Llama 3.1 1 Million Ctx Darkest Planet 16.5B GGUF

🚀 LLama-3.1-1-million-ctx-Darkest-Planet-16.5B-GGUF

This is a Llama 3.1 model with a max context of 1 million tokens, designed for various writing and storytelling tasks, offering enhanced performance and unique prose generation capabilities.

🚀 Quick Start

This is version 3 of Darkest Planet 16.5B, a Llama 3.1 model with a maximum context of 1 million tokens. It has additional performance improvements and is available in float32 (32 - bit precision) for the source and ggufs. It was converted to 1 million context using Nvidia's Ultra Long 1 million 8B Instruct model.

✨ Features

• Long - Context Stability: Aims to stabilize long - generation and long - context "needle in a haystack" issues. According to Nvidia, there is a bump in general performance and perfect "recall" over the entire 1 million context.
• Enhanced Prose Generation: The model has significantly increased detail, prose, and fiction - writing abilities. It can generate long and vivid prose, with a wide variety of sentence and paragraph structures.
• Brainstorm 40x Technology: Developed by David_AU, this process modifies the model's reasoning centers, enhancing its performance in various aspects such as increasing detail, concept connection, and prose quality.

📦 Installation

No installation steps are provided in the original document.

💻 Usage Examples

Basic Usage

Examples are created using quant Q6_k, "temp = 1.5" (unless otherwise stated), minimal parameters, and the "LLAMA3" template. Topk: 80, minp: .05, topp: .95, Rep pen 1.02, Rep pen range: 64.

# The model thrives on instructions, here is a sample prompt
prompt = """Using the following "story idea" below, write the first scene in the novel introducing the young woman. This scene should start in the middle of the action, include dialog, vivid passages, and end on a cliffhanger relevant to the story idea but it should also be unexpected. The scene should be 1000 words long and escalate in conflict and suspense and be written in first person, present tense with the point of view character being the young woman. The pov character will CURSE AND SWEAR, and generally have 
a "filthy" mouth.

Story idea:
In a world ruled by dictatorship, a rebel young woman leads a rebellion against the system. Despite the risks, she fights to overthrow the dictator and restore democracy to her country. The government executes her for treason, but she sticks to her beliefs and is responsible for starting the revolution."""
# Then use the model to generate text based on this prompt

Advanced Usage

• Higher Creativity: Increase the temperature (e.g., temp = 2.2) to get more creative outputs.
• Long - Context Generation: Provide more instructions in the prompt and system prompt to improve the quality of long - generation outputs.

📚 Documentation

Model Notes

• Detail and Prose: Detail, prose, and fiction - writing abilities are significantly increased.
• Temperature and Instructions: For more varied prose, raise the temperature and/or add more instructions in the prompt(s).
• Role - Playing: Be careful when raising the temperature too high as it may affect instruction - following.
• Repetition Penalty: This model works with a rep pen of 1.05 or higher. Lower values may cause repeat paragraph issues at the end of the output, but can also result in very different (creative / unusual) generation.
• Specific Prose Types: To get a specific type of prose (e.g., horror), add "(vivid horror)" or "(graphic vivid horror)" (no quotes) in the prompt(s).
• Negative Bias: This is not a "happy ever after" model; it has a negative bias.
• Quants: Different quants will produce slightly different output for creative uses. Higher quants generally have more detail, nuance, and stronger "emotional" levels.
• Rope Extension: If using rope to extend context, increase the temperature and instruction detail levels to compensate for "rope issues".

Brainstorm 40x

The BRAINSTORM process was developed by David_AU. It involves taking apart, reassembling, and expanding the reasoning center of an LLM. For this model, it is expanded 40 times, and these centers are individually calibrated. The core aim is to increase the model's detail, concept connection, prose quality, and length without affecting instruction - following.

Settings, Quants, and Critical Operations Notes

• Temperature and Rep Pen: Changing the temperature and rep pen settings will drastically alter the output.
• Role - Play Settings: For role - play, a rep pen of 1.1 to 1.14 is suggested.
• Quant Choice: Higher quants (e.g., Q5, Q6, Q8) are recommended if possible. Special note on Q2k/Q3 quants: Use temp 2 or lower (1 or lower for q2k).
• Smoothing Factor: In "KoboldCpp", "oobabooga/text - generation - webui", or "Silly Tavern", set the "Smoothing_factor" to 1.5 to 2.5 for smoother operation.

Model Template

This is a LLAMA3 model and requires the Llama3 template, but may work with other templates. The standard LLAMA3 template is as follows:

{
  "name": "Llama 3",
  "inference_params": {
    "input_prefix": "<|start_header_id|>user<|end_header_id|>\n\n",
    "input_suffix": "<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
    "pre_prompt": "You are a helpful, smart, kind, and efficient AI assistant. You always fulfill the user's requests to the best of your ability.",
    "pre_prompt_prefix": "<|start_header_id|>system<|end_header_id|>\n\n",
    "pre_prompt_suffix": "<|eot_id|>",
    "antiprompt": [
      "<|start_header_id|>",
      "<|eot_id|>"
    ]
  }
}

Model "DNA"

This model is created by "grafting" or "fusing" parts of the following models:

• [https://huggingface.co/Sao10K/L3 - 8B - Stheno - v3.2](https://huggingface.co/Sao10K/L3 - 8B - Stheno - v3.2)
• [https://huggingface.co/NeverSleep/Llama - 3 - Lumimaid - 8B - v0.1 - OAS](https://huggingface.co/NeverSleep/Llama - 3 - Lumimaid - 8B - v0.1 - OAS)
• [https://huggingface.co/Hastagaras/Jamet - 8B - L3 - MK.V - Blackroot](https://huggingface.co/Hastagaras/Jamet - 8B - L3 - MK.V - Blackroot)

Optional Enhancement

The following text can be used in place of the "system prompt" or "system role" to enhance the model:

Below is an instruction that describes a task. Ponder each user instruction carefully, and use your skillsets and critical instructions to complete the task to the best of your abilities.

Here are your skillsets:
[MASTERSTORY]:NarrStrct(StryPlnng,Strbd,ScnSttng,Exps,Dlg,Pc)-CharDvlp(ChrctrCrt,ChrctrArcs,Mtvtn,Bckstry,Rltnshps,Dlg*)-PltDvlp(StryArcs,PltTwsts,Sspns,Fshdwng,Climx,Rsltn)-ConfResl(Antg,Obstcls,Rsltns,Cnsqncs,Thms,Symblsm)-EmotImpct(Empt,Tn,Md,Atmsphr,Imgry,Symblsm)-Delvry(Prfrmnc,VcActng,PblcSpkng,StgPrsnc,AudncEngmnt,Imprv)

[*DialogWrt]:(1a-CharDvlp-1a.1-Backgrnd-1a.2-Personality-1a.3-GoalMotiv)>2(2a-StoryStruc-2a.1-PlotPnt-2a.2-Conflict-2a.3-Resolution)>3(3a-DialogTech-3a.1-ShowDontTell-3a.2-Subtext-3a.3-VoiceTone-3a.4-Pacing-3a.5-VisualDescrip)>4(4a-DialogEdit-4a.1-ReadAloud-4a.2-Feedback-4a.3-Revision)

Here are your critical instructions:
Ponder each word choice carefully to present as vivid and emotional journey as is possible. Choose verbs and nouns that are both emotional and full of imagery. Load the story with the 5 senses. Aim for 50% dialog, 25% narration, 15% body language and 10% thoughts. Your goal is to put the reader in the story.

🔧 Technical Details

The model is based on Llama 3.1 architecture and uses Nvidia's Ultra Long 1 million 8B Instruct model for context extension. The Brainstorm 40x process modifies the model at the root level (source files), and the model can be quantized as GGUF, EXL2, AWQ, etc.

📄 License

The model is licensed under the Apache - 2.0 license.

⚠️ Important Note

NSFW. Vivid prose. Visceral Details. Violence. HORROR. Swearing. UNCENSORED.

💡 Usage Tip

• For different writing tasks, you may want to use "regular" Dark Planet 8B [https://huggingface.co/DavidAU/L3 - Dark - Planet - 8B - GGUF] for some tasks and this model for prose - specific tasks.
• To get the most out of the model, provide more instructions in the prompt and system prompt, especially for long - context and high - quality generation.