em-80/qwen3-coder-reap-25b-a3b-rust-gguf overview

{
  "metadata": {},
  "card_data": {
    "license": "apache-2.0",
    "base_model": "cerebras/Qwen3-Coder-REAP-25B-A3B",
    "tags": [
      "code-generation",
      "quantization",
      "qwen3",
      "qwen-coder",
      "cerebras",
      "importance-matrix",
      "imatrix",
      "gguf",
      "coding",
      "MOE",
      "moe",
      "rust",
      "python"
    ],
    "model_creator": "Em-80",
    "model_type": "causal-lm",
    "pipeline_tag": "text-generation",
    "language": [
      "en"
    ],
    "frontmatter": {
      "license": "apache-2.0",
      "base_model": "cerebras/Qwen3-Coder-REAP-25B-A3B",
      "tags": [
        "code-generation",
        "quantization",
        "qwen3",
        "qwen-coder",
        "cerebras",
        "importance-matrix",
        "imatrix",
        "gguf",
        "coding",
        "MOE",
        "moe",
        "rust",
        "python"
      ],
      "model_creator": "Em-80",
      "model_type": "causal-lm",
      "pipeline_tag": "text-generation",
      "language": [
        "en"
      ]
    },
    "hero_image_url": "",
    "summary": "This repository provides high-precision quantized versions of the Qwen3-Coder-REAP-25B-A3B model, featuring a custom importance matrix generated by Em-80, and specifically optimized for Rust code generation. # Work in Progress I included the .imatrix if you need to quant your own version. Just please link to me if you use the imatrix. Once all the quants are uploaded I'll spend a weekend doing benchmarks. # Highlights # Importance Matrix (Imatrix) Details The included .imatrix file was developed to ensure that lower-bit quants retain as much intelligence as possible. Unlike standard \"blind\" quants that use generic calibration data, this imatrix was derived from a curated 3,000-sample dataset(21.8 MB) covering: By using this importance matrix during the quantization process, we ensure that the weights most critical for code generation and complex problem-solving are preserved with higher fidelity. # Files Included # Licensing and Attribution This work is a derivative of the Qwen3-Coder-REAP-25B-A3B model by Cerebras Systems and Qwen3-Coder-30B-A3B-Instruct by Alibaba Cloud. Please refer to the LICENSE file in this repository for full legal terms and modification notices. # Usage To use these quants with or quant your own Qwen3-Coder-REAP-25B-A3B with the provided imatrix in llama.cpp: # Example for running a quant ./main -m Qwen3-Coder-REAP-25B-A3B-Rust-IQ3_M.gguf -p \"Write a thread-safe singleton in Rust.\" Notice: This model is provided \"as-is\" without warranty of any kind. Use at your own risk.",
    "quick_links": [],
    "benchmark_table_html": "",
    "readme_markdown": "---\nlicense: apache-2.0\nbase_model: cerebras/Qwen3-Coder-REAP-25B-A3B\ntags:\n- code-generation\n- quantization\n- qwen3\n- qwen-coder\n- cerebras\n- importance-matrix\n- imatrix\n- gguf\n- coding\n- MOE\n- moe\n- rust\n- python\nmodel_creator: Em-80\nmodel_type: causal-lm\npipeline_tag: text-generation\nlanguage:\n- en\n---\n\n# Qwen3-coder-REAP-25B-A3B-Rust-GGUF\n\nThis repository provides high-precision quantized versions of the Qwen3-Coder-REAP-25B-A3B model, featuring a custom importance matrix generated by Em-80, and specifically optimized for Rust code generation.\n\n# Work in Progress \n\nI included the .imatrix if you need to quant your own version. Just please link to me if you use the imatrix. Once all the quants are uploaded I'll spend a weekend doing benchmarks. \n\n- All the quants I'm planning to do are up.\n\n- Benchmarks I'm shooting for the week of april first.\n  \n# Highlights\n\n- Model Architecture: Based on the Cerebras REAP (Router-weighted Expert Activation Pruning) variant of Qwen3-Coder-30B-A3B-Instruct.\n\n- Custom Imatrix: Includes Qwen3-Coder-REAP-25B-A3B-Rust.imatrix, generated using a diverse and high-density calibration set.\n\n- Optimized for Logic: The quantization process focused heavily on maintaining the model's multi-lingual coding and mathematical reasoning capabilities.\n\n# Importance Matrix (Imatrix) Details\n\nThe included .imatrix file was developed to ensure that lower-bit quants retain as much intelligence as possible. Unlike standard \"blind\" quants that use generic calibration data, this imatrix was derived from a curated 3,000-sample dataset(21.8 MB) covering:\n\n- Programming: Deep coverage of Rust and Python syntax, logic, and idiomatic patterns.\n\n- Reasoning: Advanced Mathematics and logical proofs.\n\n- Linguistic Quality: High-quality English prose.\n\nBy using this importance matrix during the quantization process, we ensure that the weights most critical for code generation and complex problem-solving are preserved with higher fidelity.\n\n# Files Included\n\n- Weights: Multiple quantization levels (GGUF).\n\n- Metadata: Qwen3-Coder-REAP-25B-A3B-Rust.imatrix for users who wish to perform their own custom quantization runs.\n\n# Licensing and Attribution\n\nThis work is a derivative of the Qwen3-Coder-REAP-25B-A3B model by Cerebras Systems and Qwen3-Coder-30B-A3B-Instruct by Alibaba Cloud.\n\n- Weights & Imatrix: Released under the Apache License 2.0.\n\n- Attribution: Modifications, quantization, and imatrix generation performed by Em-80.\n\nPlease refer to the LICENSE file in this repository for full legal terms and modification notices.\n\n# Usage\n\nTo use these quants with or quant your own Qwen3-Coder-REAP-25B-A3B with the provided imatrix in llama.cpp:\n\n# Example for running a quant\n./main -m Qwen3-Coder-REAP-25B-A3B-Rust-IQ3_M.gguf -p \"Write a thread-safe singleton in Rust.\"\n\n\nNotice: This model is provided \"as-is\" without warranty of any kind. Use at your own risk.",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "code-generation",
    "quantization",
    "qwen3",
    "qwen-coder",
    "cerebras",
    "importance-matrix",
    "imatrix",
    "coding",
    "MOE",
    "moe",
    "rust",
    "python",
    "text-generation",
    "en",
    "base_model:cerebras/Qwen3-Coder-REAP-25B-A3B",
    "base_model:quantized:cerebras/Qwen3-Coder-REAP-25B-A3B",
    "license:apache-2.0",
    "endpoints_compatible",
    "region:us",
    "conversational"
  ],
  "likes": 7,
  "downloads": 4011,
  "gated": false,
  "private": false,
  "last_modified": "2026-03-26T16:17:13.000Z",
  "created_at": "2026-01-21T19:25:38.000Z",
  "pipeline_tag": "text-generation",
  "library_name": ""
}

{
  "_id": "69712832a8bd6a529faea4c2",
  "id": "Em-80/Qwen3-coder-REAP-25B-A3B-Rust-GGUF",
  "modelId": "Em-80/Qwen3-coder-REAP-25B-A3B-Rust-GGUF",
  "sha": "0c47d1fd9e5c7bf3057a73668cfedbd2f58325d4",
  "createdAt": "2026-01-21T19:25:38.000Z",
  "lastModified": "2026-03-26T16:17:13.000Z",
  "author": "Em-80",
  "downloads": 4011,
  "likes": 7,
  "gated": false,
  "private": false,
  "pipeline_tag": "text-generation",
  "library_name": "",
  "siblings_count": 29
}