Unable to finetune using llama cpp
I am getting error while finetuning phi-2 .
Same is working when trying to finetune llama-7b.
Using this command to finetune -> llama.cpp/finetune --model-base phi-2.Q2_K.gguf --train-data input.jsonl --threads 14 --sample-start ""
Checked the code. Seems like expected_arch is always hardcoded to llama
Error -
llama_new_context_with_model: n_ctx = 512
llama_new_context_with_model: freq_base = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init: CPU KV buffer size = 160.00 MiB
llama_new_context_with_model: KV self size = 160.00 MiB, K (f16): 80.00 MiB, V (f16): 80.00 MiB
llama_new_context_with_model: CPU input buffer size = 7.01 MiB
llama_new_context_with_model: CPU compute buffer size = 105.00 MiB
llama_new_context_with_model: graph splits (measure): 1
load_model_hparams_gguf: arch=phi2 expected_arch=llama
GGML_ASSERT: examples/finetune/finetune.cpp:242: arch == expected_arch
zsh: abort llama.cpp/finetune --model-base phi-2.Q2_K.gguf --train-data input.jsonl 14