BPE tokenizer 32768 vocab with thinking/tool tokens

special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": "<|im_start|>",
+  "eos_token": "<|endoftext|>",
+  "pad_token": "<|pad|>",
+  "additional_special_tokens": [
+    "<|im_end|>",
+    "<|thinking|>",
+    "<|/thinking|>",
+    "<|tool_calls|>",
+    "<|/tool_calls|>",
+    "<|tool_response|>",
+    "<|/tool_response|>",
+    "<|code|>",
+    "<|/code|>",
+    "<|analysis|>",
+    "<|/analysis|>",
+    "<|result|>",
+    "<|/result|>",
+    "<|tool|>",
+    "<|system|>",
+    "<|user|>",
+    "<|assistant|>"
+  ]
+}

tokenizer_config.json

@@ -0,0 +1,11 @@
+{
+  "tokenizer_class": "PreTrainedTokenizerFast",
+  "bos_token": "<|im_start|>",
+  "eos_token": "<|endoftext|>",
+  "pad_token": "<|pad|>",
+  "model_max_length": 32768,
+  "chat_template": "{% for message in messages %}\n{% if message['role'] == 'system' %}<|im_start|>system\n{{ message['content'] }}<|im_end|>\n{% elif message['role'] == 'user' %}<|im_start|>user\n{{ message['content'] }}<|im_end|>\n{% elif message['role'] == 'assistant' %}\n{% if message.get('thinking') %}<|thinking|>\n{{ message['thinking'] }}<|/thinking|>{% endif %}\n{% if message.get('tool_calls') %}<|tool_calls|>\n{{ message['tool_calls'] }}<|/tool_calls|>{% endif %}\n<|im_start|>assistant\n{{ message['content'] }}<|im_end|>\n{% elif message['role'] == 'tool' %}<|im_start|>tool\n{{ message['content'] }}<|im_end|>\n{% endif %}\n{% endfor %}\n{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}",
+  "add_bos_token": false,
+  "add_eos_token": false,
+  "clean_up_tokenization_spaces": true
+}