Describe the Issue
Version 1.119 crashes upon trying to load Ministral-3-14B-Instruct-2512-Q4_K_M.gguf, with the error message below. Which backend I select does not seem to matter, which I find unsurprising, given the error message. Versions through 1.118.1 are able to load this model fine. I'm not changing any settings (other than the backend) from their defaults.
Additional Information:
Logs:
llama_model_loader: loaded meta data with 53 key-value pairs and 363 tensors from C:\...\Ministral-3-14B-Instruct-2512-Q4_K_M.gguf (version GGUF V3 (latest))
print_info: file format = GGUF V3 (latest)
print_info: file type = unknown, may not work
print_info: file size = 7.67 GiB (4.88 BPW)
llama_model_load: error loading model: error loading model vocabulary: invalid gguf type for tokenizer.ggml.scores
llama_model_load_from_file_impl: failed to load model
Traceback (most recent call last):
File "koboldcpp.py", line 12954, in <module>
main(launch_args=parser.parse_args(),default_args=parser.parse_args([]))
File "koboldcpp.py", line 11485, in main
kcpp_main_process(args,global_memory,using_gui_launcher)
File "koboldcpp.py", line 12205, in kcpp_main_process
loadok = load_model(modelname)
File "koboldcpp.py", line 2131, in load_model
ret = handle.load_model(inputs)
OSError: exception: access violation reading 0x000000000000000C
[8932] Failed to execute script 'koboldcpp' due to unhandled exception!
I'm using the official Q4_K_M quant from https://huggingface.co/mistralai/Ministral-3-14B-Instruct-2512-GGUF.
Describe the Issue
Version 1.119 crashes upon trying to load
Ministral-3-14B-Instruct-2512-Q4_K_M.gguf, with the error message below. Which backend I select does not seem to matter, which I find unsurprising, given the error message. Versions through 1.118.1 are able to load this model fine. I'm not changing any settings (other than the backend) from their defaults.Additional Information:
Logs:
I'm using the official Q4_K_M quant from https://huggingface.co/mistralai/Ministral-3-14B-Instruct-2512-GGUF.