mradermacher

/

DeepSeek-V3-Base-GGUF

like 3

llama_model_load: error loading model: tensor 'blk.8.ffn_up_exps.weight' data is not within the file bounds, model is corrupted or incomplete

That error looks as if you did not properly concatenate the files.

Make sure to use the following command to concatinate them after downloading:

cat DeepSeek-V3-Base.Q4_K_M.gguf.part1of9 DeepSeek-V3-Base.Q4_K_M.gguf.part2of9 DeepSeek-V3-Base.Q4_K_M.gguf.part3of9 DeepSeek-V3-Base.Q4_K_M.gguf.part4of9 DeepSeek-V3-Base.Q4_K_M.gguf.part5of9 DeepSeek-V3-Base.Q4_K_M.gguf.part6of9 DeepSeek-V3-Base.Q4_K_M.gguf.part7of9 DeepSeek-V3-Base.Q4_K_M.gguf.part8of9 DeepSeek-V3-Base.Q4_K_M.gguf.part9of9 > DeepSeek-V3-Base.Q4_K_M.gguf

Alternatively download the model from https://hf.tst.eu/model#DeepSeek-V3-Base-GGUF so the resulting model is already concatenated. I also highly recommend to use the much better wighted/imatrix quants available on either the download page or under https://huggingface.co/mradermacher/DeepSeek-V3-Base-i1-GGUF

Should the model really be corrupted don't worry we will soon requantize this model with MLA anyways.