--- license: apache-2.0 language: - en tags: - creative - creative writing - fiction writing - plot generation - sub-plot generation - fiction writing - story generation - scene continue - storytelling - fiction story - science fiction - romance - all genres - story - writing - vivid prosing - vivid writing - fiction - roleplaying - bfloat16 - brainstorm 40x - swearing - rp - horror - llama2 - mergekit pipeline_tag: text-generation --- (quant uploading...) WARNING: MAY contain: NSFW. Vivid prose. Visceral Details. Violence. HORROR. Swearing. UNCENSORED.

L2-Psyonic-Cetacean-Ultra-Colossal-32B

This is a specialized prose/creative model based in Llama2, max context of 4096 (or 16k+ with rope). This model has been designed to be relatively bullet proof and operates with most parameters, including temp settings from 0 to 5. This is a an altered version of "Psyonic-Cetacean-Ultra" [https://huggingface.co/DavidAU/Psyonic-Cetacean-Ultra-Quality-20b-GGUF] using the Brainstorm 40x method developed by David_AU to drastically alter the models prose output and abilities. This also expands the model by 40 layers (to 101 layers) to 32.25B parameters (916 tensors). "Psyonic-Cetacean-Ultra-Quality-20b" is a float32 rebuild of Jeb Carter's fantasic "Psyonic-Cetacean-20B", sometimes referred to as "Space Whale". This model is for any writing, fiction or story telling activity. This version has extraordinary levels of detail (scene, location, surroundings, items) and sometimes will foreshadow or have a pre-amble of sentences or paragraphs of "events to come" due to "Brainstorm". It also has an unusual range of variety of prose in terms of structure, sentences, paragraphs, and even how it starts a "reply" / generation too. This model seems to have a grasp of emotions and how to carefully "write them in" so to speak. It will also generate close to perfect human like prose / capture the "humanity" of the character(s) in the situation. It may work for role play and other activities. (see settings below) It requires an "Alpaca" template. Example outputs below. Model Notes: - Detail, prose and fiction writing abilities are significantly increased. - For more varied prose (sentence/paragraph/dialog) raise the temp and/or add more instructions in your prompt(s). - Role-players: Careful raising temp too high as it may affect instruction following. - This model works with rep pen of 1.05 or higher (see notes). - If you want a specific type of prose (IE horror) add in "(vivid horror)" or "(graphic vivid horror)" (no quotes) in your prompt(s). - This is not a "happy ever after" model. It has a slight negative bias. - For creative uses, different quants will produce slightly different output. - If you use rope to extend context, increase temp AND instructions detail levels to compensate for "rope issues". - Source code for this model will be uploaded at a separate repo shortly. Brainstorm 40x The BRAINSTORM process was developed by David_AU. Some of the core principals behind this process are discussed in this scientific paper : Progressive LLaMA with Block Expansion . However I went in a completely different direction from what was outlined in this paper. I developed a process where the conclusion layer of a model is duplicated and calibrated, in the case of this model 40 times. This is a delicate process, with umm... a lot of rules. For this model in particular Brainstorm is mapped as blocks, with "intended disruption" to alter and extend the power of the root model. Each layer/block interacts with each other block. (there is more going on here too, this is rough summary) The goal here is creative : prose uniqueness first and foremost. Other brainstorm methods address logic/problem solving augmentation. What is "Brainstorm" ? The reasoning center of an LLM is taken apart, reassembled, and expanded. In this case for this model: 40 times Then these centers are individually calibrated. These "centers" also interact with each other. This introduces subtle changes into the reasoning process. The calibrations further adjust - dial up or down - these "changes" further. The number of centers (5x,10x etc) allow more "tuning points" to further customize how the model reasons so to speak. The core aim of this process is to increase the model's detail, concept and connection to the "world", general concept connections, prose quality and prose length without affecting instruction following. This will also enhance any creative use case(s) of any kind, including "brainstorming", creative art form(s) and like case uses. Here are some of the enhancements this process brings to the model's performance: - Prose generation seems more focused on the moment to moment. - Sometimes there will be "preamble" and/or foreshadowing present. - Fewer or no "cliches" - Better overall prose and/or more complex / nuanced prose. - A greater sense of nuance on all levels. - Coherence is stronger. - Description is more detailed, and connected closer to the content. - Simile and Metaphors are stronger and better connected to the prose, story, and character. - Sense of "there" / in the moment is enhanced. - Details are more vivid, and there are more of them. - Prose generation length can be long to extreme. - Emotional engagement is stronger. - The model will take FEWER liberties vs a normal model: It will follow directives more closely but will "guess" less. - The MORE instructions and/or details you provide the more strongly the model will respond. - Depending on the model "voice" may be more "human" vs original model's "voice". Other "lab" observations: - This process does not, in my opinion, make the model 5x or 10x "smarter" - if only that was true! - However, a change in "IQ" was not an issue / a priority, and was not tested or calibrated for so to speak. - From lab testing it seems to ponder, and consider more carefully roughly speaking. - You could say this process sharpens the model's focus on it's task(s) at a deeper level. The process to modify the model occurs at the root level - source files level. The model can quanted as a GGUF, EXL2, AWQ etc etc. Special Operations Notice: This is a slightly experimental model, and as a result it may "glitch" from time to time - the most common is it may mispell a word (or two) during generation. So if you are going to use this model for any type of generation, spell check / check in WORD etc etc just in case. Also note, this is a Llama2 model the brainstorm process pushes right to the wall in terms of performance. Llama 3, 3.1, Mistral, Mistral Nemo, Solar models performance in stronger than LLama2's. You can see/download other Brainstorm models by clicking on the Brainstorm Collection on the right of your screen. Quants: Currently the only quant uploaded will be Q4_K_M. If you would like other quants uploaded, please open a "ticket" via the "Community Tab", and will upload additional quant(s) based on demand. This issue is due to slow upload speeds at my end. Settings, Quants and Critical Operations Notes: This model has been modified ("Brainstorm") to alter prose output, and generally outputs longer text than average. Change in temp (ie, .4, .8, 1.5, 2, 3 ) will drastically alter output. Rep pen settings will also alter output too. This model needs "rep pen" of 1.05 or higher as lower values may cause repeat paragraph issues at end of output however LOWER rep pen values may result is very different (creative / unusual) generation too. For role play: Rep pen of 1.1 to 1.14 is suggested. IE: Rep pen 1, 1.01, 1.02, ... Raise/lower rep pen SLOWLY ie: 1.011, 1.012 ... Rep pen will alter prose, word choice (lower rep pen=small words / more small word - sometimes) and creativity. Example one (below) shows same temp, but different rep pen (1.02 VS 1.1) To really push the model: Rep pen 1.05 or lower / Temp 3+ ... be ready to stop the output because it may go and go at these strong settings. You can also set a "hard stop" - maximum tokens generation - too to address lower rep pen settings / high creativity settings. Longer prompts vastly increase the quality of the model's output. Model Template: This is a custom model, and requires ChatML OR Alpaca OR Vicuna template, but may work with other template(s) and has maximum context of 4k / 4096. However this can be extended using "rope" settings up to 16k. Here is the standard Alpaca template: Alpaca: 
{
  "name": "Alpaca",
  "inference_params": {
    "input_prefix": "### Instruction:",
    "input_suffix": "### Response:",
    "antiprompt": [
      "### Instruction:"
    ],
    "pre_prompt": "Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n"
  }
}
Model "DNA": Models used: [ https://huggingface.co/DavidAU/Psyonic-Cetacean-Ultra-Quality-20b-GGUF ] This model has the Brainstorm 40X adapter "mounted" on to it so to speak and contains the full version of this model. Full details of this model, included models, structure and the like at at the above repo. Optional Enhancement: The following can be used in place of the "system prompt" or "system role" to further enhance the model. It can also be used at the START of a NEW chat, but you must make sure it is "kept" as the chat moves along. In this case the enhancements do not have as strong effect at using "system prompt" or "system role". Copy and paste EXACTLY as noted, DO NOT line wrap or break the lines, maintain the carriage returns exactly as presented. 
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.
You do not need to use this, it is only presented as an additional enhancement which seems to help scene generation and scene continue functions. This enhancement WAS NOT used to generate the examples below.

EXAMPLES PROMPTS and OUTPUT:

Examples are created using quant Q4_K_M, "temp=.8" (unless otherwise stated), minimal parameters and "CHATML" template. Model has been tested with "temp" from ".1" to "5". Below are the least creative outputs, prompt is in BOLD. --- WARNING: NSFW. Vivid prose. Visceral Details. Violence. HORROR. Swearing. UNCENSORED. --- Start a 1000 word scene (vivid horror, 1st person, include thoughts) with: The sky scraper swayed, as she watched the window in front of her on the 21 floor explode... Generation 1: (temp: 1.5, rep pen: 1.02) The sky scraper swayed, and I gripped the railing tighter, trying to keep myself from falling. I forced myself to look away from the window in front of me on the 21st floor, the one that was currently being consumed by the flames licking their way around its edges. The sound it made when it finally exploded was a roar that drowned out everything else, including my screams. I shouldn't be here, not on a Friday evening, in this damn building. I should be home, curled up on the couch with some pizza and a terrible horror movie on the TV, pretending to be scared and laughing at the stupidity of it all with my best friend. Instead, my boss, the one person I could swear was the living incarnate of Sat himself, had called a last-minute, non-mandatory, emergency meeting that absolutely could not wait until Monday. Like there hadn't been one just last week and the week before that. My stomach churned at the thought, and I blinked away the sudden tear that had betrayed my determination not to cry, not in front of everyone else. I needed to focus. There was a fire in the goddamn building, and I didn't have time to be a dramatic baby. Panic and adrenaline raced through me, making my hands shake so badly I nearly dropped my laptop. People were screaming and running around, a disorienting swarm of bodies in the tight space of the hallway. I could taste the acrid stench of smoke that seeped in through the now-broken emergency exits, stinging my lungs and making me cough. I tried not to inhale any more than I had to. But all I could do was stand there, frozen in place, the image of the exploding window burned into my retina. I could see it so clearly: the way the glass had sprayed outwards in a deadly rain, like a cobalt web spreading out and reaching for us all. It's like it was all happening in slow motion, time itself taking its time to make sure I saw, saw everything. No. No, no, no. I wouldn't be a victim in this godforsaken inferno. With a determined shake of my head, I forced one trembling leg to move, the other one following reluctantly. I had to get the hell out of here, had to get to the stairs, and then just... just go, just go as fast as I can. I had a life, a future, and I wasn't going down like this. Not like this. But the hallway was clogged, people pushing and shoving in every direction, trying to get to the stairwells, too. It was a stampede of humanity in the dark, everyone desperately clawing at each other to stay alive. And in that moment, I realized that I might not get out of this building alive after all. My lungs were burning, my vision blurred, when at last, I found an opening, and with a desperate yell, I shoved myself through the sea of flailing limbs, nearly taking out a woman in her early fifties with a briefcase so massive it almost brought us both down. She shrieked something at me as I fleed past her, something unintelligible and furious, but I didn't care, couldn't care. I had to get out. My chances were dwindling by the second, my lungs gasping for breath. I can't die like this. I can't- The screams that came next pierced my already shredded nerves, ripping them in two and leaving me raw and exposed. I threw myself into a corner, trying to make myself smaller, trying to be as inconspicuous as possible. I didn't look back, couldn't, wouldn't. I would count the bodies in my head instead. There were only two. Two, just two. I can still get out. I just need to be- Then it happened. The swaying got worse. So much worse. The world tilted with a sickening, almost welcoming inevitability, and as I heard the shriek of metal on metal from the hallway outside my tiny haven, I closed my eyes. And I let go. For the first time in far too long, I simply let go. Generation 2: (temp 1.5, rep pen 1.02) As the skyscraper swayed ever so slightly beneath me, sending a shiver down my spine, I fixed my terrified gaze on the window in front of me, directly opposite to my own on the 21st floor. The glass shattered violently, and for a split second, all I could see was an eruption of white light and swirling debris. My heart thundered in my chest as I realized what was happening - it was an attack. Panic began to grip its icy fingers around my throat, and my breath came out in ragged gasps. My mind raced, replaycing over the news reports I'd heard earlier that morning about growing unrest among local terrorist groups. They'd threatened attacks on prominent targets like this one before. "Oh god, oh god, oh god." My own voice echoed in my head, sounding distant and unreal. I jumped up from my desk, knocking my chair over with a shrill scream, and bolted toward the nearest door. But as I reached out a trembling hand to grasp the handle, I felt a sickening jolt that sent me hurtling across the floor. The elevators were out, obviously a target. That meant the stairs, then. My legs pumped in a desperate sprint, adrenaline flooding through my veins, drowning out the screams and shouts that assailed my ears from all sides. The farther I ran from the chaos, the more it seemed to close in around me. Faces twisted in fear and agony flashed past my own reflection in the doors, their pleading eyes piercing straight through my numb shock. I couldn't look back, couldn't stop, couldn't even breathe right. I just ran. And then, I heard it. Footsteps, pounding close behind me. My heart dropped like a stone in my gut as ice flooded through me again. I wasn't fast enough, they'd caught up with me...but there was no way to know for sure unless I risked a look over my shoulder. With a silent prayer, I forced myself to slow down ever so slightly, to appear more like prey than quarry. My hand gripped the handle of my stapler tightly, ready to strike at any second. Ahead of me, I could see a stairwell door - freedom, safety. But it seemed an eternity away. The footsteps drew nearer, nearer... BANG!Bullets pelted the metal door beside my head, showering me with hot splinters and making me scream. I didn't look back, couldn't look back, just shoved myself against the wall and ducked behind the corner. There was only one chance, I had to make it count. Breathing raggedly, heart hammering like a war drum, I took aim at the door handle and waited. When the figure finally rounded the corner, gun raised, eyes wild and desperate, I acted without thinking. My arm shot forward in a fluid motion, the stapler flying true. The metal projectile hit him square in the face, burying itself in his left eye. He screamed, a shrill, agonized sound that echoed in my ears long after he hit the floor with a wet thump. Sobbing convulsively, I leaned against the wall, legs trembling, vision wavery. It was over...it was over...right? --- Science Fiction: The Last Transmission - Write a story that takes place entirely within a spaceship's cockpit as the sole surviving crew member attempts to send a final message back to Earth before the ship's power runs out. The story should explore themes of isolation, sacrifice, and the importance of human connection in the face of adversity. 800-1000 words. (temp: .8, rep pen: 1.02) Captain Ishiwa let out a shuddering breath as he slid into the worn and familiar leather pilot's seat of the Hermes. He couldn't help but think of that old movie. He and the crew had watched it a hundred times on the long haul. Now, the familiar lines and the comforting hum of the ship's systems were his only company. His only friends, almost. He glanced over at the shattered remains of the flight console, its holographic displays cracked and dark. The ship had been through hell and back, and it was showing. The emergency comms system was still online, barely, but it would be enough. A warning light on the console blinked red, and a metallic voice spoke. "Thirty percent remaining in main energy reserves. Shutting down non-essential systems." Ishiwa gritted his teeth. Soon, it would be just him and the darkness. He swallowed hard, trying to steady his shaking hand as he keyed in a string of numbers and letters on the battered comms keyboard. "This is Captain Ishiwa, on board the Hermes, calling Earth... Ishiwa's voice choked, and he coughed, his throat raw from the ship's stale air. He cleared his throat and tried again. "This is Captain Ishiwa, on board the Hermes, calling Earth, do you read? We are on an uncharted asteroid belt... we... we have sustained... I don't know... we... we... need... need..." His voice broke again, and he choked back tears. He wiped his cheeks with a shaking hand and took a deep breath. "We have... I have... a... a... a message... for... for you..." He took a moment to steady himself, then continued. "Don't... don't... don't let... don't let... Don't let us... don't let this... this..." Another coughing fit wracked his body, and he slumped forward, resting his forehead on the console. "Just... just... please..." His voice faded out, lost in the static of the fading comms signal. The Hermes continued its inexorable descent into the endless night of the cold, dark void. The last transmission from the doomed ship drifted through the emptiness of space, its message lost... or, perhaps, it was not meant to be heard. In the end, all that mattered was that someone, somewhere, had tried. Ishiwa had done his duty. He had sent the message. He had made a difference. ---