AI/gpt4all
1
0
Fork 0
mirror of https://github.com/nomic-ai/gpt4all.git synced 2024-10-01 01:06:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

Use the default for max_tokens to avoid errors.

Browse Source
This commit is contained in:
Adam Treat 2023-05-16 09:48:21 -04:00 committed by AT
parent b24ace372b
commit 79d6243fe1
1 changed files with 5 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
gpt4all-chat/chatgpt.cpp
View File

@ -81,12 +81,16 @@ void ChatGPT::prompt(const std::string &prompt,
m_ctx = &promptCtx;
m_responseCallback = responseCallback;
// FIXME: We don't set the max_tokens on purpose because in order to do so safely without encountering
// an error we need to be able to count the tokens in our prompt. The only way to do this is to use
// the OpenAI tiktokken library or to implement our own tokenization function that matches precisely
// the tokenization used by the OpenAI model we're calling. OpenAI has not introduced any means of
// using the REST API to count tokens in a prompt.
QJsonObject root;
root.insert("model", m_modelName);
root.insert("stream", true);
root.insert("temperature", promptCtx.temp);
root.insert("top_p", promptCtx.top_p);
root.insert("max_tokens", 200);
QJsonArray messages;
for (int i = 0; i < m_context.count() && i < promptCtx.n_past; ++i) {