ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-04-28 18:32:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

Checkpoint

Browse Source 
Go to groups of 8 for iq3_kt. 2 x 8 = 16 bits for the magnitude
plus 1 bpw for the sign. It goves a visible improvement in the
PPL vs bpw plot, but that comes at the expense of much longer
quantization time (7.5 minutes for LLaMA-3.1-8B on the Ryzen-5975WX).

I also notices that the 3INST generator is not actually generating a
Gaussian distribution. But going to a better generator means
readjusting all the hyper-parameters, so leaving it for later.
This commit is contained in:
Iwan Kawrakow
2024-11-19 17:31:07 +02:00
parent 2be4cffe66
commit 3a9926b932
4 changed files with 237 additions and 129 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/llama.cpp
View File

@@ -15976,7 +15976,7 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n
} else {
if (ftype == LLAMA_FTYPE_MOSTLY_Q2_K ) new_type = GGML_TYPE_Q3_K;
else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_XXS) new_type = GGML_TYPE_IQ3_S;
//else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_KT ) new_type = GGML_TYPE_IQ3_K;
else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_KT && qs.model.hparams.n_gqa() >= 4) new_type = GGML_TYPE_IQ3_K;
else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M ) new_type = GGML_TYPE_Q4_K;
else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L ) new_type = GGML_TYPE_Q5_K;
else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_M ) new_type = GGML_TYPE_IQ4_K;