Upload tokenizer

special_tokens_map.json

@@ -15,7 +15,7 @@
       "single_word": false
     },
     {
-      "content": "<unk>",
+      "content": "</s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -24,7 +24,7 @@
   ],
   "bos_token": "<s>",
   "eos_token": "</s>",
-  "pad_token": "<unk>",
+  "pad_token": "</s>",
   "unk_token": {
     "content": "<unk>",
     "lstrip": false,

tokenizer_config.json

@@ -31,16 +31,16 @@
   "additional_special_tokens": [
     "<s>",
     "</s>",
-    "<unk>"
+    "</s>"
   ],
   "bos_token": "<s>",
   "chat_template": "{%- set system_message_present = messages | selectattr('role', 'equalto', 'system') | list -%}{%- if not system_message_present -%}{%- set messages = [{ \"content\": \"You are an AI agent acting as a human assistant.\", \"role\": \"system\" }] + messages -%}{%- endif -%}{%- for message in messages -%}<|{{ message.role }}|>{{ '\n' }}{%- if message.role == 'system' -%}{{ message.content }}{%- if tools and tools | length > 0 -%}{{ '\n\n' }}You are aware of the following tools in your environment:{{ '\n' }}{\n  \"tools\": [{{ '\n' }}{%- for tool in tools -%}{{ '    ' }}{\n      \"function\": {\n        \"description\": \"{{ tool.function.description }}\",{{ '\n' }}        \"name\": \"{{ tool.function.name }}\",{{ '\n' }}        \"parameters\": {{ tool.function.parameters | tojson }}{{ '\n' }}      },{{ '\n' }}      \"type\": \"{{ tool.type }}\"{{ '\n' }}    }{%- if not loop.last -%},{%- endif -%}{{ '\n' }}{%- endfor -%}{{ '  ' }}]{{ '\n' }}}{{ '\n\n' }}If you would like to suggest one or more tool calls, please respond in the following format:{{ '\n' }}{\n  \"finish_reason\": \"tool_calls\",{{ '\n' }}  \"tool_calls\": [{{ '\n' }}{{ '    ' }}{\n      \"arguments\": \"{\\\"parameter_name\\\": \\\"parameter_value\\\"}\",{{ '\n' }}      \"id\": \"call_id\",{{ '\n' }}      \"name\": \"tool_name\"{{ '\n' }}    }{{ '\n' }}  ]{{ '\n' }}}{%- endif -%}{{ eos_token }}{{ '\n' }}{%- endif -%}{%- if message.role == 'user' -%}{{ message.content }}{{ eos_token }}{{ '\n' }}{%- endif -%}{%- if message.role == 'assistant' -%}{% generation %}{%- if message.tool_calls and message.tool_calls | length > 0 -%}{\n  \"finish_reason\": \"tool_calls\",{{ '\n' }}  \"tool_calls\": [{{ '\n' }}{%- for tool_call in message.tool_calls -%}{{ '    ' }}{\n      \"arguments\": {{ tool_call.function.arguments | tojson }},{{ '\n' }}      \"id\": \"{{ tool_call.id }}\",{{ '\n' }}      \"name\": \"{{ tool_call.function.name }}\"{{ '\n' }}    }{%- if not loop.last -%},{%- endif -%}{{ '\n' }}{%- endfor -%}{{ '  ' }}]{{ '\n' }}}{%- else -%}{{ message.content }}{%- endif -%}{% endgeneration %}{{ eos_token }}{{ '\n' }}{%- endif -%}{%- if message.role == 'tool' -%}{\n  \"content\": {{ message.content | tojson }},{{ '\n' }}  \"name\": \"{{ message.name }}\",{{ '\n' }}  \"tool_call_id\": \"{{ message.tool_call_id }}\"{{ '\n' }}}{{ eos_token }}{{ '\n' }}{%- endif -%}{%- endfor -%}{%- if add_generation_prompt -%}{{ '<|assistant|>\n' }}{%- endif -%}",
   "clean_up_tokenization_spaces": false,
   "eos_token": "</s>",
   "legacy": false,
-  "model_max_length": 4096,
-  "pad_token": "<unk>",
-  "padding_side": "left",
+  "model_max_length": 2048,
+  "pad_token": "</s>",
+  "padding_side": "right",
   "sp_model_kwargs": {},
   "tokenizer_class": "LlamaTokenizer",
   "unk_token": "<unk>",