@@ -15479,11 +15479,11 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n
15479
15479
if (arch == LLM_ARCH_FALCON || nx % QK_K != 0) {
15480
15480
new_type = GGML_TYPE_Q8_0;
15481
15481
}
15482
- else if (ftype == LLAMA_FTYPE_MOSTLY_IQ1_XS ||ftype == LLAMA_FTYPE_MOSTLY_IQ1_S || ftype == LLAMA_FTYPE_MOSTLY_IQ1_M ||
15483
- ftype == LLAMA_FTYPE_MOSTLY_IQ1_XL) {
15482
+ else if (ftype == LLAMA_FTYPE_MOSTLY_IQ1_XS || ftype == LLAMA_FTYPE_MOSTLY_IQ1_S || ftype == LLAMA_FTYPE_MOSTLY_IQ1_M ||
15483
+ ftype == LLAMA_FTYPE_MOSTLY_IQ1_XL || ftype == LLAMA_FTYPE_MOSTLY_IQ2_XXS || ftype == LLAMA_FTYPE_MOSTLY_IQ2_XS ) {
15484
15484
new_type = GGML_TYPE_Q4_K;
15485
15485
}
15486
- else if (ftype == LLAMA_FTYPE_MOSTLY_IQ2_XXS || ftype == LLAMA_FTYPE_MOSTLY_IQ2_XS || ftype == LLAMA_FTYPE_MOSTLY_IQ3_XXS ||
15486
+ else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_XXS ||
15487
15487
ftype == LLAMA_FTYPE_MOSTLY_IQ2_XL || ftype == LLAMA_FTYPE_MOSTLY_IQ2_S || ftype == LLAMA_FTYPE_MOSTLY_IQ2_M ||
15488
15488
ftype == LLAMA_FTYPE_MOSTLY_Q2_K_S || ftype == LLAMA_FTYPE_MOSTLY_Q2_K || ftype == LLAMA_FTYPE_MOSTLY_Q2_K_L) {
15489
15489
new_type = GGML_TYPE_Q5_K;
@@ -15496,11 +15496,10 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n
15496
15496
if (qs.params->token_embedding_type < GGML_TYPE_COUNT) {
15497
15497
new_type = qs.params->token_embedding_type;
15498
15498
} else {
15499
- if (ftype == LLAMA_FTYPE_MOSTLY_IQ1_XS ||ftype == LLAMA_FTYPE_MOSTLY_IQ1_S || ftype == LLAMA_FTYPE_MOSTLY_IQ1_M ||
15500
- ftype == LLAMA_FTYPE_MOSTLY_IQ1_XL) {
15499
+ if (ftype == LLAMA_FTYPE_MOSTLY_IQ1_XS || ftype == LLAMA_FTYPE_MOSTLY_IQ1_S || ftype == LLAMA_FTYPE_MOSTLY_IQ1_M ||
15500
+ ftype == LLAMA_FTYPE_MOSTLY_IQ1_XL || ftype == LLAMA_FTYPE_MOSTLY_IQ2_XXS || ftype == LLAMA_FTYPE_MOSTLY_IQ2_XS ) {
15501
15501
new_type = GGML_TYPE_IQ2_S;
15502
15502
}
15503
- else if (ftype == LLAMA_FTYPE_MOSTLY_IQ2_XXS || ftype == LLAMA_FTYPE_MOSTLY_IQ2_XS) new_type = GGML_TYPE_Q2_K;
15504
15503
else if (ftype == LLAMA_FTYPE_MOSTLY_IQ2_S || ftype == LLAMA_FTYPE_MOSTLY_IQ2_M) new_type = GGML_TYPE_IQ3_XXS;
15505
15504
else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_XXS) new_type = GGML_TYPE_IQ3_S;
15506
15505
else if (new_type == GGML_TYPE_Q4_0_4_4 || new_type == GGML_TYPE_Q4_0_4_8 || new_type == GGML_TYPE_Q4_0_8_8) {
0 commit comments