{
  "user": {
    "created_at": "2023-01-05T10:36:53.000Z",
    "default_profile_image": false,
    "description": "Shanghai AI Lab, General Purpose Vision Team. We created InternImage, BEVFormer, VideoMAEv2, LLaMA-Adapter-V2, Ask-Anything, and many more!",
    "fast_followers_count": 0,
    "favourites_count": 17,
    "followers_count": 400,
    "friends_count": 23,
    "has_custom_timelines": false,
    "is_translator": false,
    "listed_count": 7,
    "location": "Shanghai ",
    "media_count": 15,
    "name": "OpenGVLab",
    "normal_followers_count": 400,
    "possibly_sensitive": false,
    "profile_banner_url": "https://pbs.twimg.com/profile_banners/1610948392489979904/1684122162",
    "profile_image_url_https": "https://pbs.twimg.com/profile_images/1657956546763161600/cihLDJvj_normal.jpg",
    "screen_name": "opengvlab",
    "statuses_count": 37,
    "translator_type": "none",
    "url": "https://t.co/pMel5nhAIV",
    "verified": false,
    "withheld_in_countries": [],
    "id_str": "1610948392489979904"
  },
  "id": "1697091590630862954",
  "conversation_id": "1697091590630862954",
  "full_text": "Thank AK @_akhaliq  for the post.\n🔥 Excited to introduce OmniQuant - An advanced open-source algorithm for compressing large language models!\n📜 Paper: https://t.co/i62FV71HAg\n🔗 Code: https://t.co/oPxDdySKtU\n💡 Key Features:\n🚀Omnidirectional Calibration: Enables easier weight and activation quantization through block-wise differentiation.\n🛠 Diverse Precisions: Supports both weight-only quantization (W4A16/W3A16/W2A16) and weight-activation quantization (W6A6, W4A4).\n⚡ Efficient: Quantize LLaMa-2 family (7B-70B) in just 1 to 16 hours using 128 samples.\n🤖 LLM Models: Works with diverse model families, including OPT, WizardLM @WizardLM_AI, LLaMA, LLaMA-2, and LLaMA-2-chat.\n🔑 Deployment: Offers out-of-the-box deployment cases for GPUs and mobile phones.\n🏃Comming Soon: Multi-modal models and CodeLLaMa quantization!",
  "reply_count": 1,
  "retweet_count": 24,
  "favorite_count": 71,
  "hashtags": [],
  "symbols": [],
  "user_mentions": [
    {
      "id_str": "2465283662",
      "name": "AK",
      "screen_name": "_akhaliq",
      "profile": "https://twitter.com/_akhaliq"
    }
  ],
  "urls": [
    {
      "url": "https://t.co/IqqJtJrKFQ",
      "expanded_url": "https://arxiv.org/abs/2308.13137",
      "display_url": "arxiv.org/abs/2308.13137"
    },
    {
      "url": "https://t.co/bO8io64cF5",
      "expanded_url": "https://github.com/OpenGVLab/OmniQuant",
      "display_url": "github.com/OpenGVLab/Omni…"
    }
  ],
  "media": [
    {
      "media_url": "https://pbs.twimg.com/media/F41HBJSbwAAH8GX.jpg",
      "type": "photo"
    }
  ],
  "url": "https://twitter.com/opengvlab/status/1697091590630862954",
  "created_at": "2023-08-31T03:39:00.000Z",
  "#sort_index": "1697091590630862954",
  "view_count": 20746,
  "quote_count": 1,
  "is_quote_tweet": true,
  "is_retweet": false,
  "is_pinned": false,
  "is_truncated": true,
  "quoted_tweet": {
    "user": {
      "created_at": "2014-04-27T00:20:12.000Z",
      "default_profile_image": false,
      "description": "AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗)\n\ndm for promo",
      "fast_followers_count": 0,
      "favourites_count": 26248,
      "followers_count": 229989,
      "friends_count": 1862,
      "has_custom_timelines": true,
      "is_translator": false,
      "listed_count": 3046,
      "location": "subscribe → ",
      "media_count": 13651,
      "name": "AK",
      "normal_followers_count": 229989,
      "possibly_sensitive": false,
      "profile_banner_url": "https://pbs.twimg.com/profile_banners/2465283662/1610997549",
      "profile_image_url_https": "https://pbs.twimg.com/profile_images/1451191636810092553/kpM5Fe12_normal.jpg",
      "screen_name": "_akhaliq",
      "statuses_count": 21274,
      "translator_type": "none",
      "url": "https://t.co/TbGnXZJwEc",
      "verified": false,
      "withheld_in_countries": [],
      "id_str": "2465283662"
    },
    "id": "1696057203978100775",
    "conversation_id": "1696057203978100775",
    "full_text": "OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models\n\npaper page: https://t.co/UoeEJ6xGDs\n\nLarge language models (LLMs) have revolutionized natural language processing tasks. However, their practical deployment is hindered by their immense memory and computation requirements. Although recent post-training quantization (PTQ) methods are effective in reducing memory footprint and improving the computational efficiency of LLM, they hand-craft quantization parameters, which leads to low performance and fails to deal with extremely low-bit quantization. To tackle this issue, we introduce an Omnidirectionally calibrated Quantization (OmniQuant) technique for LLMs, which achieves good performance in diverse quantization settings while maintaining the computational efficiency of PTQ by efficiently optimizing various quantization parameters. OmniQuant comprises two innovative components including Learnable Weight Clipping (LWC) and Learnable Equivalent Transformation (LET). LWC modulates the extreme values of weights by optimizing the clipping threshold. Meanwhile, LET tackles activation outliers by shifting the challenge of quantization from activations to weights through a learnable equivalent transformation. Operating within a differentiable framework using block-wise error minimization, OmniQuant can optimize the quantization process efficiently for both weight-only and weight-activation quantization. For instance, the LLaMA-2 model family with the size of 7-70B can be processed with OmniQuant on a single A100-40G GPU within 1-16 hours using 128 samples. Extensive experiments validate OmniQuant's superior performance across diverse quantization configurations such as W4A4, W6A6, W4A16, W3A16, and W2A16. Additionally, OmniQuant demonstrates effectiveness in instruction-tuned models and delivers notable improvements in inference speed and memory reduction on real devices.",
    "reply_count": 4,
    "retweet_count": 33,
    "favorite_count": 140,
    "hashtags": [],
    "symbols": [],
    "user_mentions": [],
    "urls": [
      {
        "url": "https://t.co/bVs75DbqQM",
        "expanded_url": "https://huggingface.co/papers/2308.13137",
        "display_url": "huggingface.co/papers/2308.13…"
      }
    ],
    "media": [
      {
        "media_url": "https://pbs.twimg.com/media/F4mbAdoawAAOREJ.png",
        "type": "photo"
      }
    ],
    "url": "https://twitter.com/_akhaliq/status/1696057203978100775",
    "created_at": "2023-08-28T07:08:43.000Z",
    "#sort_index": "1697091590630862800",
    "view_count": 54397,
    "quote_count": 2,
    "is_quote_tweet": false,
    "is_retweet": false,
    "is_pinned": false,
    "is_truncated": true
  },
  "startUrl": "https://twitter.com/opengvlab/status/1697091590630862954"
}