21 lines
883 B
Plaintext
21 lines
883 B
Plaintext
*funcional para quantizer Q8_0 e MXFP4_MOE
|
||
|
||
|
||
|
||
|
||
./llama-quantize /home/userk21/GGUFS/ΩFFΣLLIαHLX_Qwen3.5-35B_MXFP4_MOE.gguf_f16.gguf /home/userk21/GGUFS/ΩFFΣLLIαHLX_Qwen3.5-35B_MXFP4_MOE.gguf MXFP4_MOE
|
||
|
||
python /home/userk21/llama-pure/convert_hf_to_gguf.py /home/userk21/GGUFS/Qwen3.5-27B/ --outtype f32 --outfile /home/userk21/GGUFS/ΩFFΣLLIαHLX_Qwen3.5-27B_f32.gguf
|
||
|
||
|
||
python /home/userk21/llama-pure/convert_hf_to_gguf.py /home/userk21/GGUFS/Qwen3.5-27B/ --outtype f16 --outfile /home/userk21/GGUFS/ΩFFΣLLIαHLX_Qwen3.5-27B_f16.gguf
|
||
|
||
python /home/userk21/llama-pure/convert_hf_to_gguf.py /home/userk21/GGUFS/Qwen3.5-27B/ --outtype q8_0 --outfile /home/userk21/GGUFS/ΩFFΣLLIαHLX_Qwen3.5-27B_q8_0.gguf
|
||
|
||
|
||
/home/userk21/llama-pure/build/bin/llama-server \
|
||
-m /home/userk21/GGUFS/ΩFFΣLLIαHLX_Qwen3.5-35B_MXFP4_MOE.gguf
|
||
--ignore-eos \
|
||
--temp 0.8 \
|
||
--repeat-penalty 1.15 \
|