Update README.md
Browse files
README.md
CHANGED
|
@@ -17,6 +17,7 @@ tags:
|
|
| 17 |
`-imat` quants use an imatrix made with the prompt trained and a few caption generated by myself with those prompts.
|
| 18 |
|
| 19 |
The `tokenizer.chat_template` gguf metadata value has been updated in all quants to include the right template, the following is not needed anymore.
|
|
|
|
| 20 |
~~You have to add `--reasoning off` or `--chat-template-file <CHANGE_PATH_TO_DIR>/chat_template.jinja`, else the entire message will be output in the reasoning part.~~
|
| 21 |
|
| 22 |
Example `llama-server` command (change path and flags to your liking):
|
|
|
|
| 17 |
`-imat` quants use an imatrix made with the prompt trained and a few caption generated by myself with those prompts.
|
| 18 |
|
| 19 |
The `tokenizer.chat_template` gguf metadata value has been updated in all quants to include the right template, the following is not needed anymore.
|
| 20 |
+
|
| 21 |
~~You have to add `--reasoning off` or `--chat-template-file <CHANGE_PATH_TO_DIR>/chat_template.jinja`, else the entire message will be output in the reasoning part.~~
|
| 22 |
|
| 23 |
Example `llama-server` command (change path and flags to your liking):
|