matt/ollama37
1
0
Fork 0
mirror of https://github.com/dogkeeper886/ollama37.git synced 2025-12-11 00:07:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

update llama.cpp submodule to f364eb6 (#4060)

Browse Source
This commit is contained in:
Jeffrey Morgan
2024-04-30 17:25:39 -04:00
committed by GitHub
parent 8488388cbd
commit 18d9a7e1f1
2 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
llm/ext_server/server.cpp vendored
View File

@@ -1032,7 +1032,7 @@ struct llama_server_context
slot.has_next_token = false; slot.has_next_token = false;
} }
if (!slot.cache_tokens.empty() && result.tok == llama_token_eos(model)) if (llama_token_is_eog(model, result.tok))
{ {
slot.stopped_eos = true; slot.stopped_eos = true;
slot.has_next_token = false; slot.has_next_token = false;
@@ -1144,12 +1144,15 @@ struct llama_server_context
res.result_json = json res.result_json = json
{ {
{"content", tkn.text_to_send},
{"stop", false}, {"stop", false},
{"slot_id", slot.id}, {"slot_id", slot.id},
{"multimodal", multimodal} {"multimodal", multimodal}
}; };
if (!llama_token_is_eog(model, tkn.tok)) {
res.result_json["content"] = tkn.text_to_send;
}
if (slot.sparams.n_probs > 0) if (slot.sparams.n_probs > 0)
{ {
std::vector<completion_token_output> probs_output = {}; std::vector<completion_token_output> probs_output = {};

2
llm/llama.cpp

Submodule llm/llama.cpp updated: f4ab2a4147...f364eb6fb5