{"v":1,"id":"model:hf:llm-jp/llm-jp-4-33b-thinking-gguf","slug":"model-llm-jp-llm-jp-4-33b-thinking-gguf","kind":"model","category":"llm","title":"llm-jp-4-33b-thinking-gguf","summary":"LLM-jp-4 is a series of large language models developed by the Research and Development Center for Large Language Models at the National Institute of Informatics.","source":{"provider":"hf","ref":"llm-jp/llm-jp-4-33b-thinking-gguf","url":"https://huggingface.co/llm-jp/llm-jp-4-33b-thinking-gguf","rev":"f80f6860a584a44f9119435a80966a7c3a4c12da","fetchedAt":"2026-10-02T21:01:08.615Z","etag":"W/\"4ed1-iLSmexBI1yyq+Q6y8NHductHINE\""},"author":{"name":"llm-jp","url":"https://huggingface.co/llm-jp"},"license":{"spdx":"apache-2.0","raw":"apache-2.0","open":true},"metrics":{"downloads":310411,"downloadsWeek":327838,"likes":11,"takenAt":"2026-10-02T21:01:08.615Z"},"tags":["transformers","gguf","text-generation","en","ja","conversational"],"pipeline":"text-generation","links":{"github":"ggml-org/llama.cpp","npm":"node-llama-cpp"},"updatedAt":"2026-08-20T01:57:25.000Z","collectedAt":"2026-10-02T21:01:08.615Z","review":{"numbers":["310,411 downloads on Hugging Face","11 likes","license apache-2.0","19 GB for llm-jp-4-33b-thinking-Q4_K_M.gguf","327,838 npm downloads a week for node-llama-cpp","latest node-llama-cpp@3.22.1"],"log":null},"trust":"unlabeled","health":{"status":"alive","checkedAt":"2026-10-02T21:01:08.615Z","http":200},"description":"# llm-jp-4-33b-thinking-gguf\n\nLLM-jp-4 is a series of large language models developed by the [Research and Development Center for Large Language Models](https://llmc.nii.ac.jp/) at the [National Institute of Informatics](https://www.nii.ac.jp/en/).\n\nThis repository provides the **llm-jp-4-33b-thinking-gguf**.\nFor an overview of the LLM-jp-4 models across different parameter sizes, please refer to:\n  - [LLM-jp-4 Models](https://huggingface.co/collections/llm-jp/llm-jp-4-models)\n\nBase models are trained with pre-training and mid-training only.\nPost-trained models are aligned using supervised fine-tuning (SFT) and direct preference optimization (DPO), without reinforcement learning.\n> [!NOTE]\n> While the **thinking** variants are trained with both SFT and DPO, this **instruct** model is trained using SFT only, without DPO.\n\nFor practical usage examples and detailed instructions on how to use the models, please also refer to our [cookbook](https://github.com/llm-jp/llm-jp-4-cookbook).\n\nTo support the continued development of LLM-jp, we would greatly appreciate it if you could share how you utilize LLM-jp outcomes via the [survey form](https://forms.gle/AvbNXTNT2ADsssHq5).\n\n## Usage\n\nPlease refer to our [cookbook](https://github.com/llm-jp/llm-jp-4-cookbook) for practical usage examples and detailed instructions on how to use the models.\n\n> [!IMPORTANT]\n> Running this model with `llama.cpp` currently requires the [LLM-jp fork of `llama.cpp`](https://github.com/llm-jp/llama.cpp). The upstream `ggml-org/llama.cpp` does not yet include the required tokenizer-handling fixes, so chat parsing will fail for this model when using it as-is. See the [LLM-jp-4 llama.cpp guide](https://github.com/llm-jp/llm-jp-4-cookbook/tree/main/llmjp4_llama-cpp) for build and usage instructions.\n\n## Model Details\n\n- **Model type:** Transformer-based Language Model\n- **Architectures:**\n\nDense model:\n|Params|Layers|Hidden size|Heads…\n\nSource: https://huggingface.co/llm-jp/llm-jp-4-33b-thinking-gguf","install":{"kind":"model","hfId":"llm-jp/llm-jp-4-33b-thinking-gguf","gated":false,"format":"gguf","files":[{"name":"llm-jp-4-33b-thinking-BF16.gguf","size":66445272576,"quant":"BF16","sha256":"7bb8465702b5c4a5d94e03e62921d5917181a2eda56da31edfdcd47fab2e9964"},{"name":"llm-jp-4-33b-thinking-Q4_K_M.gguf","size":20163749728,"quant":"Q4_K_M","sha256":"9e48892c0d5ec256d05fc3258c9e50738852c39636fbb9fe1f4b1cde2f5e7520"}],"totalBytes":86609022304,"suggestedFile":"llm-jp-4-33b-thinking-Q4_K_M.gguf","requirements":{"ramGb":23,"diskBytes":20163749728,"note":"estimate: suggested file size × 1.15 + 0.5 GB; a real measurement comes with lsh models"},"runWith":["llama.cpp","ollama"],"command":"lsh models install hf:llm-jp/llm-jp-4-33b-thinking-gguf"}}