davidau/qwen3-42b-a3b-yoyo-v5-total-recall-neo-imatrix-gguf overview

{
  "metadata": {},
  "card_data": {
    "license": "apache-2.0",
    "language": [
      "en",
      "fr",
      "zh",
      "de"
    ],
    "tags": [
      "programming",
      "code generation",
      "code",
      "codeqwen",
      "programming",
      "code generation",
      "code",
      "codeqwen",
      "moe",
      "coding",
      "coder",
      "qwen2",
      "chat",
      "qwen",
      "qwen-coder",
      "chat",
      "qwen",
      "qwen-coder",
      "moe",
      "Qwen3-Coder-30B-A3B-Instruct",
      "Qwen3-30B-A3B",
      "mixture of experts",
      "128 experts",
      "8 active experts",
      "256k context",
      "qwen3",
      "gguf",
      "NEO Imatrix",
      "brainstorm 20x",
      "brainstorm",
      "optional thinking",
      "qwen3_moe"
    ],
    "pipeline_tag": "text-generation",
    "frontmatter": {
      "license": "apache-2.0",
      "language": [
        "en",
        "fr",
        "zh",
        "de"
      ],
      "tags": [
        "programming",
        "code generation",
        "code",
        "codeqwen",
        "programming",
        "code generation",
        "code",
        "codeqwen",
        "moe",
        "coding",
        "coder",
        "qwen2",
        "chat",
        "qwen",
        "qwen-coder",
        "chat",
        "qwen",
        "qwen-coder",
        "moe",
        "Qwen3-Coder-30B-A3B-Instruct",
        "Qwen3-30B-A3B",
        "mixture of experts",
        "128 experts",
        "8 active experts",
        "256k context",
        "qwen3",
        "gguf",
        "NEO Imatrix",
        "brainstorm 20x",
        "brainstorm",
        "optional thinking",
        "qwen3_moe"
      ],
      "pipeline_tag": "text-generation"
    },
    "hero_image_url": "qwen3-total-recall.gif",
    "summary": "",
    "quick_links": [],
    "benchmark_table_html": "",
    "readme_markdown": "---\nlicense: apache-2.0\nlanguage:\n- en\n- fr\n- zh\n- de\ntags:\n- programming\n- code generation\n- code\n- codeqwen\n- programming\n- code generation\n- code\n- codeqwen\n- moe\n- coding\n- coder\n- qwen2\n- chat\n- qwen\n- qwen-coder\n- chat\n- qwen\n- qwen-coder\n- moe\n- Qwen3-Coder-30B-A3B-Instruct\n- Qwen3-30B-A3B\n- mixture of experts\n- 128 experts\n- 8 active experts\n- 256k context\n- qwen3\n- gguf\n- NEO Imatrix\n- brainstorm 20x\n- brainstorm\n- optional thinking\n- qwen3_moe\npipeline_tag: text-generation\n---\n\n(Coming soon: Benchmarks / Review)\n\n<h2>Qwen3-42B-A3B-YOYO-V5-TOTAL-RECALL-NEO-imatrix-GGUF // 256k context</h2>\n\n<img src=\"qwen3-total-recall.gif\" style=\"float:right; width:300px; height:300px; padding:10px;\">\n\nThis model is for CODING and programming in all major programming languages and many minor ones too AND GENERAL USAGE.\n\nGGUFs generated from fine tune by DavidAU with Brainstorm 20x and Yoyo's excellent \"Qwen3-30B-A3B-YOYO-V5\" .\n\nThis model is based on Qwen3-Coder-30B-A3B-Instruct (MOE, 128 experts, 8 activated), with Brainstorm 20X \n(by DavidAU) - details at bottom of this page.\n\nThis model is a result of merged model (3 step, 3 models) from:\n\nhttps://huggingface.co/YOYO-AI/Qwen3-30B-A3B-YOYO-V5\n\n(you may want to visit this repo for settings/info too)\n\nThe Brainstorm adapter will improve general performance and \"out of the box\" thinking.\n\nThis creates a model of 42B parameters, 67 layers and 807 tensors.\n\nThis version has the NATIVE context of 256k (262,144) context.\n\nThis is a thinking block model. \n\nI have included an optional system prompt to invoke \"thinking\" in this model, if you want to activate it.\n\nSETTINGS:\n\nFor coding, programming set expert to:\n- 6-8 for general work.\n- 10 for moderate work.\n- 12-16 for complex work, long projects, complex coding.\n- Suggest min context window 4k to 8k.\n- And for longer context, and/or multi-turn -> increase experts by 1-2 to help with longer context/multi turn understanding.\n\nRecommended settings - general:\n- Rep pen 1.05 to 1.1 ; however rep pen of 1 will work well (may need to raise it for lower quants/fewer activated experts)\n- Temp .3 to .6 (+- .2)\n- Topk of 20, 40 or 100\n- Topp of .95 / min p of .05\n- Suggest min context window 4k to 8k.\n- System prompt (optional) to focus the model better.\n\nThis is the refined version -V1.4- from this project (see this repo for all settings, details, system prompts, example generations etc etc):\n\nhttps://huggingface.co/DavidAU/Qwen3-55B-A3B-TOTAL-RECALL-Deep-40X-GGUF/\n\nThis version 2 is slightly smaller, with further refinements to the Brainstorm adapter and uses the new \"Qwen3-30B-A3B-Instruct-2507\".\n\nReview and Specialized Settings for this model (V 1.4):\n\nhttps://www.linkedin.com/posts/gchesler_davidauqwen3-53b-a3b-total-recall-v14-128k-activity-7344938636141858816-ILCM/\n\nhttps://www.linkedin.com/posts/gchesler_haskell-postgres-agentic-activity-7347103276141596672-_zbo/\n\nYou may also want to see (root model of Total Recall series - Version 1):\n\nhttps://huggingface.co/Qwen/Qwen3-30B-A3B\n\nAND Version 2 root model:\n\nhttps://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct\n\nFor additional settings, tool use, and other model settings.\n\nSummary of root model below, followed by FULL HELP SECTION, then info on Brainstorm 40x.\n\nOPTIONAL SYSTEM PROMPT - INVOKE \"Thinking\":\n\n```\nEnable deep thinking subroutine. You are a deep thinking AI, you may use extremely long chains of thought to deeply consider the problem and deliberate with yourself via systematic reasoning processes to help come to a correct solution prior to answering. You should enclose your thoughts and internal monologue inside ###ponder### ###/ponder### tags, and then provide your solution or response to the problem.\n```\n\nUse this to INVOKE \"thinking\" block(s) in the model. These will be a lot shorter than 1000s of tokens generally in most \"thinking\" models.\n\nIn you use this prompt, you may need to raise \"rep pen\" to 1.08 to 1.1, to prevent \"loops\" in the \"thought block(s)\" ; especially in lower quants.\n\nIf you change \"ponder\" to a different word/phrase this will affect model \"thinking\" too.\n\n---\n\nQUANTS\n\n---\n\nQuants are build using NEO Imatrix dataset (by DavidAU), and optimized specifically for thinking block models \nwith the output tensor at 16 bits.\n\nI have found this specific combo to be unbeatable even at IQ2_M.\n\n---\n\nMLX Quants By Nightmedia in partnership with DavidAU:\n\nhttps://huggingface.co/nightmedia/Qwen3-42B-A3B-YOYO-V5-TOTAL-RECALL-qx86x-hi-mlx\n\nhttps://huggingface.co/nightmedia/Qwen3-42B-A3B-YOYO-V5-TOTAL-RECALL-qx86-hi-mlx\n\nhttps://huggingface.co/nightmedia/Qwen3-42B-A3B-YOYO-V5-TOTAL-RECALL-mxfp4-mlx\n\n---\n\n# Qwen3-Coder-3B-A3B-Instruct\n\n## Highlights\n\n**Qwen3-Coder** is available in multiple sizes. Today, we're excited to introduce **Qwen3-Coder-30B-A3B-Instruct**. This streamlined model maintains impressive performance and efficiency, featuring the following key enhancements:  \n\n- **Significant Performance** among open models on **Agentic Coding**, **Agentic Browser-Use**, and other foundational coding tasks.\n- **Long-context Capabilities** with native support for **256K** tokens, extendable up to **1M** tokens using Yarn, optimized for repository-scale understanding.\n- **Agentic Coding** supporting for most platform such as **Qwen Code**, **CLINE**, featuring a specially designed function call format.\n\n![image/jpeg](https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-Coder/qwen3-coder-30a3-main.jpg)\n\n## Model Overview\n\n**Qwen3-Coder-30B-A3B-Instruct** has the following features:\n- Type: Causal Language Models\n- Training Stage: Pretraining & Post-training\n- Number of Parameters: 30.5B in total and 3.3B activated\n- Number of Layers: 48\n- Number of Attention Heads (GQA): 32 for Q and 4 for KV\n- Number of Experts: 128\n- Number of Activated Experts: 8\n- Context Length: **262,144 natively**. \n\n**NOTE: This model supports only non-thinking mode and does not generate ``<think></think>`` blocks in its output. Meanwhile, specifying `enable_thinking=False` is no longer required.**\n\nFor more details, including benchmark evaluation, hardware requirements, and inference performance, please refer to our [blog](https://qwenlm.github.io/blog/qwen3-coder/), [GitHub](https://github.com/QwenLM/Qwen3-Coder), and [Documentation](https://qwen.readthedocs.io/en/latest/).\n\n\n## Quickstart\n\nWe advise you to use the latest version of `transformers`.\n\nWith `transformers<4.51.0`, you will encounter the following error:\n```\nKeyError: 'qwen3_moe'\n```\n\nThe following contains a code snippet illustrating how to use the model generate content based on given inputs. \n```python\nfrom transformers import AutoModelForCausalLM, AutoTokenizer\n\nmodel_name = \"Qwen/Qwen3-Coder-30B-A3B-Instruct\"\n\n# load the tokenizer and the model\ntokenizer = AutoTokenizer.from_pretrained(model_name)\nmodel = AutoModelForCausalLM.from_pretrained(\n    model_name,\n    torch_dtype=\"auto\",\n    device_map=\"auto\"\n)\n\n# prepare the model input\nprompt = \"Write a quick sort algorithm.\"\nmessages = [\n    {\"role\": \"user\", \"content\": prompt}\n]\ntext = tokenizer.apply_chat_template(\n    messages,\n    tokenize=False,\n    add_generation_prompt=True,\n)\nmodel_inputs = tokenizer([text], return_tensors=\"pt\").to(model.device)\n\n# conduct text completion\ngenerated_ids = model.generate(\n    **model_inputs,\n    max_new_tokens=65536\n)\noutput_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() \n\ncontent = tokenizer.decode(output_ids, skip_special_tokens=True)\n\nprint(\"content:\", content)\n```\n\n**Note: If you encounter out-of-memory (OOM) issues, consider reducing the context length to a shorter value, such as `32,768`.**\n\nFor local use, applications such as Ollama, LMStudio, MLX-LM, llama.cpp, and KTransformers have also supported Qwen3.\n\n## Agentic Coding\n\nQwen3-Coder excels in tool calling capabilities. \n\nYou can simply define or use any tools as following example.\n```python\n# Your tool implementation\ndef square_the_number(num: float) -> dict:\n    return num ** 2\n\n# Define Tools\ntools=[\n    {\n        \"type\":\"function\",\n        \"function\":{\n            \"name\": \"square_the_number\",\n            \"description\": \"output the square of the number.\",\n            \"parameters\": {\n                \"type\": \"object\",\n                \"required\": [\"input_num\"],\n                \"properties\": {\n                    'input_num': {\n                        'type': 'number', \n                        'description': 'input_num is a number that will be squared'\n                        }\n                },\n            }\n        }\n    }\n]\n\nimport OpenAI\n# Define LLM\nclient = OpenAI(\n    # Use a custom endpoint compatible with OpenAI API\n    base_url='http://localhost:8000/v1',  # api_base\n    api_key=\"EMPTY\"\n)\n \nmessages = [{'role': 'user', 'content': 'square the number 1024'}]\n\ncompletion = client.chat.completions.create(\n    messages=messages,\n    model=\"Qwen3-Coder-30B-A3B-Instruct\",\n    max_tokens=65536,\n    tools=tools,\n)\n\nprint(completion.choice[0])\n```\n\n## Best Practices\n\nTo achieve optimal performance, we recommend the following settings:\n\n1. **Sampling Parameters**:\n   - We suggest using `temperature=0.7`, `top_p=0.8`, `top_k=20`, `repetition_penalty=1.05`.\n\n2. **Adequate Output Length**: We recommend using an output length of 65,536 tokens for most queries, which is adequate for instruct models.\n\n---\n\n<H2>Help, Adjustments, Samplers, Parameters and More</H2>\n\n---\n\n<B>CHANGE THE NUMBER OF ACTIVE EXPERTS:</B>\n\nSee this document:\n\nhttps://huggingface.co/DavidAU/How-To-Set-and-Manage-MOE-Mix-of-Experts-Model-Activation-of-Experts\n\n<B>Settings: CHAT / ROLEPLAY and/or SMOOTHER operation of this model:</B>\n\nIn \"KoboldCpp\" or  \"oobabooga/text-generation-webui\" or \"Silly Tavern\" ;\n\nSet the \"Smoothing_factor\" to 1.5 \n\n: in KoboldCpp -> Settings->Samplers->Advanced-> \"Smooth_F\"\n\n: in text-generation-webui -> parameters -> lower right.\n\n: In Silly Tavern this is called: \"Smoothing\"\n\n\nNOTE: For \"text-generation-webui\" \n\n-> if using GGUFs you need to use \"llama_HF\" (which involves downloading some config files from the SOURCE version of this model)\n\nSource versions (and config files) of my models are here:\n\nhttps://huggingface.co/collections/DavidAU/d-au-source-files-for-gguf-exl2-awq-gptq-hqq-etc-etc-66b55cb8ba25f914cbf210be\n\nOTHER OPTIONS:\n\n- Increase rep pen to 1.1 to 1.15 (you don't need to do this if you use \"smoothing_factor\")\n\n- If the interface/program you are using to run AI MODELS supports \"Quadratic Sampling\" (\"smoothing\") just make the adjustment as noted.\n\n<B>Highest Quality Settings / Optimal Operation Guide / Parameters and Samplers</B>\n\nThis a \"Class 1\" model:\n\nFor all settings used for this model (including specifics for its \"class\"), including example generation(s) and for advanced settings guide (which many times addresses any model issue(s)), including methods to improve model performance for all use case(s) as well as chat, roleplay and other use case(s) please see:\n\n[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]\n\nYou can see all parameters used for generation, in addition to advanced parameters and samplers to get the most out of this model here:\n\n[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]\n\n---\n\n<H2>What is Brainstorm?</H2>\n\n---\n\n<B>Brainstorm 20x</B>\n\nThe BRAINSTORM process was developed by David_AU.\n\nSome of the core principals behind this process are discussed in this <a href=\"https://arxiv.org/pdf/2401.02415\"> \nscientific paper : Progressive LLaMA with Block Expansion </a>. \n\nHowever I went in a completely different direction from what was outlined in this paper.\n\nWhat is \"Brainstorm\" ?\n\nThe reasoning center of an LLM is taken apart, reassembled, and expanded.\n\nIn this case for this model: 20 times\n\nThen these centers are individually calibrated. These \"centers\" also interact with each other. \nThis introduces subtle changes into the reasoning process. \nThe calibrations further adjust - dial up or down - these \"changes\" further. \nThe number of centers (5x,10x etc) allow more \"tuning points\" to further customize how the model reasons so to speak.\n\nThe core aim of this process is to increase the model's detail, concept and connection to the \"world\", \ngeneral concept connections, prose quality and prose length without affecting instruction following. \n\nThis will also enhance any creative use case(s) of any kind, including \"brainstorming\", creative art form(s) and like case uses.\n\nHere are some of the enhancements this process brings to the model's performance:\n\n- Prose generation seems more focused on the moment to moment. \n- Sometimes there will be \"preamble\" and/or foreshadowing present.\n- Fewer or no \"cliches\"\n- Better overall prose and/or more complex / nuanced prose.\n- A greater sense of nuance on all levels.\n- Coherence is stronger.\n- Description is more detailed, and connected closer to the content.\n- Simile and Metaphors are stronger and better connected to the prose, story, and character.\n- Sense of \"there\" / in the moment is enhanced.\n- Details are more vivid, and there are more of them.\n- Prose generation length can be long to extreme.\n- Emotional engagement is stronger.\n- The model will take FEWER liberties vs a normal model: It will follow directives more closely but will \"guess\" less.\n- The MORE instructions and/or details you provide the more strongly the model will respond.\n- Depending on the model \"voice\" may be more \"human\" vs original model's \"voice\".\n\nOther \"lab\" observations:\n\n- This process does not, in my opinion, make the model 5x or 10x \"smarter\" - if only that was true! \n- However, a change in \"IQ\" was not an issue / a priority, and was not tested or calibrated for so to speak.\n- From lab testing it seems to ponder, and consider more carefully roughly speaking.\n- You could say this process sharpens the model's focus on it's task(s) at a deeper level.\n\nThe process to modify the model occurs at the root level - source files level. The model can quanted as a GGUF, EXL2, AWQ etc etc.\n\n---\n",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "programming",
    "code generation",
    "code",
    "codeqwen",
    "moe",
    "coding",
    "coder",
    "qwen2",
    "chat",
    "qwen",
    "qwen-coder",
    "Qwen3-Coder-30B-A3B-Instruct",
    "Qwen3-30B-A3B",
    "mixture of experts",
    "128 experts",
    "8 active experts",
    "256k context",
    "qwen3",
    "NEO Imatrix",
    "brainstorm 20x",
    "brainstorm",
    "optional thinking",
    "qwen3_moe",
    "text-generation",
    "en",
    "fr",
    "zh",
    "de",
    "arxiv:2401.02415",
    "license:apache-2.0",
    "region:us"
  ],
  "likes": 4,
  "downloads": 143,
  "gated": false,
  "private": false,
  "last_modified": "2025-11-14T23:29:41.000Z",
  "created_at": "2025-11-14T06:47:06.000Z",
  "pipeline_tag": "text-generation",
  "library_name": ""
}

{
  "_id": "6916d06ab38ecaf595d4a26b",
  "id": "DavidAU/Qwen3-42B-A3B-YOYO-V5-TOTAL-RECALL-NEO-imatrix-GGUF",
  "modelId": "DavidAU/Qwen3-42B-A3B-YOYO-V5-TOTAL-RECALL-NEO-imatrix-GGUF",
  "sha": "360700308a23f31dfb59fc1576f36e249a6afcf0",
  "createdAt": "2025-11-14T06:47:06.000Z",
  "lastModified": "2025-11-14T23:29:41.000Z",
  "author": "DavidAU",
  "downloads": 143,
  "likes": 4,
  "gated": false,
  "private": false,
  "pipeline_tag": "text-generation",
  "library_name": "",
  "siblings_count": 12
}