vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 01:24:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

some cleanup before starting on TTS

Browse source
This commit is contained in:
Concedo 2025-01-10 22:13:44 +08:00
parent 93b2bebc2f
commit bd38665e1f
3 changed files with 16 additions and 15 deletions
Download patch file Download diff file Expand all files Collapse all files

7
src/llama-quant.cpp
View file

@ -397,7 +397,10 @@ static ggml_type llama_tensor_get_type(quantize_state_impl & qs, ggml_type new_t
case GGML_TYPE_Q4_K: new_type = GGML_TYPE_Q5_0; break;
case GGML_TYPE_Q5_K: new_type = GGML_TYPE_Q5_1; break;
case GGML_TYPE_Q6_K: new_type = GGML_TYPE_Q8_0; break;
default: throw std::runtime_error("\nUnsupported tensor size encountered\n");
default:
printf("\nUnsupported tensor size encountered! Will use %s for %s\n",ggml_type_name(tensor->type),name.c_str()) ;
new_type = tensor->type;
break;
}
if (tensor->ne[0] % ggml_blck_size(new_type) != 0) {
new_type = GGML_TYPE_F16;
@ -629,7 +632,7 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
if (llama_model_has_encoder(&model)) {
n_attn_layer *= 3;
}
GGML_ASSERT((qs.n_attention_wv == n_attn_layer) && "n_attention_wv is unexpected");
GGML_ASSERT_CONTINUE((qs.n_attention_wv == n_attn_layer) && "n_attention_wv is unexpected");
}
size_t total_size_org = 0;