[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-663":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":15,"stars7d":16,"stars30d":17,"stars90d":15,"forks30d":15,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":23,"hasPages":21,"topics":24,"createdAt":9,"pushedAt":9,"updatedAt":25,"readmeContent":26,"aiSummary":27,"trendingCount":15,"starSnapshotCount":15,"syncStatus":28,"lastSyncTime":29,"discoverSource":30},663,"grok-1","xai-org\u002Fgrok-1","xai-org","Grok open release",null,"Python",51682,8480,637,125,0,10,165,4,45,"Apache License 2.0",false,"main",true,[],"2026-06-12 02:00:16","# Grok-1\n\nThis repository contains JAX example code for loading and running the Grok-1 open-weights model.\n\nMake sure to download the checkpoint and place the `ckpt-0` directory in `checkpoints` - see [Downloading the weights](#downloading-the-weights)\n\nThen, run\n\n```shell\npip install -r requirements.txt\npython run.py\n```\n\nto test the code.\n\nThe script loads the checkpoint and samples from the model on a test input.\n\nDue to the large size of the model (314B parameters), a machine with enough GPU memory is required to test the model with the example code.\nThe implementation of the MoE layer in this repository is not efficient. The implementation was chosen to avoid the need for custom kernels to validate the correctness of the model.\n\n# Model Specifications\n\nGrok-1 is currently designed with the following specifications:\n\n- **Parameters:** 314B\n- **Architecture:** Mixture of 8 Experts (MoE)\n- **Experts Utilization:** 2 experts used per token\n- **Layers:** 64\n- **Attention Heads:** 48 for queries, 8 for keys\u002Fvalues\n- **Embedding Size:** 6,144\n- **Tokenization:** SentencePiece tokenizer with 131,072 tokens\n- **Additional Features:**\n  - Rotary embeddings (RoPE)\n  - Supports activation sharding and 8-bit quantization\n- **Maximum Sequence Length (context):** 8,192 tokens\n\n# Downloading the weights\n\nYou can download the weights using a torrent client and this magnet link:\n\n```\nmagnet:?xt=urn:btih:5f96d43576e3d386c9ba65b883210a393b68210e&tr=https%3A%2F%2Facademictorrents.com%2Fannounce.php&tr=udp%3A%2F%2Ftracker.coppersurfer.tk%3A6969&tr=udp%3A%2F%2Ftracker.opentrackr.org%3A1337%2Fannounce\n```\n\nor directly using [HuggingFace 🤗 Hub](https:\u002F\u002Fhuggingface.co\u002Fxai-org\u002Fgrok-1):\n```\ngit clone https:\u002F\u002Fgithub.com\u002Fxai-org\u002Fgrok-1.git && cd grok-1\npip install huggingface_hub[hf_transfer]\nhuggingface-cli download xai-org\u002Fgrok-1 --repo-type model --include ckpt-0\u002F* --local-dir checkpoints --local-dir-use-symlinks False\n```\n\n# License\n\nThe code and associated Grok-1 weights in this release are licensed under the\nApache 2.0 license. The license only applies to the source files in this\nrepository and the model weights of Grok-1.\n","Grok-1 是一个基于 JAX 的开源模型项目，旨在加载和运行 Grok-1 模型。该项目的核心功能包括使用 Mixture of 8 Experts (MoE) 架构、支持激活分片和8位量化等特性，具备314B参数规模。技术特点上，Grok-1采用了旋转位置编码（RoPE），并且其最大序列长度可达8,192个token。由于模型体积庞大，需要具有足够GPU内存的机器才能进行测试。此项目适用于需要高效处理大规模文本数据的应用场景，如自然语言处理任务中的文本生成与理解。",2,"2026-06-11 02:38:30","top_all"]