richarderkhov/athirdpath_-_llama-3.1-base_nsfw-pretrained_e-0.5-gguf overview

{
  "metadata": {},
  "card_data": {
    "frontmatter": {},
    "hero_image_url": "",
    "summary": "Quantization made by Richard Erkhov. Github Discord Request more models Llama-3.1-Base_NSFW-pretrained_e-0.5 - GGUF | Name | Quant method | Size | | ---- | ---- | ---- | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q2_K.gguf | Q2_K | 2.96GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_XS.gguf | IQ3_XS | 3.28GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_S.gguf | IQ3_S | 3.43GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_S.gguf | Q3_K_S | 3.41GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_M.gguf | IQ3_M | 3.52GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K.gguf | Q3_K | 3.74GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_M.gguf | Q3_K_M | 3.74GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_L.gguf | Q3_K_L | 4.03GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ4_XS.gguf | IQ4_XS | 4.18GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_0.gguf | Q4_0 | 4.34GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ4_NL.gguf | IQ4_NL | 4.38GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K_S.gguf | Q4_K_S | 4.37GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K.gguf | Q4_K | 4.58GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K_M.gguf | Q4_K_M | 4.58GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_1.gguf | Q4_1 | 4.78GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_0.gguf | Q5_0 | 5.21GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K_S.gguf | Q5_K_S | 5.21GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K.gguf | Q5_K | 5.34GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K_M.gguf | Q5_K_M | 5.34GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_1.gguf | Q5_1 | 5.65GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q6_K.gguf | Q6_K | 6.14GB | | Llama-3.1-Base_NSFW-pretrained_e-0.5.Q8_0.gguf | Q8_0 | 7.95GB | Original model description: --- license: llama3.1 --- Llama 3.1 **Base**, continually pretrained with 0.5 Epochs (2100 steps @ total batch 64) of the same 1.5gb private dataset that underpins Iambe Mostly a proof of concept, but outputs are better than expected. It'd likely be quite good with some instruction tuning. ----- Why do this? I have a niche use case where I cannot increase compute over 8b, and L3/3.1 are the only models in this size category that meet my needs for logic. However, both versions of L3/3.1 have the damn repetition/token overconfidence problem, and this is meant to disrupt that certainty without disrupting the model's ability to function. By the way, I *think* it's the lm_head that is causing the looping, but it might be the embeddings being too separated. I'm not going to pay two more times to test them separately, however :p",
    "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\nLlama-3.1-Base_NSFW-pretrained_e-0.5 - GGUF\n- Model creator: https://huggingface.co/athirdpath/\n- Original model: https://huggingface.co/athirdpath/Llama-3.1-Base_NSFW-pretrained_e-0.5/\n\n\n| Name | Quant method | Size |\n| ---- | ---- | ---- |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q2_K.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q2_K.gguf) | Q2_K | 2.96GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_XS.gguf) | IQ3_XS | 3.28GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_S.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_S.gguf) | IQ3_S | 3.43GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_S.gguf) | Q3_K_S | 3.41GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_M.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ3_M.gguf) | IQ3_M | 3.52GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K.gguf) | Q3_K | 3.74GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_M.gguf) | Q3_K_M | 3.74GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q3_K_L.gguf) | Q3_K_L | 4.03GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ4_XS.gguf) | IQ4_XS | 4.18GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_0.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_0.gguf) | Q4_0 | 4.34GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.IQ4_NL.gguf) | IQ4_NL | 4.38GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K_S.gguf) | Q4_K_S | 4.37GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K.gguf) | Q4_K | 4.58GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_K_M.gguf) | Q4_K_M | 4.58GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_1.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q4_1.gguf) | Q4_1 | 4.78GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_0.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_0.gguf) | Q5_0 | 5.21GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K_S.gguf) | Q5_K_S | 5.21GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K.gguf) | Q5_K | 5.34GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_K_M.gguf) | Q5_K_M | 5.34GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_1.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q5_1.gguf) | Q5_1 | 5.65GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q6_K.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q6_K.gguf) | Q6_K | 6.14GB |\n| [Llama-3.1-Base_NSFW-pretrained_e-0.5.Q8_0.gguf](https://huggingface.co/RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf/blob/main/Llama-3.1-Base_NSFW-pretrained_e-0.5.Q8_0.gguf) | Q8_0 | 7.95GB |\n\n\n\n\nOriginal model description:\n---\nlicense: llama3.1\n---\n\nLlama 3.1 **Base**, continually pretrained with 0.5 Epochs (2100 steps @ total batch 64) of the same 1.5gb private dataset that underpins Iambe\n\nMostly a proof of concept, but outputs are better than expected. It'd likely be quite good with some instruction tuning.\n\n-----\n\nWhy do this? I have a niche use case where I cannot increase compute over 8b, and L3/3.1 are the only models in this size category that meet my needs for logic. However, both versions of L3/3.1 have the damn repetition/token overconfidence problem, and this is meant to disrupt that certainty without disrupting the model's ability to function.\n\nBy the way, I *think* it's the lm_head that is causing the looping, but it might be the embeddings being too separated. I'm not going to pay two more times to test them separately, however :p\n\n",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "endpoints_compatible",
    "region:us"
  ],
  "likes": 0,
  "downloads": 121,
  "gated": false,
  "private": false,
  "last_modified": "2024-08-09T12:58:07.000Z",
  "created_at": "2024-08-09T11:34:29.000Z",
  "pipeline_tag": "",
  "library_name": ""
}

{
  "_id": "66b5fec594096020e0b43450",
  "id": "RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf",
  "modelId": "RichardErkhov/athirdpath_-_Llama-3.1-Base_NSFW-pretrained_e-0.5-gguf",
  "sha": "0644937f35ddde16421ad6936ce0c84b544535f0",
  "createdAt": "2024-08-09T11:34:29.000Z",
  "lastModified": "2024-08-09T12:58:07.000Z",
  "author": "RichardErkhov",
  "downloads": 121,
  "likes": 0,
  "gated": false,
  "private": false,
  "pipeline_tag": "",
  "library_name": "",
  "siblings_count": 24
}