**Source:** [gguf_init_from_file](https://github.com/ggerganov/llama.cpp/blob/master/ggml.c) **Examples:** [all-MiniLM-L6-v2.F16.gguf.zip](https://github.com/user-attachments/files/25527189/all-MiniLM-L6-v2.F16.gguf.zip) [deepseek-ocr-Q4_K_M.gguf.zip](https://github.com/user-attachments/files/27738911/deepseek-ocr-Q4_K_M.gguf.zip) [DeepSeek-V4-Flash-Q2-0731.gguf.zip](https://github.com/user-attachments/files/30635564/DeepSeek-V4-Flash-Q2-0731.gguf.zip) [Falcon-H1-Tiny-90M-Instruct-Q4_K_M.gguf.zip](https://github.com/user-attachments/files/25527191/Falcon-H1-Tiny-90M-Instruct-Q4_K_M.gguf.zip) [flan-t5-small-q2_k.gguf.zip](https://github.com/user-attachments/files/25527190/flan-t5-small-q2_k.gguf.zip) [gemma-3-1b-it-Q4_K_M.gguf.zip](https://github.com/user-attachments/files/27738910/gemma-3-1b-it-Q4_K_M.gguf.zip) [gemma-4-E2B-it-Q8_0.gguf.zip](https://github.com/user-attachments/files/27541805/gemma-4-E2B-it-Q8_0.gguf.zip) [gpt-oss-20b-MXFP4.gguf.zip](https://github.com/user-attachments/files/21702620/gpt-oss-20b-MXFP4.gguf.zip) [HunyuanOCR-Q8_0.gguf.zip](https://github.com/user-attachments/files/27738913/HunyuanOCR-Q8_0.gguf.zip) [Kimi-K3-UD-Q8_K_XL.gguf.zip](https://github.com/user-attachments/files/30564609/Kimi-K3-UD-Q8_K_XL.gguf.zip) [mamba-130m-hf.Q8_0.gguf.zip](https://github.com/user-attachments/files/27738912/mamba-130m-hf.Q8_0.gguf.zip) [Mistral-Small-4-119B-2603-IQ1_S.gguf.zip](https://github.com/user-attachments/files/27738914/Mistral-Small-4-119B-2603-IQ1_S.gguf.zip) [mixtral-8x7b-v0.1.Q4_K_M.gguf.zip](https://github.com/lutzroeder/netron/files/13802180/mixtral-8x7b-v0.1.Q4_K_M.gguf.zip) [phi-2.Q2_K.gguf.zip](https://github.com/lutzroeder/netron/files/13825087/phi-2.Q2_K.gguf.zip) [qwen2.5-coder-7b-instruct-q4_k_m.gguf.zip](https://github.com/user-attachments/files/25527192/qwen2.5-coder-7b-instruct-q4_k_m.gguf.zip) [Qwen3-14B-speculator.eagle3-Q4_K_M.gguf.zip](https://github.com/user-attachments/files/30016436/Qwen3-14B-speculator.eagle3-Q4_K_M.gguf.zip) [vicuna-33b.Q2_K.gguf.zip](https://github.com/lutzroeder/netron/files/13825089/vicuna-33b.Q2_K.gguf.zip) **Instructions:** - Run `./tools/gguf sync install schema metadata` for latest updates - If there are `gguf-metadata.json` changes investigate the reference implementation if related code changes are required. - The reference implementation is under `./third_party/source/llama.cpp` - `npm run test gguf` to run test validation.
Source:
gguf_init_from_file
Examples:
all-MiniLM-L6-v2.F16.gguf.zip
deepseek-ocr-Q4_K_M.gguf.zip
DeepSeek-V4-Flash-Q2-0731.gguf.zip
Falcon-H1-Tiny-90M-Instruct-Q4_K_M.gguf.zip
flan-t5-small-q2_k.gguf.zip
gemma-3-1b-it-Q4_K_M.gguf.zip
gemma-4-E2B-it-Q8_0.gguf.zip
gpt-oss-20b-MXFP4.gguf.zip
HunyuanOCR-Q8_0.gguf.zip
Kimi-K3-UD-Q8_K_XL.gguf.zip
mamba-130m-hf.Q8_0.gguf.zip
Mistral-Small-4-119B-2603-IQ1_S.gguf.zip
mixtral-8x7b-v0.1.Q4_K_M.gguf.zip
phi-2.Q2_K.gguf.zip
qwen2.5-coder-7b-instruct-q4_k_m.gguf.zip
Qwen3-14B-speculator.eagle3-Q4_K_M.gguf.zip
vicuna-33b.Q2_K.gguf.zip
Instructions:
./tools/gguf sync install schema metadatafor latest updatesgguf-metadata.jsonchanges investigate the reference implementation if related code changes are required../third_party/source/llama.cppnpm run test ggufto run test validation.