory/Llama-3.1-8B-ArliAI-RPMax-v1.2-GGUF",
  "modelId": "QuantFactory/Llama-3.1-8B-ArliAI-RPMax-v1.2-GGUF",
  "sha": "cea4141cb030b5d48bdac9d6eb6362c9f57689e3",
  "createdAt": "2024-10-16T04:10:08.000Z",
  "lastModified": "2024-10-16T04:52:31.000Z",
  "author": "QuantFactory",
  "downloads": 149,
  "likes": 2,
  "gated": false,
  "private": false,
  "pipeline_tag": "",
  "library_name": "",
  "siblings_count": 16
}