Can't load DeepSeek-R1-Distill-Qwen-32B-Q8_0.gguf to LM Studio
#5
by
vsenn
- opened
🥲 Failed to load the model
Failed to load model
llama.cpp error: 'error loading model vocabulary: unknown pre-tokenizer type: 'deepseek-r1-qwen''
Same issue here
Just update to the last beta version LM Studio + runtimes.
Main lmstudio should be fine, 0.3.7, with latest runtimes
Thank you. Confirmed model loading on LM Studio 0.3.7 (Build 2) Runtime v1.9.2
vsenn
changed discussion status to
closed
LM Studio 0.3.7 (Build 2) Runtime v1.9.2 has the same error, my model is DeepSeek-R1-Distill-Qwen-32B-Q6_K_L.gguf, llama.cpp error: 'error loading model vocabulary: unknown pre-tokenizer type: 'deepseek-r1-qwen''
Yeah it gives the same error message when loading the model in even in 0.3.7
Can you confirm you're loading 1.9.2? Maybe a screenshot?
My model quant is Q8_0 as mentioned in the Title. It works in 0.3.7 (Build 2) Runtime v1.9.2. Colleagues mention Q6_K_L quant.
There is already 0.3.8 b2 (1.9.2) in beta channel.