prithivmlmods/sage-mm-qwen2.5-vl-7b-sft-gguf overview

{
  "metadata": {},
  "card_data": {
    "license": "apache-2.0",
    "language": [
      "en"
    ],
    "base_model": [
      "allenai/SAGE-MM-Qwen2.5-VL-7B-SFT"
    ],
    "pipeline_tag": "video-text-to-text",
    "library_name": "transformers",
    "tags": [
      "text-generation-inference",
      "llama.cpp"
    ],
    "frontmatter": {
      "license": "apache-2.0",
      "language": [
        "en"
      ],
      "base_model": [
        "allenai/SAGE-MM-Qwen2.5-VL-7B-SFT"
      ],
      "pipeline_tag": "video-text-to-text",
      "library_name": "transformers",
      "tags": [
        "text-generation-inference",
        "llama.cpp"
      ]
    },
    "hero_image_url": "https://www.nethype.de/huggingface_embed/quantpplgraph.png",
    "summary": "> The SAGE-MM-Qwen2.5-VL-7B-SFT from allenai is a 7B-parameter vision-language model fine-tuned from Qwen/Qwen2.5-VL-7B-Instruct, serving as the core decision-maker in the SAGE (Smart Any-Horizon Agent) system for long video reasoning through a two-stage process: Stage-1 analyzes initial sampled frames and metadata to classify queries as single-turn (immediate answers) or multi-turn (tool-required), while Stage-2 iteratively generates JSON-formatted tool calls for web-search, speech transcription on timestamps, event grounding, video part extraction, and detailed visual analysis to build context progressively [attached_file:1 equivalent]. Designed to handle arbitrary-length videos beyond fixed horizons—like sports events, narratives, or complex timelines—it requires the SAGE GitHub runtime for tool parsing/execution and observation feedback, enabling robust Q&A via dynamic tool orchestration under Apache 2.0 for research/educational use per Ai2 guidelines, with GGUF quantizations for efficient deployment. This SFT variant powers the SAGE framework's superior performance on benchmarks like MINERVA, outperforming prior Qwen3-VL-4B baselines in extended video comprehension.",
    "quick_links": [],
    "benchmark_table_html": "",
    "readme_markdown": "---\nlicense: apache-2.0\nlanguage:\n- en\nbase_model:\n- allenai/SAGE-MM-Qwen2.5-VL-7B-SFT\npipeline_tag: video-text-to-text\nlibrary_name: transformers\ntags:\n- text-generation-inference\n- llama.cpp\n---\n\n# **SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF**\n\n> The SAGE-MM-Qwen2.5-VL-7B-SFT from allenai is a 7B-parameter vision-language model fine-tuned from Qwen/Qwen2.5-VL-7B-Instruct, serving as the core decision-maker in the SAGE (Smart Any-Horizon Agent) system for long video reasoning through a two-stage process: Stage-1 analyzes initial sampled frames and metadata to classify queries as single-turn (immediate answers) or multi-turn (tool-required), while Stage-2 iteratively generates JSON-formatted tool calls for web-search, speech transcription on timestamps, event grounding, video part extraction, and detailed visual analysis to build context progressively [attached_file:1 equivalent]. Designed to handle arbitrary-length videos beyond fixed horizons—like sports events, narratives, or complex timelines—it requires the SAGE GitHub runtime for tool parsing/execution and observation feedback, enabling robust Q&A via dynamic tool orchestration under Apache 2.0 for research/educational use per Ai2 guidelines, with GGUF quantizations for efficient deployment. This SFT variant powers the SAGE framework's superior performance on benchmarks like MINERVA, outperforming prior Qwen3-VL-4B baselines in extended video comprehension.\n\n## SAGE-MM-Qwen2.5-VL-7B-SFT [GGUF]\n\n| File Name | Quant Type | File Size | File Link |\n| - | - | - | - |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.IQ4_XS.gguf | IQ4_XS | 4.25 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.IQ4_XS.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q2_K.gguf | Q2_K | 3.02 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q2_K.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q3_K_L.gguf | Q3_K_L | 4.09 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q3_K_L.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q3_K_M.gguf | Q3_K_M | 3.81 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q3_K_M.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q3_K_S.gguf | Q3_K_S | 3.49 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q3_K_S.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q4_K_M.gguf | Q4_K_M | 4.68 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q4_K_M.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q4_K_S.gguf | Q4_K_S | 4.46 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q4_K_S.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q5_K_M.gguf | Q5_K_M | 5.44 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q5_K_M.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q5_K_S.gguf | Q5_K_S | 5.32 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q5_K_S.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q6_K.gguf | Q6_K | 6.25 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q6_K.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.Q8_0.gguf | Q8_0 | 8.1 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.Q8_0.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.f16.gguf | F16 | 15.2 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.f16.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.mmproj-Q8_0.gguf | mmproj-Q8_0 | 856 MB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.mmproj-Q8_0.gguf) |\n| SAGE-MM-Qwen2.5-VL-7B-SFT.mmproj-f16.gguf | mmproj-f16 | 1.35 GB | [Download](https://huggingface.co/prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF/blob/main/SAGE-MM-Qwen2.5-VL-7B-SFT.mmproj-f16.gguf) |\n\n## Quants Usage \n\n(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)\n\nHere is a handy graph by ikawrakow comparing some lower-quality quant\ntypes (lower is better):\n\n![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)",
    "related_quantizations": []
  },
  "tags": [
    "transformers",
    "gguf",
    "qwen2_5_vl",
    "text-generation-inference",
    "llama.cpp",
    "video-text-to-text",
    "en",
    "base_model:allenai/SAGE-MM-Qwen2.5-VL-7B-SFT",
    "base_model:quantized:allenai/SAGE-MM-Qwen2.5-VL-7B-SFT",
    "license:apache-2.0",
    "endpoints_compatible",
    "region:us",
    "conversational"
  ],
  "likes": 1,
  "downloads": 87,
  "gated": false,
  "private": false,
  "last_modified": "2025-12-20T05:04:15.000Z",
  "created_at": "2025-12-18T09:08:21.000Z",
  "pipeline_tag": "video-text-to-text",
  "library_name": "transformers"
}

{
  "_id": "6943c485c2caeb482b3026f3",
  "id": "prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF",
  "modelId": "prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF",
  "sha": "f05cc74ad4bcee296d8d5041d1cdf9778802bcd5",
  "createdAt": "2025-12-18T09:08:21.000Z",
  "lastModified": "2025-12-20T05:04:15.000Z",
  "author": "prithivMLmods",
  "downloads": 87,
  "likes": 1,
  "gated": false,
  "private": false,
  "pipeline_tag": "video-text-to-text",
  "library_name": "transformers",
  "siblings_count": 17
}