Update README.md
Browse files
README.md
CHANGED
|
@@ -21,22 +21,22 @@ This tokenizer supports the following special tokens:
|
|
| 21 |
- `|||IP_ADDRESS|||`: Not used. Kept for compatibility with Dolma 1.x.
|
| 22 |
- `<|im_start|>`: Indicates the beginning of a message (turn in a conversation).
|
| 23 |
- `<|im_end|>`: Indicates the end of a message (turn in a conversation).
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
- `<|extra_id_1|>`: Not used.
|
| 25 |
- `<|extra_id_2|>`: Not used.
|
| 26 |
-
- `<think>`: Indicates the beginning of model thoughts.
|
| 27 |
-
- `</think>`: Indicates the end of model thoughts.
|
| 28 |
- `<|extra_id_3|>`: Not used.
|
| 29 |
- `<|extra_id_4|>`: Not used.
|
| 30 |
- `<|extra_id_5|>`: Not used.
|
| 31 |
- `<|extra_id_6|>`: Not used.
|
| 32 |
-
-
|
| 33 |
-
-
|
|
|
|
|
|
|
| 34 |
- `<|endofprompt|>`: Not Used.
|
| 35 |
- `<|pad|>`: Symbol to pad input sequences.
|
| 36 |
-
- `<functions>`: Indicates start of function definitions in the system prompt for tool use.
|
| 37 |
-
- `</functions>`: Indicates end of function definitions in the sytem prompt.
|
| 38 |
-
- `<function_calls>`: Indicates start of function calls made by the model.
|
| 39 |
-
- `</function_calls>`: Indicates end of function calls made by the model.
|
| 40 |
|
| 41 |
|
| 42 |
## Chat template
|
|
|
|
| 21 |
- `|||IP_ADDRESS|||`: Not used. Kept for compatibility with Dolma 1.x.
|
| 22 |
- `<|im_start|>`: Indicates the beginning of a message (turn in a conversation).
|
| 23 |
- `<|im_end|>`: Indicates the end of a message (turn in a conversation).
|
| 24 |
+
- `<functions>`: Indicates start of function definitions in the system prompt for tool use.
|
| 25 |
+
- `</functions>`: Indicates end of function definitions in the sytem prompt.
|
| 26 |
+
- `<function_calls>`: Indicates start of function calls made by the model.
|
| 27 |
+
- `</function_calls>`: Indicates end of function calls made by the model.
|
| 28 |
- `<|extra_id_1|>`: Not used.
|
| 29 |
- `<|extra_id_2|>`: Not used.
|
|
|
|
|
|
|
| 30 |
- `<|extra_id_3|>`: Not used.
|
| 31 |
- `<|extra_id_4|>`: Not used.
|
| 32 |
- `<|extra_id_5|>`: Not used.
|
| 33 |
- `<|extra_id_6|>`: Not used.
|
| 34 |
+
- `<|extra_id_7|>`: Not used.
|
| 35 |
+
- `<|extra_id_8|>`: Not used.
|
| 36 |
+
- `<|extra_id_9|>`: Not used.
|
| 37 |
+
- `<|extra_id_10|>`: Not used.
|
| 38 |
- `<|endofprompt|>`: Not Used.
|
| 39 |
- `<|pad|>`: Symbol to pad input sequences.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
|
| 42 |
## Chat template
|