vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

squash! common : add default embeddings presets

Browse source 
Update the remaining presets to use the models from ggml-org.
This commit is contained in:
Daniel Bevenius 2025-02-06 10:43:16 +01:00
parent 61f410f9e4
commit d1d0a61b00
1 changed files with 3 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

6
common/arg.cpp
View file

@ -2328,7 +2328,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
{"--embd-bge-small-en-default"}, {"--embd-bge-small-en-default"},
string_format("use default bge-small-en-v1.5 model (note: can download weights from the internet)"), string_format("use default bge-small-en-v1.5 model (note: can download weights from the internet)"),
[](common_params & params) { [](common_params & params) {
params.hf_repo = "CompendiumLabs/bge-small-en-v1.5-gguf"; params.hf_repo = "ggml-org/bge-small-en-v1.5-Q8_0-GGUF";
params.hf_file = "bge-small-en-v1.5-q8_0.gguf"; params.hf_file = "bge-small-en-v1.5-q8_0.gguf";
params.pooling_type = LLAMA_POOLING_TYPE_NONE; params.pooling_type = LLAMA_POOLING_TYPE_NONE;
params.embd_normalize = 2; params.embd_normalize = 2;
@ -2342,8 +2342,8 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
{"--embd-e5-small-en-default"}, {"--embd-e5-small-en-default"},
string_format("use default e5-small-v2 model (note: can download weights from the internet)"), string_format("use default e5-small-v2 model (note: can download weights from the internet)"),
[](common_params & params) { [](common_params & params) {
params.hf_repo = "ChristianAzinn/e5-small-v2-gguf"; params.hf_repo = "ggml-org/e5-small-v2-Q8_0-GGUF";
params.hf_file = "e5-small-v2.Q8_0.gguf"; params.hf_file = "e5-small-v2-q8_0.gguf";
params.pooling_type = LLAMA_POOLING_TYPE_NONE; params.pooling_type = LLAMA_POOLING_TYPE_NONE;
params.embd_normalize = 2; params.embd_normalize = 2;
params.n_ctx = 512; params.n_ctx = 512;