feat: use tokenizer.apply_chat_template() in vLLM (#1990)

Use tokenizer.apply_chat_template() in vLLM

Signed-off-by: Ludovic LEROUX <ludovic@inpher.io>
This commit is contained in:
Ludovic Leroux 2024-04-11 13:20:22 -04:00 committed by GitHub
parent cbda06fb96
commit 12c0d9443e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
34 changed files with 3088 additions and 989 deletions

View file

@ -107,6 +107,8 @@ message PredictOptions {
string NegativePrompt = 40;
int32 NDraft = 41;
repeated string Images = 42;
bool UseTokenizerTemplate = 43;
repeated Message Messages = 44;
}
// The response message containing the result
@ -256,3 +258,8 @@ message StatusResponse {
State state = 1;
MemoryUsageData memory = 2;
}
message Message {
string role = 1;
string content = 2;
}